diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,99549 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.8612544834222726,
+  "eval_steps": 500,
+  "global_step": 16588,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.7199,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 0.805,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 0.9467,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 0.6916,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 0.8694,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 0.7637,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014,
+      "loss": 0.9346,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 0.6136,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018,
+      "loss": 0.8,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9198,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999831925614,
+      "loss": 0.9382,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999327702467,
+      "loss": 1.0295,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999998487330568,
+      "loss": 0.8922,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997310809952,
+      "loss": 0.8105,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999995798140657,
+      "loss": 1.0031,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999394932273,
+      "loss": 0.6206,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999176435624,
+      "loss": 0.9784,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999989243241255,
+      "loss": 0.7659,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998638597786,
+      "loss": 0.6858,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999983192566157,
+      "loss": 0.8177,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999979663006245,
+      "loss": 0.8118,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999975797298247,
+      "loss": 0.7845,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999997159544229,
+      "loss": 0.9051,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996705743852,
+      "loss": 0.8631,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999996218328709,
+      "loss": 0.6533,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999995697298816,
+      "loss": 0.8376,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999951426541906,
+      "loss": 0.7084,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999994554394851,
+      "loss": 1.1035,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999993932520818,
+      "loss": 0.8748,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999932770321116,
+      "loss": 0.8335,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999925879287544,
+      "loss": 0.993,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999918652107695,
+      "loss": 0.898,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999911088781805,
+      "loss": 0.8519,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999903189310135,
+      "loss": 1.0362,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999894953692953,
+      "loss": 0.6559,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999988638193053,
+      "loss": 0.8633,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999877474023155,
+      "loss": 0.7078,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999986822997113,
+      "loss": 0.7114,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999858649774764,
+      "loss": 0.769,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999848733434378,
+      "loss": 0.6608,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999983848095031,
+      "loss": 0.9467,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999827892322897,
+      "loss": 0.8933,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199998169675525,
+      "loss": 0.7929,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999805706639488,
+      "loss": 0.7625,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999794109584236,
+      "loss": 0.7197,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999782176387135,
+      "loss": 0.7544,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999769907048584,
+      "loss": 0.7698,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999757301569,
+      "loss": 0.9213,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999744359948805,
+      "loss": 0.878,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999973108218843,
+      "loss": 0.726,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999717468288325,
+      "loss": 0.7409,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999703518248948,
+      "loss": 0.9382,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999968923207077,
+      "loss": 0.9218,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999674609754264,
+      "loss": 0.8698,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999659651299926,
+      "loss": 0.6652,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999644356708261,
+      "loss": 0.6995,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999962872597978,
+      "loss": 0.7243,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999612759115008,
+      "loss": 0.8642,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999596456114484,
+      "loss": 0.8807,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999579816978755,
+      "loss": 0.896,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999956284170838,
+      "loss": 0.4963,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999954553030393,
+      "loss": 0.6397,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999527882765987,
+      "loss": 0.7187,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999950989909514,
+      "loss": 0.8534,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999491579292004,
+      "loss": 0.8356,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999472923357185,
+      "loss": 0.7319,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999453931291315,
+      "loss": 0.6541,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999943460309503,
+      "loss": 0.6762,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999941493876898,
+      "loss": 0.7663,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999939493831383,
+      "loss": 0.8119,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999374601730244,
+      "loss": 0.9552,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999353929018915,
+      "loss": 0.5735,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999332920180528,
+      "loss": 0.8131,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999311575215797,
+      "loss": 0.8975,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999289894125438,
+      "loss": 0.8001,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999267876910182,
+      "loss": 0.6619,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999245523570763,
+      "loss": 0.7864,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999222834107933,
+      "loss": 1.0257,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999919980852246,
+      "loss": 0.8059,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999176446815117,
+      "loss": 0.7944,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999152748986684,
+      "loss": 0.7643,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999128715037965,
+      "loss": 0.893,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999104344969762,
+      "loss": 0.5073,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999079638782893,
+      "loss": 0.959,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999054596478194,
+      "loss": 0.8839,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999029218056508,
+      "loss": 0.7206,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999900350351868,
+      "loss": 1.0036,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999897745286558,
+      "loss": 0.8536,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998951066098084,
+      "loss": 0.5982,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998924343217077,
+      "loss": 0.9153,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998897284223457,
+      "loss": 0.9578,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998869889118135,
+      "loss": 1.0133,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998842157902035,
+      "loss": 0.8965,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999881409057608,
+      "loss": 0.7524,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998785687141223,
+      "loss": 0.8401,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999875694759841,
+      "loss": 0.9346,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998727871948618,
+      "loss": 0.8993,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998698460192813,
+      "loss": 0.9417,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998668712331995,
+      "loss": 0.6835,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998638628367153,
+      "loss": 0.7598,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998608208299302,
+      "loss": 0.5782,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999857745212947,
+      "loss": 0.7008,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998546359858686,
+      "loss": 0.9307,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998514931487992,
+      "loss": 0.8149,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998483167018454,
+      "loss": 0.8283,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999845106645113,
+      "loss": 1.1012,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998418629787104,
+      "loss": 0.7821,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998385857027468,
+      "loss": 0.8203,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998352748173316,
+      "loss": 0.6769,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999831930322577,
+      "loss": 0.9319,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999828552218595,
+      "loss": 0.6304,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998251405054987,
+      "loss": 0.8134,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999821695183404,
+      "loss": 0.9526,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998182162524257,
+      "loss": 0.6363,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999814703712681,
+      "loss": 0.8205,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998111575642878,
+      "loss": 0.9742,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998075778073656,
+      "loss": 0.7705,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998039644420347,
+      "loss": 0.7763,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998003174684167,
+      "loss": 0.6184,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997966368866336,
+      "loss": 0.827,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997929226968098,
+      "loss": 0.5372,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997891748990698,
+      "loss": 0.6692,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997853934935393,
+      "loss": 0.8195,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997815784803463,
+      "loss": 0.79,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999777729859618,
+      "loss": 0.7324,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997738476314848,
+      "loss": 0.5056,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997699317960766,
+      "loss": 0.8393,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997659823535246,
+      "loss": 0.8404,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997619993039625,
+      "loss": 0.9393,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997579826475236,
+      "loss": 0.8566,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999753932384343,
+      "loss": 0.8063,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997498485145568,
+      "loss": 0.4418,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997457310383027,
+      "loss": 0.7619,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999741579955719,
+      "loss": 0.7905,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997373952669446,
+      "loss": 0.5905,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999733176972121,
+      "loss": 0.7844,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997289250713892,
+      "loss": 0.9462,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997246395648928,
+      "loss": 0.6628,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997203204527754,
+      "loss": 0.9415,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997159677351826,
+      "loss": 0.826,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000199971158141226,
+      "loss": 0.6456,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019997071614841563,
+      "loss": 0.9123,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999702707951019,
+      "loss": 1.0632,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999698220812998,
+      "loss": 0.7061,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996937000702443,
+      "loss": 0.7193,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199968914572291,
+      "loss": 0.8488,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996845577711477,
+      "loss": 0.9048,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996799362151122,
+      "loss": 0.6541,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996752810549584,
+      "loss": 0.8182,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996705922908432,
+      "loss": 1.0219,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996658699229238,
+      "loss": 0.4944,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996611139513592,
+      "loss": 0.6013,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999656324376309,
+      "loss": 0.537,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996515011979347,
+      "loss": 0.7879,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999646644416398,
+      "loss": 0.8204,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996417540318624,
+      "loss": 0.8174,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996368300444922,
+      "loss": 0.7133,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996318724544527,
+      "loss": 0.696,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996268812619107,
+      "loss": 1.046,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996218564670342,
+      "loss": 0.8863,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999616798069992,
+      "loss": 0.9716,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999611706070954,
+      "loss": 0.5357,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999606580470091,
+      "loss": 0.7239,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996014212675763,
+      "loss": 0.7839,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995962284635827,
+      "loss": 0.7606,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995910020582849,
+      "loss": 0.7373,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999585742051858,
+      "loss": 0.5467,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995804484444798,
+      "loss": 0.9355,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995751212363276,
+      "loss": 0.5586,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995697604275806,
+      "loss": 1.0643,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995643660184191,
+      "loss": 0.9176,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995589380090244,
+      "loss": 0.9771,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999553476399579,
+      "loss": 0.8295,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995479811902661,
+      "loss": 0.9399,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999542452381271,
+      "loss": 0.8542,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995368899727793,
+      "loss": 0.7516,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999531293964978,
+      "loss": 0.8793,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999525664358055,
+      "loss": 0.7973,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995200011521997,
+      "loss": 0.5836,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999514304347603,
+      "loss": 0.7604,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995085739444552,
+      "loss": 0.8096,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019995028099429498,
+      "loss": 0.8494,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994970123432806,
+      "loss": 0.684,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999491181145642,
+      "loss": 1.0991,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994853163502306,
+      "loss": 0.8043,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999479417957243,
+      "loss": 1.0486,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994734859668775,
+      "loss": 0.8378,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999467520379334,
+      "loss": 0.6527,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994615211948126,
+      "loss": 0.7909,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994554884135153,
+      "loss": 0.8528,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994494220356443,
+      "loss": 0.7075,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999443322061404,
+      "loss": 0.7444,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994371884909994,
+      "loss": 0.6695,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994310213246368,
+      "loss": 0.8869,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994248205625233,
+      "loss": 0.7034,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994185862048672,
+      "loss": 0.713,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019994123182518782,
+      "loss": 0.8529,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999406016703767,
+      "loss": 0.986,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019993996815607457,
+      "loss": 0.8744,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019993933128230272,
+      "loss": 0.7906,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999386910490825,
+      "loss": 0.5993,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993804745643546,
+      "loss": 0.8262,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993740050438329,
+      "loss": 0.8209,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993675019294767,
+      "loss": 0.9562,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999360965221505,
+      "loss": 0.7123,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999354394920137,
+      "loss": 0.4869,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993477910255943,
+      "loss": 0.8085,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999341153538098,
+      "loss": 0.7958,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999334482457872,
+      "loss": 0.8509,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993277777851403,
+      "loss": 1.1027,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993210395201287,
+      "loss": 0.7983,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993142676630627,
+      "loss": 0.5059,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999307462214171,
+      "loss": 0.723,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993006231736814,
+      "loss": 0.5704,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999293750541824,
+      "loss": 0.8502,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992868443188306,
+      "loss": 0.844,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999279904504933,
+      "loss": 1.0006,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992729311003638,
+      "loss": 1.0059,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992659241053578,
+      "loss": 0.7027,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999258883520151,
+      "loss": 1.0108,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992518093449798,
+      "loss": 0.942,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992447015800818,
+      "loss": 0.7798,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999237560225696,
+      "loss": 0.9415,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992303852820625,
+      "loss": 1.1018,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992231767494226,
+      "loss": 0.9409,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992159346280186,
+      "loss": 0.8469,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999208658918094,
+      "loss": 0.9706,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992013496198928,
+      "loss": 0.6583,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999194006733661,
+      "loss": 0.8953,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991866302596458,
+      "loss": 0.9293,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999179220198095,
+      "loss": 0.8865,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991717765492574,
+      "loss": 0.7254,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991642993133838,
+      "loss": 0.7549,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991567884907247,
+      "loss": 0.8362,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991492440815333,
+      "loss": 0.6564,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999141666086063,
+      "loss": 0.8251,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999134054504568,
+      "loss": 1.0015,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991264093373052,
+      "loss": 0.4737,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991187305845308,
+      "loss": 0.9589,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991110182465032,
+      "loss": 0.7726,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991032723234815,
+      "loss": 0.9139,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999095492815726,
+      "loss": 0.7454,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990876797234988,
+      "loss": 0.5997,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990798330470622,
+      "loss": 0.8792,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990719527866795,
+      "loss": 0.9483,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990640389426162,
+      "loss": 0.5828,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999056091515138,
+      "loss": 0.547,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999048110504512,
+      "loss": 0.9757,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999040095911007,
+      "loss": 0.8175,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990320477348918,
+      "loss": 0.8095,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999023965976437,
+      "loss": 0.9248,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990158506359148,
+      "loss": 0.931,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990077017135975,
+      "loss": 0.8933,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989995192097593,
+      "loss": 0.9807,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998991303124675,
+      "loss": 0.7464,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998983053458621,
+      "loss": 0.7187,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989747702118743,
+      "loss": 0.7342,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989664533847135,
+      "loss": 0.8852,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989581029774186,
+      "loss": 0.7199,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199894971899027,
+      "loss": 0.9691,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989413014235492,
+      "loss": 0.7748,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989328502775395,
+      "loss": 0.7666,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989243655525247,
+      "loss": 0.8289,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989158472487903,
+      "loss": 0.886,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019989072953666227,
+      "loss": 0.7552,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998898709906309,
+      "loss": 0.7991,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988900908681382,
+      "loss": 0.3021,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988814382523997,
+      "loss": 0.7897,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988727520593846,
+      "loss": 0.7632,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988640322893847,
+      "loss": 0.7282,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988552789426933,
+      "loss": 1.0337,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988464920196047,
+      "loss": 0.85,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988376715204137,
+      "loss": 0.8555,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988288174454171,
+      "loss": 0.7457,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988199297949134,
+      "loss": 0.8836,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988110085692,
+      "loss": 1.2996,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988020537685776,
+      "loss": 0.8198,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998793065393347,
+      "loss": 0.8061,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987840434438104,
+      "loss": 0.743,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987749879202707,
+      "loss": 0.6841,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998765898823033,
+      "loss": 0.9369,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987567761524022,
+      "loss": 0.9303,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987476199086855,
+      "loss": 0.8315,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199873843009219,
+      "loss": 0.6697,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987292067032253,
+      "loss": 0.9709,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998719949742101,
+      "loss": 0.7439,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987106592091284,
+      "loss": 0.847,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987013351046199,
+      "loss": 0.5929,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986919774288888,
+      "loss": 0.6776,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986825861822497,
+      "loss": 0.7044,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986731613650185,
+      "loss": 0.944,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986637029775116,
+      "loss": 0.9283,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998654211020047,
+      "loss": 0.6722,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986446854929443,
+      "loss": 0.84,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998635126396523,
+      "loss": 0.6897,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998625533731105,
+      "loss": 1.2113,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998615907497012,
+      "loss": 0.806,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986062476945687,
+      "loss": 0.8526,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985965543240988,
+      "loss": 0.8153,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985868273859287,
+      "loss": 0.4516,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985770668803853,
+      "loss": 0.5378,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985672728077965,
+      "loss": 0.7201,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985574451684914,
+      "loss": 0.8745,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998547583962801,
+      "loss": 0.6647,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998537689191056,
+      "loss": 0.4259,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985277608535897,
+      "loss": 1.004,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985177989507356,
+      "loss": 0.5194,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998507803482828,
+      "loss": 0.8122,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984977744502038,
+      "loss": 0.8439,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984877118531995,
+      "loss": 0.5752,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984776156921536,
+      "loss": 0.9515,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984674859674058,
+      "loss": 0.5376,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984573226792957,
+      "loss": 0.6689,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984471258281659,
+      "loss": 0.9413,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019984368954143586,
+      "loss": 0.9275,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019984266314382177,
+      "loss": 0.9343,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019984163339000887,
+      "loss": 0.7005,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019984060028003172,
+      "loss": 0.9917,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983956381392507,
+      "loss": 0.8871,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998385239917238,
+      "loss": 0.892,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998374808134628,
+      "loss": 0.8219,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983643427917716,
+      "loss": 0.5366,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983538438890204,
+      "loss": 0.9305,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983433114267277,
+      "loss": 0.8928,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983327454052474,
+      "loss": 0.6436,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983221458249345,
+      "loss": 0.8005,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983115126861456,
+      "loss": 0.7446,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019983008459892378,
+      "loss": 0.9599,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982901457345703,
+      "loss": 0.3609,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982794119225016,
+      "loss": 0.8321,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982686445533938,
+      "loss": 0.5267,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982578436276082,
+      "loss": 0.7302,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998247009145508,
+      "loss": 0.867,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982361411074572,
+      "loss": 1.0449,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982252395138212,
+      "loss": 0.6984,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982143043649668,
+      "loss": 0.752,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019982033356612613,
+      "loss": 0.6426,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981923334030731,
+      "loss": 0.7104,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981812975907728,
+      "loss": 0.7237,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981702282247308,
+      "loss": 0.7542,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981591253053193,
+      "loss": 0.7955,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981479888329115,
+      "loss": 0.6992,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981368188078818,
+      "loss": 0.7745,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019981256152306058,
+      "loss": 0.8471,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199811437810146,
+      "loss": 0.7395,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998103107420822,
+      "loss": 0.811,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998091803189071,
+      "loss": 0.7575,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980804654065866,
+      "loss": 0.8558,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980690940737502,
+      "loss": 0.7215,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980576891909441,
+      "loss": 0.7259,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980462507585513,
+      "loss": 1.1169,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980347787769566,
+      "loss": 0.8584,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980232732465456,
+      "loss": 0.7948,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998011734167705,
+      "loss": 1.0297,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019980001615408228,
+      "loss": 1.1306,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979885553662876,
+      "loss": 1.0863,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199797691564449,
+      "loss": 0.8862,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979652423758213,
+      "loss": 0.8676,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979535355606735,
+      "loss": 0.8583,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979417951994407,
+      "loss": 0.8834,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979300212925164,
+      "loss": 0.7554,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979182138402977,
+      "loss": 0.8743,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019979063728431813,
+      "loss": 0.7672,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978944983015643,
+      "loss": 0.7503,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978825902158468,
+      "loss": 0.6295,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978706485864286,
+      "loss": 0.7739,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978586734137115,
+      "loss": 0.6409,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978466646980976,
+      "loss": 0.8245,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978346224399908,
+      "loss": 0.6862,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001997822546639796,
+      "loss": 0.5554,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019978104372979187,
+      "loss": 0.9106,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977982944147666,
+      "loss": 0.8567,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977861179907476,
+      "loss": 0.79,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977739080262707,
+      "loss": 0.8113,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977616645217467,
+      "loss": 0.8957,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977493874775873,
+      "loss": 0.9909,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977370768942045,
+      "loss": 0.8737,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019977247327720128,
+      "loss": 0.8458,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997712355111427,
+      "loss": 0.9375,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997699943912863,
+      "loss": 0.8369,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997687499176738,
+      "loss": 0.5132,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019976750209034705,
+      "loss": 0.7266,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199766250909348,
+      "loss": 0.9774,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019976499637471866,
+      "loss": 0.9387,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019976373848650124,
+      "loss": 1.0494,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019976247724473806,
+      "loss": 0.9941,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019976121264947145,
+      "loss": 0.7196,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019975994470074397,
+      "loss": 1.1757,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997586733985982,
+      "loss": 0.9028,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997573987430769,
+      "loss": 0.7315,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997561207342229,
+      "loss": 0.9611,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019975483937207918,
+      "loss": 0.8954,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997535546566888,
+      "loss": 0.7228,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019975226658809492,
+      "loss": 1.0516,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019975097516634088,
+      "loss": 0.7737,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997496803914701,
+      "loss": 0.7747,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997483822635261,
+      "loss": 0.6684,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019974708078255245,
+      "loss": 0.5897,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019974577594859296,
+      "loss": 1.0293,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019974446776169149,
+      "loss": 0.9366,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199743156221892,
+      "loss": 0.6153,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997418413292386,
+      "loss": 0.9184,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019974052308377544,
+      "loss": 0.9181,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019973920148554685,
+      "loss": 0.9252,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019973787653459733,
+      "loss": 0.9011,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997365482309713,
+      "loss": 0.8916,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997352165747135,
+      "loss": 0.7552,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019973388156586865,
+      "loss": 0.7198,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019973254320448163,
+      "loss": 1.3034,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019973120149059743,
+      "loss": 0.9143,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019972985642426115,
+      "loss": 0.759,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019972850800551806,
+      "loss": 0.6614,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997271562344134,
+      "loss": 0.7605,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019972580111099264,
+      "loss": 0.7142,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019972444263530138,
+      "loss": 0.8482,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001997230808073852,
+      "loss": 0.8234,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019972171562728996,
+      "loss": 0.8095,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019972034709506148,
+      "loss": 0.5827,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971897521074583,
+      "loss": 0.6554,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971759997438907,
+      "loss": 0.8072,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971622138603747,
+      "loss": 0.8314,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971483944573735,
+      "loss": 0.9174,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971345415353516,
+      "loss": 1.0227,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971206550947748,
+      "loss": 0.9134,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019971067351361095,
+      "loss": 0.9234,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019970927816598242,
+      "loss": 0.893,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019970787946663876,
+      "loss": 0.999,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199706477415627,
+      "loss": 0.7828,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019970507201299425,
+      "loss": 0.6189,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019970366325878779,
+      "loss": 0.6158,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019970225115305493,
+      "loss": 1.1153,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019970083569584318,
+      "loss": 1.0481,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996994168872001,
+      "loss": 0.8387,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969799472717333,
+      "loss": 0.775,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996965692158108,
+      "loss": 0.7172,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996951403531603,
+      "loss": 0.8834,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969370813927,
+      "loss": 0.7321,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969227257418787,
+      "loss": 0.5709,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019969083365796228,
+      "loss": 0.7434,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996893913906416,
+      "loss": 0.7158,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968794577227429,
+      "loss": 0.8811,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968649680290895,
+      "loss": 0.7923,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968504448259424,
+      "loss": 0.7722,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968358881137905,
+      "loss": 0.8936,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996821297893123,
+      "loss": 0.5117,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019968066741644295,
+      "loss": 1.1236,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996792016928203,
+      "loss": 0.7458,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996777326184935,
+      "loss": 0.796,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199676260193512,
+      "loss": 0.4112,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019967478441792525,
+      "loss": 0.889,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019967330529178289,
+      "loss": 0.8379,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019967182281513464,
+      "loss": 0.855,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996703369880303,
+      "loss": 0.6757,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966884781051985,
+      "loss": 0.9854,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966735528265334,
+      "loss": 1.0439,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966585940448095,
+      "loss": 0.9692,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966436017605297,
+      "loss": 1.0572,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966285759741976,
+      "loss": 0.8713,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019966135166863184,
+      "loss": 0.7563,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965984238973984,
+      "loss": 0.7805,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965832976079454,
+      "loss": 0.6105,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996568137818467,
+      "loss": 0.4788,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996552944529473,
+      "loss": 0.8044,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996537717741475,
+      "loss": 0.839,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965224574549834,
+      "loss": 0.7095,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019965071636705124,
+      "loss": 0.7777,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964918363885755,
+      "loss": 0.8643,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996476475609688,
+      "loss": 0.6839,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964610813343666,
+      "loss": 0.8375,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964456535631286,
+      "loss": 0.7154,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964301922964923,
+      "loss": 0.8813,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019964146975349776,
+      "loss": 0.8356,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963991692791054,
+      "loss": 0.8204,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963836075293978,
+      "loss": 0.5341,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963680122863774,
+      "loss": 0.957,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963523835505692,
+      "loss": 0.5644,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963367213224982,
+      "loss": 1.1211,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963210256026905,
+      "loss": 0.6793,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019963052963916741,
+      "loss": 0.8589,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996289533689978,
+      "loss": 0.6171,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019962737374981315,
+      "loss": 0.5957,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001996257907816666,
+      "loss": 0.7759,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019962420446461134,
+      "loss": 0.7412,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019962261479870069,
+      "loss": 0.9193,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019962102178398808,
+      "loss": 0.5367,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019961942542052712,
+      "loss": 0.909,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001996178257083714,
+      "loss": 0.671,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019961622264757473,
+      "loss": 0.6067,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019961461623819099,
+      "loss": 0.695,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019961300648027418,
+      "loss": 0.7447,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001996113933738784,
+      "loss": 1.0703,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019960977691905788,
+      "loss": 0.9471,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019960815711586696,
+      "loss": 0.6767,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019960653396436007,
+      "loss": 0.6551,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019960490746459181,
+      "loss": 0.9142,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019960327761661684,
+      "loss": 0.916,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019960164442048994,
+      "loss": 0.7945,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199600007876266,
+      "loss": 0.826,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019959836798400006,
+      "loss": 1.0404,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995967247437472,
+      "loss": 0.8519,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019959507815556275,
+      "loss": 0.7376,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019959342821950196,
+      "loss": 0.6546,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019959177493562033,
+      "loss": 0.8089,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995901183039734,
+      "loss": 0.8774,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019958845832461695,
+      "loss": 0.6759,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995867949976067,
+      "loss": 0.9788,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019958512832299858,
+      "loss": 0.6422,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019958345830084858,
+      "loss": 0.6903,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019958178493121294,
+      "loss": 0.8474,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019958010821414783,
+      "loss": 0.7654,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995784281497096,
+      "loss": 0.6064,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995767447379548,
+      "loss": 0.7496,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019957505797893994,
+      "loss": 0.8736,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019957336787272177,
+      "loss": 0.562,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019957167441935706,
+      "loss": 0.6875,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956997761890277,
+      "loss": 0.6671,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956827747141594,
+      "loss": 0.8787,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995665739769537,
+      "loss": 0.7646,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995648671355733,
+      "loss": 0.4905,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956315694733218,
+      "loss": 0.6571,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019956144341228774,
+      "loss": 0.9392,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995597265304977,
+      "loss": 0.8,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995580063020196,
+      "loss": 0.7057,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019955628272691142,
+      "loss": 0.5937,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019955455580523103,
+      "loss": 0.7622,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995528255370365,
+      "loss": 0.8164,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019955109192238598,
+      "loss": 0.4461,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019954935496133773,
+      "loss": 1.0142,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019954761465395017,
+      "loss": 0.7899,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019954587100028177,
+      "loss": 0.6982,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019954412400039119,
+      "loss": 0.6787,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995423736543371,
+      "loss": 0.491,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019954061996217836,
+      "loss": 0.9724,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019953886292397392,
+      "loss": 0.8193,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019953710253978285,
+      "loss": 0.8382,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019953533880966436,
+      "loss": 0.7396,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019953357173367763,
+      "loss": 0.8085,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001995318013118822,
+      "loss": 0.5484,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019953002754433743,
+      "loss": 0.8647,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019952825043110308,
+      "loss": 0.7887,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019952646997223883,
+      "loss": 0.877,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019952468616780448,
+      "loss": 0.796,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995228990178601,
+      "loss": 0.6093,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995211085224657,
+      "loss": 0.8335,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019951931468168148,
+      "loss": 0.7003,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019951751749556774,
+      "loss": 0.7864,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019951571696418484,
+      "loss": 0.5817,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019951391308759342,
+      "loss": 0.8349,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199512105865854,
+      "loss": 0.9724,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995102952990274,
+      "loss": 0.8377,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019950848138717447,
+      "loss": 0.8936,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019950666413035613,
+      "loss": 0.9389,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019950484352863359,
+      "loss": 0.7233,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019950301958206792,
+      "loss": 0.7883,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001995011922907205,
+      "loss": 0.7514,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019949936165465274,
+      "loss": 0.7304,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019949752767392618,
+      "loss": 0.687,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019949569034860245,
+      "loss": 0.582,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019949384967874335,
+      "loss": 0.8362,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019949200566441068,
+      "loss": 0.6451,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019949015830566654,
+      "loss": 0.9915,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019948830760257291,
+      "loss": 1.0358,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001994864535551921,
+      "loss": 0.5925,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019948459616358634,
+      "loss": 0.7669,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019948273542781816,
+      "loss": 0.9437,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019948087134795005,
+      "loss": 0.8197,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019947900392404465,
+      "loss": 0.7709,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019947713315616482,
+      "loss": 0.8589,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019947525904437336,
+      "loss": 0.6802,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001994733815887333,
+      "loss": 0.738,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019947150078930776,
+      "loss": 0.8395,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946961664615997,
+      "loss": 0.898,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946772915935322,
+      "loss": 0.6227,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946583832895103,
+      "loss": 0.6,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946394415501688,
+      "loss": 1.07,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946204663761453,
+      "loss": 0.886,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019946014577680766,
+      "loss": 0.8715,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019945824157266025,
+      "loss": 0.9691,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019945633402523627,
+      "loss": 0.8445,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019945442313459985,
+      "loss": 0.6285,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019945250890081523,
+      "loss": 0.6824,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019945059132394675,
+      "loss": 0.6677,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001994486704040589,
+      "loss": 0.853,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019944674614121618,
+      "loss": 1.0018,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019944481853548335,
+      "loss": 0.6742,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001994428875869252,
+      "loss": 0.5279,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019944095329560657,
+      "loss": 0.6715,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019943901566159253,
+      "loss": 0.5386,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019943707468494823,
+      "loss": 0.798,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019943513036573885,
+      "loss": 0.7784,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019943318270402983,
+      "loss": 0.9894,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019943123169988662,
+      "loss": 0.7868,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019942927735337477,
+      "loss": 0.7381,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019942731966455997,
+      "loss": 0.5292,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019942535863350807,
+      "loss": 0.7065,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019942339426028496,
+      "loss": 0.4654,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001994214265449567,
+      "loss": 0.5724,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019941945548758935,
+      "loss": 0.791,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019941748108824928,
+      "loss": 0.7383,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019941550334700282,
+      "loss": 0.6325,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994135222639164,
+      "loss": 0.7126,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994115378390567,
+      "loss": 0.8377,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940955007249035,
+      "loss": 1.0667,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940755896428422,
+      "loss": 0.7769,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001994055645145052,
+      "loss": 0.569,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940356672322037,
+      "loss": 0.718,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019940156559049687,
+      "loss": 0.8333,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939956111640197,
+      "loss": 0.9863,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939755330100303,
+      "loss": 0.9847,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019939554214436756,
+      "loss": 0.8955,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993935276465632,
+      "loss": 0.782,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993915098076576,
+      "loss": 0.8171,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938948862771865,
+      "loss": 0.8673,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938746410681427,
+      "loss": 0.6957,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938543624501245,
+      "loss": 1.003,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938340504238147,
+      "loss": 0.5508,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019938137049898953,
+      "loss": 1.1448,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937933261490504,
+      "loss": 0.7671,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993772913901965,
+      "loss": 0.7837,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937524682493254,
+      "loss": 0.849,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937319891918186,
+      "loss": 0.8104,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019937114767301336,
+      "loss": 0.8536,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936909308649594,
+      "loss": 0.8094,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936703515969866,
+      "loss": 0.738,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936497389269072,
+      "loss": 0.8173,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993629092855414,
+      "loss": 0.7628,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019936084133832013,
+      "loss": 0.7178,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935877005109639,
+      "loss": 0.8257,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993566954239398,
+      "loss": 0.7815,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019935461745692012,
+      "loss": 0.683,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993525361501072,
+      "loss": 0.7589,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199350451503571,
+      "loss": 0.7221,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019934836351738155,
+      "loss": 0.6757,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993462721916091,
+      "loss": 1.0615,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019934417752632394,
+      "loss": 0.6733,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019934207952159648,
+      "loss": 0.7157,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993399781774972,
+      "loss": 0.9119,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993378734940968,
+      "loss": 0.7672,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199335765471466,
+      "loss": 1.1473,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933365410967566,
+      "loss": 0.691,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019933153940879676,
+      "loss": 0.9763,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019932942136890037,
+      "loss": 0.8682,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019932729999005768,
+      "loss": 0.8139,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019932517527234002,
+      "loss": 0.4509,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019932304721581882,
+      "loss": 0.8985,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993209158205656,
+      "loss": 0.7388,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019931878108665204,
+      "loss": 0.8798,
+      "step": 647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019931664301414982,
+      "loss": 0.9034,
+      "step": 648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993145016031309,
+      "loss": 0.7005,
+      "step": 649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019931235685366719,
+      "loss": 1.0111,
+      "step": 650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993102087658308,
+      "loss": 0.9467,
+      "step": 651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199308057339694,
+      "loss": 0.873,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019930590257532908,
+      "loss": 1.1138,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019930374447280845,
+      "loss": 0.7459,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001993015830322046,
+      "loss": 0.7837,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019929941825359032,
+      "loss": 0.6505,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019929725013703828,
+      "loss": 0.8308,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992950786826214,
+      "loss": 0.914,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929290389041263,
+      "loss": 0.7682,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019929072576048515,
+      "loss": 0.7642,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019928854429291214,
+      "loss": 0.922,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019928635948776688,
+      "loss": 0.6691,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019928417134512286,
+      "loss": 0.7843,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019928197986505367,
+      "loss": 0.7682,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992797850476329,
+      "loss": 0.8448,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019927758689293438,
+      "loss": 0.9056,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019927538540103198,
+      "loss": 0.8164,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019927318057199974,
+      "loss": 0.7328,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019927097240591171,
+      "loss": 0.7196,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019926876090284218,
+      "loss": 0.699,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019926654606286544,
+      "loss": 0.7424,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019926432788605595,
+      "loss": 0.7146,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992621063724883,
+      "loss": 0.7275,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019925988152223715,
+      "loss": 0.8806,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992576533353773,
+      "loss": 0.8955,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019925542181198363,
+      "loss": 0.7847,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992531869521312,
+      "loss": 0.7662,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019925094875589505,
+      "loss": 0.8311,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924870722335048,
+      "loss": 0.7992,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924646235457282,
+      "loss": 0.7839,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924421414963754,
+      "loss": 0.8667,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019924196260862021,
+      "loss": 0.5507,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992397077315965,
+      "loss": 0.6423,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019923744951864223,
+      "loss": 0.4879,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992351879698333,
+      "loss": 0.6594,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992329230852457,
+      "loss": 0.5868,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019923065486495564,
+      "loss": 0.8982,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992283833090393,
+      "loss": 0.8859,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019922610841757306,
+      "loss": 1.0729,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992238301906334,
+      "loss": 0.9366,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019922154862829687,
+      "loss": 0.6497,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992192637306402,
+      "loss": 0.7765,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992169754977402,
+      "loss": 0.9066,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019921468392967374,
+      "loss": 0.6269,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019921238902651788,
+      "loss": 0.9968,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019921009078834978,
+      "loss": 0.7204,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920778921524667,
+      "loss": 1.0817,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920548430728597,
+      "loss": 1.0722,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019920317606454506,
+      "loss": 0.8469,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001992008644871016,
+      "loss": 0.7491,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991985495750333,
+      "loss": 0.4694,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019919623132841794,
+      "loss": 0.769,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019919390974733347,
+      "loss": 0.9488,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019919158483185792,
+      "loss": 0.824,
+      "step": 704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019918925658206943,
+      "loss": 1.0538,
+      "step": 705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991869249980463,
+      "loss": 0.8402,
+      "step": 706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019918459007986687,
+      "loss": 0.6452,
+      "step": 707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991822518276097,
+      "loss": 0.7535,
+      "step": 708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019917991024135325,
+      "loss": 0.6695,
+      "step": 709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019917756532117635,
+      "loss": 1.0074,
+      "step": 710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991752170671578,
+      "loss": 0.7899,
+      "step": 711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001991728654793765,
+      "loss": 0.6308,
+      "step": 712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019917051055791155,
+      "loss": 0.7236,
+      "step": 713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019916815230284208,
+      "loss": 0.5642,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019916579071424734,
+      "loss": 0.8199,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019916342579220678,
+      "loss": 0.8916,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019916105753679984,
+      "loss": 0.8728,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019915868594810613,
+      "loss": 0.7645,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001991563110262054,
+      "loss": 0.7251,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001991539327711775,
+      "loss": 0.6153,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001991515511831023,
+      "loss": 0.9396,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019914916626205992,
+      "loss": 0.7587,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001991467780081305,
+      "loss": 0.8474,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019914438642139434,
+      "loss": 0.694,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019914199150193184,
+      "loss": 0.7422,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019913959324982346,
+      "loss": 1.0007,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019913719166514987,
+      "loss": 0.8399,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019913478674799174,
+      "loss": 0.6413,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019913237849842998,
+      "loss": 0.8584,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019912996691654548,
+      "loss": 0.8304,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019912755200241937,
+      "loss": 0.778,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019912513375613277,
+      "loss": 0.5365,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199122712177767,
+      "loss": 0.77,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019912028726740345,
+      "loss": 0.7586,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019911785902512361,
+      "loss": 0.7581,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019911542745100914,
+      "loss": 0.7664,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019911299254514177,
+      "loss": 0.8505,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019911055430760338,
+      "loss": 0.7961,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019910811273847581,
+      "loss": 0.7992,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001991056678378413,
+      "loss": 0.7703,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019910321960578194,
+      "loss": 0.5933,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019910076804238004,
+      "loss": 1.1134,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019909831314771797,
+      "loss": 0.6413,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019909585492187832,
+      "loss": 0.7009,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001990933933649437,
+      "loss": 0.7361,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019909092847699683,
+      "loss": 0.9548,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001990884602581206,
+      "loss": 0.9132,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019908598870839797,
+      "loss": 0.9389,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199083513827912,
+      "loss": 0.8056,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019908103561674595,
+      "loss": 0.9052,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019907855407498302,
+      "loss": 0.5638,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019907606920270668,
+      "loss": 0.6155,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001990735810000005,
+      "loss": 0.9601,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019907108946694803,
+      "loss": 0.6563,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019906859460363307,
+      "loss": 0.7896,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019906609641013952,
+      "loss": 0.6224,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019906359488655132,
+      "loss": 0.7489,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019906109003295255,
+      "loss": 0.8361,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019905858184942742,
+      "loss": 0.8485,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019905607033606027,
+      "loss": 0.5607,
+      "step": 760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019905355549293546,
+      "loss": 0.5448,
+      "step": 761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019905103732013757,
+      "loss": 0.7412,
+      "step": 762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019904851581775126,
+      "loss": 0.8426,
+      "step": 763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019904599098586129,
+      "loss": 0.9433,
+      "step": 764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019904346282455246,
+      "loss": 0.8418,
+      "step": 765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019904093133390983,
+      "loss": 0.6798,
+      "step": 766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019903839651401852,
+      "loss": 0.8856,
+      "step": 767
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019903585836496363,
+      "loss": 0.7564,
+      "step": 768
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019903331688683057,
+      "loss": 0.6687,
+      "step": 769
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019903077207970475,
+      "loss": 0.7412,
+      "step": 770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019902822394367168,
+      "loss": 0.6754,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019902567247881707,
+      "loss": 0.9128,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019902311768522663,
+      "loss": 0.754,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019902055956298628,
+      "loss": 0.7924,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019901799811218202,
+      "loss": 0.8147,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019901543333289992,
+      "loss": 0.7663,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990128652252262,
+      "loss": 0.6846,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019901029378924717,
+      "loss": 0.7127,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990077190250493,
+      "loss": 0.8204,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019900514093271915,
+      "loss": 0.8834,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019900255951234337,
+      "loss": 0.6147,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989999747640087,
+      "loss": 0.6721,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989973866878021,
+      "loss": 0.8111,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019899479528381049,
+      "loss": 0.8818,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000198992200552121,
+      "loss": 0.7874,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019898960249282088,
+      "loss": 0.7466,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019898700110599745,
+      "loss": 0.9018,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019898439639173815,
+      "loss": 0.7666,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989817883501305,
+      "loss": 0.6935,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019897917698126225,
+      "loss": 0.8645,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019897656228522114,
+      "loss": 0.6968,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019897394426209505,
+      "loss": 0.6979,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000198971322911972,
+      "loss": 0.8506,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019896869823494008,
+      "loss": 0.646,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989660702310876,
+      "loss": 0.8884,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989634389005028,
+      "loss": 0.9244,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019896080424327414,
+      "loss": 0.8265,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019895816625949028,
+      "loss": 0.7816,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989555249492398,
+      "loss": 0.6012,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019895288031261153,
+      "loss": 0.8176,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019895023234969435,
+      "loss": 0.8036,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019894758106057728,
+      "loss": 0.8129,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019894492644534944,
+      "loss": 1.0069,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019894226850410008,
+      "loss": 0.4328,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989396072369185,
+      "loss": 0.5824,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989369426438942,
+      "loss": 0.8343,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019893427472511674,
+      "loss": 0.5442,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019893160348067584,
+      "loss": 0.8079,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019892892891066124,
+      "loss": 0.7785,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019892625101516282,
+      "loss": 0.6889,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989235697942707,
+      "loss": 0.7491,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989208852480749,
+      "loss": 0.6711,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891819737666576,
+      "loss": 0.7522,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891550618013354,
+      "loss": 0.7636,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891281165856873,
+      "loss": 0.8172,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891011381206194,
+      "loss": 0.7248,
+      "step": 816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019890741264070387,
+      "loss": 0.8014,
+      "step": 817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019890470814458525,
+      "loss": 0.7319,
+      "step": 818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019890200032379705,
+      "loss": 0.5669,
+      "step": 819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019889928917843025,
+      "loss": 0.9916,
+      "step": 820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019889657470857603,
+      "loss": 0.9812,
+      "step": 821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988938569143256,
+      "loss": 0.5064,
+      "step": 822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019889113579577036,
+      "loss": 0.8325,
+      "step": 823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019888841135300172,
+      "loss": 0.9372,
+      "step": 824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988856835861113,
+      "loss": 0.8401,
+      "step": 825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988829524951908,
+      "loss": 0.7735,
+      "step": 826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000198880218080332,
+      "loss": 0.8699,
+      "step": 827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019887748034162683,
+      "loss": 0.6425,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019887473927916732,
+      "loss": 0.8382,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019887199489304562,
+      "loss": 0.862,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019886924718335398,
+      "loss": 0.7198,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019886649615018473,
+      "loss": 0.7775,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001988637417936304,
+      "loss": 0.7226,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019886098411378352,
+      "loss": 0.8872,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019885822311073683,
+      "loss": 0.5678,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019885545878458314,
+      "loss": 0.4021,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019885269113541532,
+      "loss": 0.8964,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001988499201633265,
+      "loss": 0.6716,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019884714586840973,
+      "loss": 0.7754,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001988443682507583,
+      "loss": 0.6182,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019884158731046564,
+      "loss": 0.6571,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019883880304762515,
+      "loss": 0.7338,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019883601546233045,
+      "loss": 0.682,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019883322455467527,
+      "loss": 0.6045,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019883043032475336,
+      "loss": 0.6104,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001988276327726587,
+      "loss": 0.8601,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019882483189848533,
+      "loss": 0.8798,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019882202770232739,
+      "loss": 0.9044,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019881922018427913,
+      "loss": 0.4691,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019881640934443494,
+      "loss": 0.7295,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019881359518288932,
+      "loss": 0.3874,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019881077769973683,
+      "loss": 0.6201,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019880795689507222,
+      "loss": 0.821,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019880513276899028,
+      "loss": 0.6127,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019880230532158595,
+      "loss": 0.6227,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987994745529543,
+      "loss": 0.8701,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019879664046319043,
+      "loss": 0.5157,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019879380305238965,
+      "loss": 1.1488,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019879096232064736,
+      "loss": 0.8465,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000198788118268059,
+      "loss": 0.8322,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987852708947202,
+      "loss": 0.4374,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019878242020072664,
+      "loss": 1.0945,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019877956618617423,
+      "loss": 1.0049,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987767088511588,
+      "loss": 0.8623,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019877384819577646,
+      "loss": 0.7347,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987709842201234,
+      "loss": 0.8531,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987681169242958,
+      "loss": 0.4244,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019876524630839013,
+      "loss": 0.4932,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019876237237250282,
+      "loss": 0.9565,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019875949511673056,
+      "loss": 0.7818,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019875661454116995,
+      "loss": 0.8351,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019875373064591795,
+      "loss": 0.7194,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987508434310714,
+      "loss": 0.8557,
+      "step": 873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987479528967274,
+      "loss": 0.8535,
+      "step": 874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019874505904298313,
+      "loss": 0.6261,
+      "step": 875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019874216186993585,
+      "loss": 0.8137,
+      "step": 876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987392613776829,
+      "loss": 0.6698,
+      "step": 877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019873635756632185,
+      "loss": 0.7802,
+      "step": 878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987334504359503,
+      "loss": 0.8242,
+      "step": 879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987305399866659,
+      "loss": 0.5062,
+      "step": 880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987276262185666,
+      "loss": 0.6498,
+      "step": 881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019872470913175026,
+      "loss": 0.9414,
+      "step": 882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019872178872631498,
+      "loss": 0.5724,
+      "step": 883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987188650023589,
+      "loss": 0.8669,
+      "step": 884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019871593795998033,
+      "loss": 0.879,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019871300759927762,
+      "loss": 0.8564,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987100739203493,
+      "loss": 0.6158,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019870713692329403,
+      "loss": 0.8058,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001987041966082104,
+      "loss": 0.7045,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019870125297519745,
+      "loss": 0.6389,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019869830602435394,
+      "loss": 1.0022,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019869535575577904,
+      "loss": 0.785,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986924021695719,
+      "loss": 0.7706,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986894452658318,
+      "loss": 0.618,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019868648504465814,
+      "loss": 0.7717,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986835215061504,
+      "loss": 0.798,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019868055465040825,
+      "loss": 0.6867,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019867758447753136,
+      "loss": 0.577,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019867461098761963,
+      "loss": 0.7693,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019867163418077298,
+      "loss": 0.7573,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019866865405709146,
+      "loss": 0.7467,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986656706166753,
+      "loss": 0.7835,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019866268385962475,
+      "loss": 0.785,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986596937860402,
+      "loss": 0.4763,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986567003960222,
+      "loss": 0.8461,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019865370368967134,
+      "loss": 0.8391,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019865070366708836,
+      "loss": 0.721,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986477003283741,
+      "loss": 0.855,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019864469367362954,
+      "loss": 0.7115,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019864168370295574,
+      "loss": 1.0809,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019863867041645386,
+      "loss": 0.6812,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986356538142252,
+      "loss": 0.9615,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019863263389637118,
+      "loss": 0.7851,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019862961066299332,
+      "loss": 0.7698,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986265841141932,
+      "loss": 0.8155,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986235542500726,
+      "loss": 0.8602,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019862052107073332,
+      "loss": 0.9291,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019861748457627738,
+      "loss": 0.7113,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019861444476680683,
+      "loss": 0.6304,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019861140164242384,
+      "loss": 0.7359,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001986083552032307,
+      "loss": 0.6221,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019860530544932984,
+      "loss": 0.9207,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019860225238082377,
+      "loss": 0.7281,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001985991959978151,
+      "loss": 0.6007,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001985961363004066,
+      "loss": 0.6608,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001985930732887011,
+      "loss": 0.7813,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019859000696280155,
+      "loss": 0.8407,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019858693732281106,
+      "loss": 0.7505,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019858386436883278,
+      "loss": 0.6738,
+      "step": 929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019858078810097002,
+      "loss": 0.8052,
+      "step": 930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019857770851932624,
+      "loss": 0.661,
+      "step": 931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019857462562400486,
+      "loss": 0.6591,
+      "step": 932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019857153941510956,
+      "loss": 0.8069,
+      "step": 933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001985684498927441,
+      "loss": 0.6854,
+      "step": 934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019856535705701237,
+      "loss": 0.6256,
+      "step": 935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019856226090801823,
+      "loss": 0.7561,
+      "step": 936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019855916144586583,
+      "loss": 0.3545,
+      "step": 937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019855605867065933,
+      "loss": 0.8789,
+      "step": 938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019855295258250304,
+      "loss": 0.6478,
+      "step": 939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019854984318150136,
+      "loss": 0.5945,
+      "step": 940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001985467304677588,
+      "loss": 1.0899,
+      "step": 941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019854361444138007,
+      "loss": 0.632,
+      "step": 942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019854049510246982,
+      "loss": 0.8485,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019853737245113296,
+      "loss": 0.6736,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019853424648747447,
+      "loss": 0.8355,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985311172115994,
+      "loss": 0.6636,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985279846236129,
+      "loss": 1.2306,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019852484872362033,
+      "loss": 0.7098,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985217095117271,
+      "loss": 0.7661,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019851856698803873,
+      "loss": 0.6555,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985154211526608,
+      "loss": 0.8823,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019851227200569916,
+      "loss": 0.821,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985091195472596,
+      "loss": 0.9121,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985059637774481,
+      "loss": 0.9078,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001985028046963707,
+      "loss": 0.6339,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984996423041337,
+      "loss": 0.5531,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019849647660084328,
+      "loss": 0.575,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019849330758660595,
+      "loss": 0.7612,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019849013526152817,
+      "loss": 0.8145,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019848695962571664,
+      "loss": 0.8656,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019848378067927806,
+      "loss": 0.8107,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984805984223193,
+      "loss": 0.979,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019847741285494735,
+      "loss": 0.8652,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019847422397726926,
+      "loss": 0.729,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019847103178939225,
+      "loss": 0.7991,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984678362914236,
+      "loss": 0.7053,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984646374834708,
+      "loss": 0.8035,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019846143536564125,
+      "loss": 0.599,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984582299380427,
+      "loss": 0.8939,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019845502120078285,
+      "loss": 0.7244,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019845180915396956,
+      "loss": 0.8188,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019844859379771084,
+      "loss": 0.8177,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019844537513211474,
+      "loss": 0.6421,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019844215315728946,
+      "loss": 0.4589,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984389278733433,
+      "loss": 0.8798,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984356992803847,
+      "loss": 0.5024,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019843246737852216,
+      "loss": 0.8096,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019842923216786433,
+      "loss": 0.7358,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019842599364851998,
+      "loss": 0.7851,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019842275182059793,
+      "loss": 0.7571,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984195066842072,
+      "loss": 0.7295,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984162582394569,
+      "loss": 0.8569,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019841300648645613,
+      "loss": 0.8565,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019840975142531426,
+      "loss": 0.732,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001984064930561407,
+      "loss": 0.8565,
+      "step": 985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019840323137904498,
+      "loss": 0.8045,
+      "step": 986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019839996639413676,
+      "loss": 0.6276,
+      "step": 987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019839669810152577,
+      "loss": 0.8999,
+      "step": 988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019839342650132183,
+      "loss": 0.854,
+      "step": 989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000198390151593635,
+      "loss": 0.8563,
+      "step": 990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019838687337857532,
+      "loss": 0.5858,
+      "step": 991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000198383591856253,
+      "loss": 0.9544,
+      "step": 992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019838030702677833,
+      "loss": 0.7202,
+      "step": 993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001983770188902617,
+      "loss": 1.0348,
+      "step": 994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019837372744681373,
+      "loss": 0.8691,
+      "step": 995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000198370432696545,
+      "loss": 0.6115,
+      "step": 996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019836713463956625,
+      "loss": 0.6828,
+      "step": 997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019836383327598842,
+      "loss": 0.8588,
+      "step": 998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019836052860592237,
+      "loss": 1.0112,
+      "step": 999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001983572206294793,
+      "loss": 0.7231,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019835390934677032,
+      "loss": 0.6882,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983505947579068,
+      "loss": 0.8406,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019834727686300009,
+      "loss": 0.8227,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983439556621618,
+      "loss": 0.5856,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019834063115550351,
+      "loss": 0.8464,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019833730334313702,
+      "loss": 1.038,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019833397222517415,
+      "loss": 0.7166,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019833063780172694,
+      "loss": 0.5489,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019832730007290737,
+      "loss": 1.2114,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019832395903882777,
+      "loss": 0.9471,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019832061469960034,
+      "loss": 0.7228,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019831726705533755,
+      "loss": 0.6002,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019831391610615194,
+      "loss": 0.7869,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019831056185215607,
+      "loss": 0.7003,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001983072042934628,
+      "loss": 0.7489,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019830384343018495,
+      "loss": 0.9682,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019830047926243548,
+      "loss": 0.7045,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019829711179032749,
+      "loss": 0.926,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982937410139742,
+      "loss": 0.8598,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019829036693348886,
+      "loss": 0.6375,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019828698954898495,
+      "loss": 0.6854,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019828360886057594,
+      "loss": 0.8388,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019828022486837555,
+      "loss": 0.6097,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019827683757249744,
+      "loss": 0.6486,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019827344697305558,
+      "loss": 0.7766,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982700530701638,
+      "loss": 0.8723,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019826665586393635,
+      "loss": 0.5694,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019826325535448732,
+      "loss": 0.7655,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019825985154193107,
+      "loss": 0.9196,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019825644442638196,
+      "loss": 0.7941,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019825303400795456,
+      "loss": 0.7782,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019824962028676355,
+      "loss": 0.6383,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019824620326292357,
+      "loss": 0.6986,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982427829365496,
+      "loss": 0.6089,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019823935930775655,
+      "loss": 1.0154,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019823593237665952,
+      "loss": 0.6429,
+      "step": 1036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982325021433737,
+      "loss": 0.7894,
+      "step": 1037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019822906860801441,
+      "loss": 0.7868,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019822563177069707,
+      "loss": 1.0572,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019822219163153718,
+      "loss": 0.5919,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001982187481906504,
+      "loss": 0.6477,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019821530144815245,
+      "loss": 0.5849,
+      "step": 1042
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019821185140415928,
+      "loss": 0.6862,
+      "step": 1043
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019820839805878676,
+      "loss": 0.8403,
+      "step": 1044
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019820494141215104,
+      "loss": 0.8799,
+      "step": 1045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019820148146436827,
+      "loss": 0.642,
+      "step": 1046
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019819801821555479,
+      "loss": 0.7884,
+      "step": 1047
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000198194551665827,
+      "loss": 0.8503,
+      "step": 1048
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019819108181530143,
+      "loss": 1.0032,
+      "step": 1049
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019818760866409477,
+      "loss": 0.759,
+      "step": 1050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019818413221232365,
+      "loss": 0.6735,
+      "step": 1051
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019818065246010504,
+      "loss": 0.7231,
+      "step": 1052
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019817716940755586,
+      "loss": 0.7521,
+      "step": 1053
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019817368305479323,
+      "loss": 0.6835,
+      "step": 1054
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019817019340193431,
+      "loss": 0.7459,
+      "step": 1055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019816670044909638,
+      "loss": 0.8736,
+      "step": 1056
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001981632041963969,
+      "loss": 0.7302,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981597046439534,
+      "loss": 0.8067,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019815620179188347,
+      "loss": 0.665,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019815269564030493,
+      "loss": 0.6128,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019814918618933557,
+      "loss": 0.6871,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981456734390934,
+      "loss": 0.6979,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981421573896965,
+      "loss": 0.3618,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019813863804126302,
+      "loss": 0.6226,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019813511539391133,
+      "loss": 0.6349,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019813158944775974,
+      "loss": 0.745,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981280602029269,
+      "loss": 0.7407,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019812452765953135,
+      "loss": 0.5896,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981209918176919,
+      "loss": 0.7959,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019811745267752734,
+      "loss": 0.3949,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981139102391567,
+      "loss": 0.6641,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019811036450269906,
+      "loss": 0.9141,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019810681546827353,
+      "loss": 0.6788,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001981032631359995,
+      "loss": 0.7084,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019809970750599636,
+      "loss": 0.6544,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019809614857838357,
+      "loss": 0.7013,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019809258635328084,
+      "loss": 0.8141,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019808902083080787,
+      "loss": 0.6952,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019808545201108454,
+      "loss": 0.6244,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001980818798942308,
+      "loss": 0.7944,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019807830448036672,
+      "loss": 0.6225,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001980747257696125,
+      "loss": 0.7635,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019807114376208844,
+      "loss": 0.5617,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019806755845791496,
+      "loss": 0.6949,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019806396985721253,
+      "loss": 0.5059,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019806037796010184,
+      "loss": 0.7212,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019805678276670362,
+      "loss": 0.7137,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019805318427713867,
+      "loss": 0.7293,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000198049582491528,
+      "loss": 0.5604,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019804597740999267,
+      "loss": 0.5351,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019804236903265388,
+      "loss": 0.9533,
+      "step": 1091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001980387573596329,
+      "loss": 0.9348,
+      "step": 1092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019803514239105118,
+      "loss": 0.5396,
+      "step": 1093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019803152412703018,
+      "loss": 0.5694,
+      "step": 1094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019802790256769154,
+      "loss": 0.801,
+      "step": 1095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019802427771315704,
+      "loss": 0.6681,
+      "step": 1096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019802064956354846,
+      "loss": 0.4526,
+      "step": 1097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019801701811898785,
+      "loss": 0.7964,
+      "step": 1098
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001980133833795972,
+      "loss": 0.6595,
+      "step": 1099
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019800974534549872,
+      "loss": 0.7268,
+      "step": 1100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019800610401681472,
+      "loss": 0.6118,
+      "step": 1101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001980024593936676,
+      "loss": 0.4471,
+      "step": 1102
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019799881147617986,
+      "loss": 0.7201,
+      "step": 1103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019799516026447407,
+      "loss": 0.9311,
+      "step": 1104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019799150575867307,
+      "loss": 0.5656,
+      "step": 1105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019798784795889963,
+      "loss": 0.7345,
+      "step": 1106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019798418686527676,
+      "loss": 0.5614,
+      "step": 1107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019798052247792746,
+      "loss": 0.7733,
+      "step": 1108
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019797685479697497,
+      "loss": 0.7645,
+      "step": 1109
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019797318382254258,
+      "loss": 0.9442,
+      "step": 1110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019796950955475363,
+      "loss": 0.5531,
+      "step": 1111
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001979658319937317,
+      "loss": 0.5711,
+      "step": 1112
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019796215113960033,
+      "loss": 0.676,
+      "step": 1113
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019795846699248332,
+      "loss": 0.7002,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019795477955250447,
+      "loss": 0.7644,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001979510888197878,
+      "loss": 0.6527,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019794739479445726,
+      "loss": 0.7925,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019794369747663714,
+      "loss": 0.6809,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001979399968664517,
+      "loss": 0.4153,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019793629296402526,
+      "loss": 0.4566,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001979325857694824,
+      "loss": 0.7602,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001979288752829477,
+      "loss": 0.4373,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001979251615045459,
+      "loss": 0.8237,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019792144443440184,
+      "loss": 0.8305,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019791772407264048,
+      "loss": 0.9207,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019791400041938685,
+      "loss": 0.6175,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019791027347476617,
+      "loss": 0.5941,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019790654323890365,
+      "loss": 0.5771,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019790280971192475,
+      "loss": 0.7257,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019789907289395492,
+      "loss": 1.0078,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019789533278511982,
+      "loss": 0.8544,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019789158938554512,
+      "loss": 0.6674,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019788784269535672,
+      "loss": 0.6991,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001978840927146805,
+      "loss": 0.7898,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019788033944364254,
+      "loss": 0.6598,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000197876582882369,
+      "loss": 0.8502,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019787282303098617,
+      "loss": 0.8207,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019786905988962046,
+      "loss": 0.5353,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001978652934583983,
+      "loss": 0.8131,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019786152373744636,
+      "loss": 0.7868,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001978577507268913,
+      "loss": 0.929,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019785397442686003,
+      "loss": 0.8376,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019785019483747944,
+      "loss": 0.9897,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019784641195887657,
+      "loss": 0.5486,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001978426257911786,
+      "loss": 0.649,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019783883633451278,
+      "loss": 0.7654,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019783504358900653,
+      "loss": 0.9198,
+      "step": 1147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019783124755478734,
+      "loss": 0.691,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019782744823198275,
+      "loss": 0.7043,
+      "step": 1149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019782364562072053,
+      "loss": 0.9345,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019781983972112853,
+      "loss": 0.843,
+      "step": 1151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019781603053333461,
+      "loss": 0.6836,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019781221805746687,
+      "loss": 0.8437,
+      "step": 1153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019780840229365344,
+      "loss": 0.6996,
+      "step": 1154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001978045832420226,
+      "loss": 0.3419,
+      "step": 1155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019780076090270275,
+      "loss": 0.7217,
+      "step": 1156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977969352758223,
+      "loss": 0.9782,
+      "step": 1157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019779310636150994,
+      "loss": 0.7575,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977892741598943,
+      "loss": 0.8653,
+      "step": 1159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019778543867110426,
+      "loss": 0.967,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019778159989526875,
+      "loss": 0.8151,
+      "step": 1161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019777775783251677,
+      "loss": 0.7034,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019777391248297743,
+      "loss": 0.8274,
+      "step": 1163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977700638467801,
+      "loss": 0.4245,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977662119240541,
+      "loss": 0.964,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977623567149289,
+      "loss": 0.7527,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977584982195341,
+      "loss": 0.5225,
+      "step": 1167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977546364379994,
+      "loss": 0.6716,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019775077137045463,
+      "loss": 0.6995,
+      "step": 1169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001977469030170297,
+      "loss": 0.6838,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019774303137785464,
+      "loss": 0.8004,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001977391564530596,
+      "loss": 0.8444,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001977352782427748,
+      "loss": 0.8158,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019773139674713074,
+      "loss": 0.8728,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019772751196625774,
+      "loss": 0.761,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019772362390028643,
+      "loss": 0.7175,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019771973254934753,
+      "loss": 1.0432,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019771583791357186,
+      "loss": 0.7877,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001977119399930903,
+      "loss": 0.983,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001977080387880339,
+      "loss": 0.6362,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001977041342985338,
+      "loss": 0.7501,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019770022652472122,
+      "loss": 0.9432,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019769631546672756,
+      "loss": 0.8545,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019769240112468427,
+      "loss": 0.6737,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019768848349872292,
+      "loss": 0.6464,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019768456258897522,
+      "loss": 0.9758,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019768063839557294,
+      "loss": 0.7098,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019767671091864804,
+      "loss": 0.7958,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001976727801583325,
+      "loss": 0.7124,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001976688461147585,
+      "loss": 0.6185,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019766490878805822,
+      "loss": 0.6759,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019766096817836403,
+      "loss": 0.7901,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019765702428580843,
+      "loss": 0.799,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019765307711052397,
+      "loss": 0.7831,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019764912665264333,
+      "loss": 0.879,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001976451729122993,
+      "loss": 0.8613,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019764121588962478,
+      "loss": 0.6641,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019763725558475282,
+      "loss": 0.6253,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001976332919978165,
+      "loss": 0.6883,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019762932512894913,
+      "loss": 0.7297,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019762535497828393,
+      "loss": 1.046,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019762138154595446,
+      "loss": 0.9925,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019761740483209428,
+      "loss": 0.9333,
+      "step": 1203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000197613424836837,
+      "loss": 0.7105,
+      "step": 1204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019760944156031645,
+      "loss": 0.862,
+      "step": 1205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019760545500266657,
+      "loss": 0.8108,
+      "step": 1206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019760146516402128,
+      "loss": 0.5907,
+      "step": 1207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019759747204451476,
+      "loss": 0.6492,
+      "step": 1208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019759347564428126,
+      "loss": 0.9407,
+      "step": 1209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019758947596345503,
+      "loss": 0.7778,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001975854730021706,
+      "loss": 0.8018,
+      "step": 1211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019758146676056244,
+      "loss": 0.5979,
+      "step": 1212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019757745723876532,
+      "loss": 0.5514,
+      "step": 1213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019757344443691396,
+      "loss": 0.692,
+      "step": 1214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001975694283551433,
+      "loss": 0.5045,
+      "step": 1215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019756540899358826,
+      "loss": 0.8423,
+      "step": 1216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019756138635238402,
+      "loss": 0.4799,
+      "step": 1217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019755736043166575,
+      "loss": 0.7396,
+      "step": 1218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019755333123156886,
+      "loss": 0.9076,
+      "step": 1219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001975492987522287,
+      "loss": 0.7788,
+      "step": 1220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019754526299378084,
+      "loss": 0.643,
+      "step": 1221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000197541223956361,
+      "loss": 0.8829,
+      "step": 1222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001975371816401049,
+      "loss": 0.7686,
+      "step": 1223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019753313604514846,
+      "loss": 0.8192,
+      "step": 1224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019752908717162758,
+      "loss": 0.8011,
+      "step": 1225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001975250350196785,
+      "loss": 0.4586,
+      "step": 1226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001975209795894373,
+      "loss": 0.6682,
+      "step": 1227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019751692088104043,
+      "loss": 0.9186,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019751285889462423,
+      "loss": 0.7383,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019750879363032527,
+      "loss": 0.9348,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001975047250882802,
+      "loss": 0.6121,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001975006532686258,
+      "loss": 0.8626,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019749657817149893,
+      "loss": 0.9203,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019749249979703653,
+      "loss": 0.9258,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974884181453758,
+      "loss": 0.7214,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019748433321665386,
+      "loss": 0.4795,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019748024501100802,
+      "loss": 0.7754,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974761535285758,
+      "loss": 0.7755,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974720587694946,
+      "loss": 0.7605,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019746796073390218,
+      "loss": 0.7728,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019746385942193625,
+      "loss": 0.7035,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019745975483373464,
+      "loss": 0.9279,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019745564696943535,
+      "loss": 0.638,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974515358291765,
+      "loss": 0.9414,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019744742141309627,
+      "loss": 0.42,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019744330372133294,
+      "loss": 0.695,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974391827540249,
+      "loss": 0.8239,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019743505851131077,
+      "loss": 0.6649,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974309309933291,
+      "loss": 0.788,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974268002002187,
+      "loss": 0.6962,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019742266613211835,
+      "loss": 0.8056,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974185287891671,
+      "loss": 0.5673,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019741438817150395,
+      "loss": 0.8299,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019741024427926814,
+      "loss": 0.897,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019740609711259895,
+      "loss": 0.7159,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001974019466716358,
+      "loss": 0.733,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019739779295651814,
+      "loss": 0.52,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019739363596738569,
+      "loss": 0.7754,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019738947570437817,
+      "loss": 0.5988,
+      "step": 1259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019738531216763536,
+      "loss": 1.0068,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019738114535729727,
+      "loss": 0.5948,
+      "step": 1261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019737697527350394,
+      "loss": 0.7534,
+      "step": 1262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019737280191639559,
+      "loss": 0.7807,
+      "step": 1263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019736862528611245,
+      "loss": 0.5348,
+      "step": 1264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019736444538279497,
+      "loss": 0.7131,
+      "step": 1265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019736026220658364,
+      "loss": 1.0014,
+      "step": 1266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019735607575761902,
+      "loss": 0.9075,
+      "step": 1267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019735188603604194,
+      "loss": 0.6072,
+      "step": 1268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019734769304199316,
+      "loss": 0.7871,
+      "step": 1269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019734349677561364,
+      "loss": 0.499,
+      "step": 1270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019733929723704448,
+      "loss": 0.8366,
+      "step": 1271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001973350944264268,
+      "loss": 0.8726,
+      "step": 1272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019733088834390187,
+      "loss": 0.8408,
+      "step": 1273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019732667898961115,
+      "loss": 0.6158,
+      "step": 1274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019732246636369605,
+      "loss": 0.8137,
+      "step": 1275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001973182504662982,
+      "loss": 0.7739,
+      "step": 1276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019731403129755935,
+      "loss": 0.5947,
+      "step": 1277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019730980885762132,
+      "loss": 0.7493,
+      "step": 1278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000197305583146626,
+      "loss": 0.5807,
+      "step": 1279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001973013541647155,
+      "loss": 0.5449,
+      "step": 1280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019729712191203193,
+      "loss": 0.6386,
+      "step": 1281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019729288638871756,
+      "loss": 0.8725,
+      "step": 1282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001972886475949148,
+      "loss": 0.7568,
+      "step": 1283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019728440553076613,
+      "loss": 0.6274,
+      "step": 1284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001972801601964141,
+      "loss": 0.6809,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019727591159200147,
+      "loss": 0.4502,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019727165971767104,
+      "loss": 0.6277,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019726740457356572,
+      "loss": 0.8481,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019726314615982858,
+      "loss": 0.6078,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001972588844766027,
+      "loss": 0.9675,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019725461952403142,
+      "loss": 0.935,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000197250351302258,
+      "loss": 0.9323,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019724607981142607,
+      "loss": 0.8856,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001972418050516791,
+      "loss": 0.7705,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019723752702316077,
+      "loss": 0.7445,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019723324572601495,
+      "loss": 0.5362,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019722896116038555,
+      "loss": 0.8003,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019722467332641656,
+      "loss": 1.0088,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019722038222425215,
+      "loss": 0.8163,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019721608785403655,
+      "loss": 0.6927,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001972117902159141,
+      "loss": 0.8215,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019720748931002931,
+      "loss": 0.8645,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019720318513652668,
+      "loss": 0.7615,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019719887769555094,
+      "loss": 0.7591,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001971945669872469,
+      "loss": 0.8544,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019719025301175944,
+      "loss": 0.8046,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019718593576923356,
+      "loss": 0.6792,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019718161525981442,
+      "loss": 0.7994,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019717729148364724,
+      "loss": 0.8537,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019717296444087733,
+      "loss": 0.9,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019716863413165022,
+      "loss": 0.5013,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019716430055611138,
+      "loss": 0.7432,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019715996371440654,
+      "loss": 0.3935,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019715562360668148,
+      "loss": 0.85,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019715128023308206,
+      "loss": 0.7292,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019714693359375432,
+      "loss": 0.8305,
+      "step": 1316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019714258368884436,
+      "loss": 0.6072,
+      "step": 1317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019713823051849838,
+      "loss": 0.6061,
+      "step": 1318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019713387408286273,
+      "loss": 0.7059,
+      "step": 1319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019712951438208385,
+      "loss": 0.6542,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001971251514163083,
+      "loss": 0.6149,
+      "step": 1321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001971207851856827,
+      "loss": 0.5958,
+      "step": 1322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001971164156903539,
+      "loss": 0.7858,
+      "step": 1323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001971120429304687,
+      "loss": 1.0565,
+      "step": 1324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019710766690617413,
+      "loss": 0.6944,
+      "step": 1325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019710328761761726,
+      "loss": 0.7343,
+      "step": 1326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019709890506494534,
+      "loss": 0.4842,
+      "step": 1327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019709451924830566,
+      "loss": 0.7562,
+      "step": 1328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019709013016784566,
+      "loss": 0.7181,
+      "step": 1329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019708573782371288,
+      "loss": 0.7771,
+      "step": 1330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019708134221605494,
+      "loss": 0.8011,
+      "step": 1331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019707694334501966,
+      "loss": 0.5325,
+      "step": 1332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019707254121075485,
+      "loss": 0.6843,
+      "step": 1333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019706813581340848,
+      "loss": 0.6927,
+      "step": 1334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019706372715312868,
+      "loss": 0.746,
+      "step": 1335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019705931523006365,
+      "loss": 0.7625,
+      "step": 1336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019705490004436163,
+      "loss": 0.7359,
+      "step": 1337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019705048159617114,
+      "loss": 0.7998,
+      "step": 1338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019704605988564062,
+      "loss": 0.5782,
+      "step": 1339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019704163491291875,
+      "loss": 0.645,
+      "step": 1340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019703720667815422,
+      "loss": 0.8224,
+      "step": 1341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019703277518149593,
+      "loss": 0.7668,
+      "step": 1342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019702834042309287,
+      "loss": 0.8797,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019702390240309404,
+      "loss": 0.9121,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001970194611216487,
+      "loss": 0.8676,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019701501657890606,
+      "loss": 0.9002,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001970105687750156,
+      "loss": 0.7072,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001970061177101268,
+      "loss": 0.7925,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019700166338438928,
+      "loss": 0.787,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969972057979528,
+      "loss": 0.8837,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019699274495096712,
+      "loss": 0.6852,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969882808435823,
+      "loss": 0.8949,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019698381347594833,
+      "loss": 0.57,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969793428482154,
+      "loss": 0.713,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019697486896053375,
+      "loss": 0.8234,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019697039181305385,
+      "loss": 0.6079,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019696591140592616,
+      "loss": 0.3851,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019696142773930125,
+      "loss": 0.6277,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969569408133299,
+      "loss": 0.8233,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969524506281629,
+      "loss": 0.6882,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969479571839512,
+      "loss": 0.6043,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019694346048084582,
+      "loss": 0.7527,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019693896051899796,
+      "loss": 0.8431,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019693445729855887,
+      "loss": 0.567,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969299508196799,
+      "loss": 0.7149,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019692544108251256,
+      "loss": 0.7629,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019692092808720846,
+      "loss": 0.6827,
+      "step": 1367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019691641183391926,
+      "loss": 0.8046,
+      "step": 1368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001969118923227968,
+      "loss": 0.7724,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000196907369553993,
+      "loss": 0.8194,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019690284352765988,
+      "loss": 0.6622,
+      "step": 1371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001968983142439496,
+      "loss": 0.7143,
+      "step": 1372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001968937817030144,
+      "loss": 0.7166,
+      "step": 1373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019688924590500665,
+      "loss": 0.8912,
+      "step": 1374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019688470685007883,
+      "loss": 0.7511,
+      "step": 1375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001968801645383835,
+      "loss": 0.7016,
+      "step": 1376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019687561897007333,
+      "loss": 0.6442,
+      "step": 1377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019687107014530115,
+      "loss": 0.9811,
+      "step": 1378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019686651806421988,
+      "loss": 0.7203,
+      "step": 1379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001968619627269825,
+      "loss": 0.8839,
+      "step": 1380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001968574041337422,
+      "loss": 0.7619,
+      "step": 1381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019685284228465212,
+      "loss": 0.9031,
+      "step": 1382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019684827717986566,
+      "loss": 0.6003,
+      "step": 1383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019684370881953628,
+      "loss": 0.7581,
+      "step": 1384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019683913720381754,
+      "loss": 1.0594,
+      "step": 1385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019683456233286311,
+      "loss": 0.5442,
+      "step": 1386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001968299842068268,
+      "loss": 0.6661,
+      "step": 1387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019682540282586248,
+      "loss": 0.461,
+      "step": 1388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019682081819012413,
+      "loss": 0.6196,
+      "step": 1389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019681623029976588,
+      "loss": 0.8338,
+      "step": 1390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000196811639154942,
+      "loss": 0.6399,
+      "step": 1391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019680704475580671,
+      "loss": 0.543,
+      "step": 1392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019680244710251453,
+      "loss": 0.5685,
+      "step": 1393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019679784619522003,
+      "loss": 0.6246,
+      "step": 1394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019679324203407782,
+      "loss": 0.8983,
+      "step": 1395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001967886346192427,
+      "loss": 0.8199,
+      "step": 1396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001967840239508695,
+      "loss": 0.5423,
+      "step": 1397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019677941002911326,
+      "loss": 0.6731,
+      "step": 1398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019677479285412904,
+      "loss": 0.7628,
+      "step": 1399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019677017242607207,
+      "loss": 0.6065,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019676554874509768,
+      "loss": 0.8165,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019676092181136122,
+      "loss": 0.3354,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001967562916250183,
+      "loss": 0.7851,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019675165818622454,
+      "loss": 1.0028,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001967470214951357,
+      "loss": 0.8354,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019674238155190758,
+      "loss": 0.6501,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019673773835669626,
+      "loss": 0.4355,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019673309190965772,
+      "loss": 0.5625,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001967284422109482,
+      "loss": 0.9423,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019672378926072402,
+      "loss": 0.7285,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019671913305914154,
+      "loss": 0.9975,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001967144736063573,
+      "loss": 0.6966,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019670981090252792,
+      "loss": 0.7487,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019670514494781014,
+      "loss": 0.6719,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001967004757423608,
+      "loss": 0.6918,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019669580328633688,
+      "loss": 0.9668,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966911275798954,
+      "loss": 0.5601,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966864486231936,
+      "loss": 0.67,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966817664163887,
+      "loss": 0.7487,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966770809596381,
+      "loss": 0.5622,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966723922530993,
+      "loss": 0.8823,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019666770029692996,
+      "loss": 0.7593,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019666300509128774,
+      "loss": 0.3709,
+      "step": 1423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966583066363305,
+      "loss": 0.5927,
+      "step": 1424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019665360493221617,
+      "loss": 0.7536,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019664889997910278,
+      "loss": 0.8178,
+      "step": 1426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019664419177714855,
+      "loss": 0.7843,
+      "step": 1427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019663948032651164,
+      "loss": 0.5479,
+      "step": 1428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019663476562735053,
+      "loss": 0.73,
+      "step": 1429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019663004767982365,
+      "loss": 0.8032,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966253264840896,
+      "loss": 0.7058,
+      "step": 1431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019662060204030706,
+      "loss": 0.7533,
+      "step": 1432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966158743486349,
+      "loss": 0.8796,
+      "step": 1433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019661114340923196,
+      "loss": 0.7249,
+      "step": 1434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019660640922225737,
+      "loss": 0.3943,
+      "step": 1435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001966016717878702,
+      "loss": 0.9593,
+      "step": 1436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965969311062297,
+      "loss": 0.5667,
+      "step": 1437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019659218717749526,
+      "loss": 0.6479,
+      "step": 1438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965874400018263,
+      "loss": 0.7497,
+      "step": 1439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019658268957938244,
+      "loss": 0.9215,
+      "step": 1440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019657793591032333,
+      "loss": 0.6149,
+      "step": 1441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965731789948088,
+      "loss": 0.6752,
+      "step": 1442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965684188329987,
+      "loss": 0.7324,
+      "step": 1443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019656365542505315,
+      "loss": 0.7666,
+      "step": 1444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019655888877113215,
+      "loss": 0.8901,
+      "step": 1445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019655411887139598,
+      "loss": 0.7885,
+      "step": 1446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000196549345726005,
+      "loss": 1.1367,
+      "step": 1447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019654456933511962,
+      "loss": 0.5847,
+      "step": 1448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019653978969890043,
+      "loss": 0.6977,
+      "step": 1449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019653500681750808,
+      "loss": 0.8354,
+      "step": 1450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019653022069110332,
+      "loss": 0.8881,
+      "step": 1451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965254313198471,
+      "loss": 0.6379,
+      "step": 1452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019652063870390037,
+      "loss": 0.6812,
+      "step": 1453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965158428434242,
+      "loss": 0.7321,
+      "step": 1454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019651104373857989,
+      "loss": 0.7258,
+      "step": 1455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019650624138952872,
+      "loss": 0.43,
+      "step": 1456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001965014357964321,
+      "loss": 0.5969,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019649662695945156,
+      "loss": 0.6987,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001964918148787488,
+      "loss": 0.6704,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019648699955448552,
+      "loss": 0.9197,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019648218098682364,
+      "loss": 0.7429,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001964773591759251,
+      "loss": 0.788,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019647253412195204,
+      "loss": 0.5868,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019646770582506658,
+      "loss": 0.5707,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019646287428543103,
+      "loss": 0.83,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019645803950320786,
+      "loss": 0.6972,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019645320147855953,
+      "loss": 0.7722,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001964483602116487,
+      "loss": 0.7012,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019644351570263812,
+      "loss": 0.7935,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019643866795169063,
+      "loss": 0.6466,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019643381695896915,
+      "loss": 0.8716,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019642896272463678,
+      "loss": 0.5528,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019642410524885667,
+      "loss": 0.8694,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019641924453179214,
+      "loss": 0.6565,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019641438057360654,
+      "loss": 0.8269,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019640951337446345,
+      "loss": 0.9614,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019640464293452638,
+      "loss": 0.4738,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963997692539591,
+      "loss": 0.7359,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019639489233292542,
+      "loss": 0.6745,
+      "step": 1479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019639001217158932,
+      "loss": 1.0445,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963851287701148,
+      "loss": 0.6226,
+      "step": 1481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019638024212866606,
+      "loss": 0.5991,
+      "step": 1482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963753522474073,
+      "loss": 0.6762,
+      "step": 1483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019637045912650295,
+      "loss": 0.7116,
+      "step": 1484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019636556276611745,
+      "loss": 0.9473,
+      "step": 1485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963606631664154,
+      "loss": 0.6294,
+      "step": 1486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019635576032756152,
+      "loss": 0.7527,
+      "step": 1487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963508542497206,
+      "loss": 0.9952,
+      "step": 1488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963459449330576,
+      "loss": 0.5353,
+      "step": 1489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019634103237773748,
+      "loss": 0.7008,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963361165839254,
+      "loss": 0.5759,
+      "step": 1491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019633119755178662,
+      "loss": 0.3007,
+      "step": 1492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019632627528148647,
+      "loss": 0.6295,
+      "step": 1493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019632134977319043,
+      "loss": 1.0108,
+      "step": 1494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019631642102706408,
+      "loss": 0.5674,
+      "step": 1495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019631148904327307,
+      "loss": 0.8611,
+      "step": 1496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001963065538219832,
+      "loss": 0.8375,
+      "step": 1497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019630161536336036,
+      "loss": 0.9546,
+      "step": 1498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019629667366757056,
+      "loss": 0.5587,
+      "step": 1499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019629172873477995,
+      "loss": 0.777,
+      "step": 1500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001962867805651547,
+      "loss": 0.7602,
+      "step": 1501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019628182915886115,
+      "loss": 0.6696,
+      "step": 1502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019627687451606577,
+      "loss": 0.6294,
+      "step": 1503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001962719166369351,
+      "loss": 0.9283,
+      "step": 1504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019626695552163578,
+      "loss": 0.6711,
+      "step": 1505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001962619911703346,
+      "loss": 0.5375,
+      "step": 1506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001962570235831984,
+      "loss": 0.3164,
+      "step": 1507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019625205276039423,
+      "loss": 0.4285,
+      "step": 1508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019624707870208912,
+      "loss": 0.7476,
+      "step": 1509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001962421014084503,
+      "loss": 0.7474,
+      "step": 1510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019623712087964506,
+      "loss": 0.7399,
+      "step": 1511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019623213711584089,
+      "loss": 0.6412,
+      "step": 1512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019622715011720522,
+      "loss": 0.9279,
+      "step": 1513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019622215988390575,
+      "loss": 0.2824,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001962171664161102,
+      "loss": 0.6449,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019621216971398643,
+      "loss": 0.665,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019620716977770242,
+      "loss": 0.6509,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019620216660742624,
+      "loss": 0.853,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019619716020332603,
+      "loss": 0.7976,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019619215056557014,
+      "loss": 0.4875,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019618713769432693,
+      "loss": 0.6281,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019618212158976494,
+      "loss": 0.643,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001961771022520527,
+      "loss": 0.8109,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019617207968135905,
+      "loss": 0.6839,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019616705387785275,
+      "loss": 0.6261,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019616202484170277,
+      "loss": 0.8303,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019615699257307813,
+      "loss": 0.6959,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019615195707214803,
+      "loss": 0.7496,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019614691833908172,
+      "loss": 0.7998,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019614187637404857,
+      "loss": 0.7852,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019613683117721806,
+      "loss": 0.7718,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001961317827487598,
+      "loss": 0.6489,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001961267310888435,
+      "loss": 0.7862,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019612167619763895,
+      "loss": 0.7806,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019611661807531606,
+      "loss": 0.8979,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019611155672204488,
+      "loss": 0.7107,
+      "step": 1536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019610649213799554,
+      "loss": 0.7993,
+      "step": 1537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001961014243233383,
+      "loss": 0.9208,
+      "step": 1538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960963532782435,
+      "loss": 0.9539,
+      "step": 1539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960912790028816,
+      "loss": 0.7529,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019608620149742318,
+      "loss": 0.6729,
+      "step": 1541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960811207620389,
+      "loss": 0.6834,
+      "step": 1542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019607603679689956,
+      "loss": 1.0641,
+      "step": 1543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019607094960217607,
+      "loss": 0.6726,
+      "step": 1544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960658591780394,
+      "loss": 0.6814,
+      "step": 1545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019606076552466071,
+      "loss": 0.5797,
+      "step": 1546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019605566864221123,
+      "loss": 1.0194,
+      "step": 1547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019605056853086222,
+      "loss": 0.8006,
+      "step": 1548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960454651907852,
+      "loss": 0.801,
+      "step": 1549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019604035862215164,
+      "loss": 0.6555,
+      "step": 1550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019603524882513327,
+      "loss": 0.5997,
+      "step": 1551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960301357999018,
+      "loss": 0.8051,
+      "step": 1552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960250195466291,
+      "loss": 0.8312,
+      "step": 1553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019601990006548722,
+      "loss": 0.6324,
+      "step": 1554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001960147773566482,
+      "loss": 0.5873,
+      "step": 1555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019600965142028425,
+      "loss": 0.7693,
+      "step": 1556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019600452225656767,
+      "loss": 0.6885,
+      "step": 1557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019599938986567087,
+      "loss": 0.6737,
+      "step": 1558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001959942542477664,
+      "loss": 0.69,
+      "step": 1559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019598911540302688,
+      "loss": 0.7466,
+      "step": 1560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019598397333162507,
+      "loss": 1.1076,
+      "step": 1561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019597882803373376,
+      "loss": 0.6121,
+      "step": 1562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019597367950952596,
+      "loss": 0.6531,
+      "step": 1563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019596852775917475,
+      "loss": 0.7374,
+      "step": 1564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019596337278285326,
+      "loss": 0.5047,
+      "step": 1565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001959582145807348,
+      "loss": 0.8166,
+      "step": 1566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019595305315299276,
+      "loss": 0.7448,
+      "step": 1567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019594788849980064,
+      "loss": 0.763,
+      "step": 1568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019594272062133206,
+      "loss": 0.5926,
+      "step": 1569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001959375495177607,
+      "loss": 0.6834,
+      "step": 1570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019593237518926044,
+      "loss": 0.7782,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019592719763600517,
+      "loss": 0.9419,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019592201685816895,
+      "loss": 1.119,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019591683285592593,
+      "loss": 0.5984,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001959116456294504,
+      "loss": 0.737,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019590645517891667,
+      "loss": 0.5895,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019590126150449924,
+      "loss": 0.7623,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001958960646063727,
+      "loss": 0.6754,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019589086448471176,
+      "loss": 0.9595,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019588566113969117,
+      "loss": 0.6381,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001958804545714859,
+      "loss": 0.6348,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019587524478027093,
+      "loss": 0.6563,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019587003176622141,
+      "loss": 0.7654,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019586481552951254,
+      "loss": 0.4789,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019585959607031973,
+      "loss": 0.5537,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019585437338881834,
+      "loss": 0.7667,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000195849147485184,
+      "loss": 0.8178,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019584391835959238,
+      "loss": 0.7333,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019583868601221916,
+      "loss": 0.876,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019583345044324037,
+      "loss": 0.9461,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019582821165283192,
+      "loss": 0.8852,
+      "step": 1591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019582296964116988,
+      "loss": 0.8449,
+      "step": 1592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019581772440843052,
+      "loss": 0.5933,
+      "step": 1593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019581247595479016,
+      "loss": 0.6756,
+      "step": 1594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019580722428042517,
+      "loss": 0.4773,
+      "step": 1595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019580196938551213,
+      "loss": 0.7628,
+      "step": 1596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957967112702277,
+      "loss": 0.6257,
+      "step": 1597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019579144993474857,
+      "loss": 0.5405,
+      "step": 1598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019578618537925165,
+      "loss": 0.5364,
+      "step": 1599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019578091760391386,
+      "loss": 0.448,
+      "step": 1600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019577564660891233,
+      "loss": 0.7389,
+      "step": 1601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957703723944242,
+      "loss": 0.659,
+      "step": 1602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957650949606268,
+      "loss": 0.6607,
+      "step": 1603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019575981430769752,
+      "loss": 0.8335,
+      "step": 1604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019575453043581384,
+      "loss": 0.7941,
+      "step": 1605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019574924334515338,
+      "loss": 0.3651,
+      "step": 1606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957439530358939,
+      "loss": 0.4948,
+      "step": 1607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019573865950821322,
+      "loss": 0.7429,
+      "step": 1608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019573336276228927,
+      "loss": 0.5875,
+      "step": 1609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957280627983001,
+      "loss": 0.7385,
+      "step": 1610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001957227596164239,
+      "loss": 0.4506,
+      "step": 1611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019571745321683888,
+      "loss": 0.815,
+      "step": 1612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019571214359972343,
+      "loss": 0.6231,
+      "step": 1613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019570683076525604,
+      "loss": 0.9732,
+      "step": 1614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019570151471361532,
+      "loss": 1.0063,
+      "step": 1615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019569619544498,
+      "loss": 0.9787,
+      "step": 1616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019569087295952877,
+      "loss": 0.6684,
+      "step": 1617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019568554725744064,
+      "loss": 0.4639,
+      "step": 1618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001956802183388946,
+      "loss": 0.6805,
+      "step": 1619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019567488620406983,
+      "loss": 0.8489,
+      "step": 1620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019566955085314547,
+      "loss": 0.5071,
+      "step": 1621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019566421228630097,
+      "loss": 0.9618,
+      "step": 1622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001956588705037157,
+      "loss": 0.6408,
+      "step": 1623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019565352550556927,
+      "loss": 0.4381,
+      "step": 1624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019564817729204138,
+      "loss": 0.7761,
+      "step": 1625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019564282586331173,
+      "loss": 0.6612,
+      "step": 1626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019563747121956028,
+      "loss": 0.5248,
+      "step": 1627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019563211336096696,
+      "loss": 0.7366,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019562675228771196,
+      "loss": 0.7408,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001956213879999754,
+      "loss": 0.7406,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019561602049793768,
+      "loss": 0.7268,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019561064978177917,
+      "loss": 1.0472,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019560527585168043,
+      "loss": 0.7709,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955998987078221,
+      "loss": 0.791,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019559451835038494,
+      "loss": 0.5354,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019558913477954977,
+      "loss": 0.6325,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019558374799549761,
+      "loss": 0.6899,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019557835799840953,
+      "loss": 0.9536,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955729647884667,
+      "loss": 0.8195,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955675683658504,
+      "loss": 0.7359,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019556216873074203,
+      "loss": 0.7953,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019555676588332315,
+      "loss": 0.7115,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955513598237753,
+      "loss": 0.7958,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019554595055228026,
+      "loss": 0.7337,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019554053806901983,
+      "loss": 0.5382,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000195535122374176,
+      "loss": 0.9897,
+      "step": 1646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019552970346793075,
+      "loss": 0.5713,
+      "step": 1647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019552428135046626,
+      "loss": 0.7817,
+      "step": 1648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955188560219648,
+      "loss": 0.544,
+      "step": 1649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019551342748260877,
+      "loss": 0.7191,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955079957325806,
+      "loss": 0.8655,
+      "step": 1651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001955025607720629,
+      "loss": 1.0011,
+      "step": 1652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019549712260123837,
+      "loss": 0.8422,
+      "step": 1653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019549168122028982,
+      "loss": 0.81,
+      "step": 1654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019548623662940013,
+      "loss": 0.7316,
+      "step": 1655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019548078882875236,
+      "loss": 0.7346,
+      "step": 1656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019547533781852963,
+      "loss": 0.5497,
+      "step": 1657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019546988359891513,
+      "loss": 0.9065,
+      "step": 1658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019546442617009223,
+      "loss": 0.5447,
+      "step": 1659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019545896553224438,
+      "loss": 1.1997,
+      "step": 1660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019545350168555516,
+      "loss": 0.7105,
+      "step": 1661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019544803463020824,
+      "loss": 0.8952,
+      "step": 1662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019544256436638735,
+      "loss": 0.7405,
+      "step": 1663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001954370908942764,
+      "loss": 0.3988,
+      "step": 1664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019543161421405937,
+      "loss": 0.8528,
+      "step": 1665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019542613432592038,
+      "loss": 1.0274,
+      "step": 1666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001954206512300436,
+      "loss": 0.6382,
+      "step": 1667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019541516492661338,
+      "loss": 0.6665,
+      "step": 1668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019540967541581414,
+      "loss": 0.6635,
+      "step": 1669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019540418269783038,
+      "loss": 0.7705,
+      "step": 1670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019539868677284674,
+      "loss": 0.6781,
+      "step": 1671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019539318764104802,
+      "loss": 0.7975,
+      "step": 1672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000195387685302619,
+      "loss": 0.7484,
+      "step": 1673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019538217975774465,
+      "loss": 0.5476,
+      "step": 1674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001953766710066101,
+      "loss": 0.8322,
+      "step": 1675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019537115904940045,
+      "loss": 0.3415,
+      "step": 1676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019536564388630102,
+      "loss": 0.5554,
+      "step": 1677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001953601255174972,
+      "loss": 0.5822,
+      "step": 1678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001953546039431745,
+      "loss": 0.5067,
+      "step": 1679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019534907916351853,
+      "loss": 0.7384,
+      "step": 1680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019534355117871495,
+      "loss": 0.7097,
+      "step": 1681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019533801998894968,
+      "loss": 0.7074,
+      "step": 1682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019533248559440853,
+      "loss": 0.8789,
+      "step": 1683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019532694799527764,
+      "loss": 0.7264,
+      "step": 1684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019532140719174311,
+      "loss": 0.9648,
+      "step": 1685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019531586318399119,
+      "loss": 0.6601,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019531031597220828,
+      "loss": 0.4265,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001953047655565808,
+      "loss": 0.4957,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019529921193729534,
+      "loss": 0.6088,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001952936551145386,
+      "loss": 0.6363,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019528809508849736,
+      "loss": 0.52,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019528253185935852,
+      "loss": 0.5551,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001952769654273091,
+      "loss": 0.6482,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001952713957925362,
+      "loss": 0.5867,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019526582295522707,
+      "loss": 0.851,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019526024691556896,
+      "loss": 0.7987,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019525466767374942,
+      "loss": 0.543,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019524908522995592,
+      "loss": 0.6134,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019524349958437615,
+      "loss": 0.6185,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019523791073719783,
+      "loss": 0.6195,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019523231868860889,
+      "loss": 0.5148,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019522672343879723,
+      "loss": 0.8587,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019522112498795099,
+      "loss": 0.8449,
+      "step": 1703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019521552333625836,
+      "loss": 0.8895,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019520991848390756,
+      "loss": 0.4958,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019520431043108712,
+      "loss": 0.582,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001951986991779855,
+      "loss": 0.8931,
+      "step": 1707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019519308472479128,
+      "loss": 0.7187,
+      "step": 1708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019518746707169324,
+      "loss": 0.7265,
+      "step": 1709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001951818462188802,
+      "loss": 0.775,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001951762221665411,
+      "loss": 0.629,
+      "step": 1711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000195170594914865,
+      "loss": 0.9414,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019516496446404108,
+      "loss": 1.0298,
+      "step": 1713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019515933081425856,
+      "loss": 0.664,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019515369396570683,
+      "loss": 0.7742,
+      "step": 1715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001951480539185754,
+      "loss": 0.7416,
+      "step": 1716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019514241067305384,
+      "loss": 0.6306,
+      "step": 1717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019513676422933186,
+      "loss": 0.7377,
+      "step": 1718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001951311145875992,
+      "loss": 0.762,
+      "step": 1719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019512546174804585,
+      "loss": 0.8241,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019511980571086183,
+      "loss": 0.7174,
+      "step": 1721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001951141464762372,
+      "loss": 0.7167,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019510848404436223,
+      "loss": 0.6945,
+      "step": 1723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019510281841542728,
+      "loss": 0.8973,
+      "step": 1724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950971495896228,
+      "loss": 0.7424,
+      "step": 1725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019509147756713934,
+      "loss": 0.5846,
+      "step": 1726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950858023481675,
+      "loss": 0.7811,
+      "step": 1727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950801239328982,
+      "loss": 0.878,
+      "step": 1728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019507444232152215,
+      "loss": 0.6186,
+      "step": 1729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019506875751423044,
+      "loss": 0.4407,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019506306951121413,
+      "loss": 0.8489,
+      "step": 1731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019505737831266445,
+      "loss": 0.7596,
+      "step": 1732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019505168391877268,
+      "loss": 0.642,
+      "step": 1733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950459863297302,
+      "loss": 0.5812,
+      "step": 1734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019504028554572864,
+      "loss": 0.4458,
+      "step": 1735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019503458156695956,
+      "loss": 0.5411,
+      "step": 1736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950288743936147,
+      "loss": 0.7037,
+      "step": 1737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950231640258859,
+      "loss": 0.6677,
+      "step": 1738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019501745046396517,
+      "loss": 0.9169,
+      "step": 1739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019501173370804447,
+      "loss": 0.8951,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019500601375831606,
+      "loss": 0.4503,
+      "step": 1741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001950002906149722,
+      "loss": 0.6895,
+      "step": 1742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019499456427820523,
+      "loss": 0.5467,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019498883474820767,
+      "loss": 0.5299,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019498310202517208,
+      "loss": 0.9,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019497736610929125,
+      "loss": 0.7107,
+      "step": 1746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001949716270007579,
+      "loss": 0.7205,
+      "step": 1747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000194965884699765,
+      "loss": 0.696,
+      "step": 1748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019496013920650558,
+      "loss": 0.6631,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019495439052117275,
+      "loss": 0.6448,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019494863864395976,
+      "loss": 0.7112,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019494288357505997,
+      "loss": 0.4895,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001949371253146668,
+      "loss": 0.5596,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019493136386297387,
+      "loss": 0.8337,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001949255992201748,
+      "loss": 0.5491,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019491983138646336,
+      "loss": 1.0699,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001949140603620335,
+      "loss": 0.9438,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019490828614707916,
+      "loss": 0.7331,
+      "step": 1758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019490250874179449,
+      "loss": 0.5584,
+      "step": 1759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001948967281463736,
+      "loss": 0.6628,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019489094436101089,
+      "loss": 0.7084,
+      "step": 1761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019488515738590078,
+      "loss": 0.7838,
+      "step": 1762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019487936722123773,
+      "loss": 0.6442,
+      "step": 1763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019487357386721644,
+      "loss": 0.7513,
+      "step": 1764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019486777732403166,
+      "loss": 0.8298,
+      "step": 1765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001948619775918782,
+      "loss": 0.6894,
+      "step": 1766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019485617467095102,
+      "loss": 0.5935,
+      "step": 1767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019485036856144516,
+      "loss": 0.6752,
+      "step": 1768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019484455926355588,
+      "loss": 0.7098,
+      "step": 1769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019483874677747836,
+      "loss": 0.5668,
+      "step": 1770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019483293110340806,
+      "loss": 0.6962,
+      "step": 1771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019482711224154045,
+      "loss": 0.8719,
+      "step": 1772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001948212901920711,
+      "loss": 0.7827,
+      "step": 1773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019481546495519577,
+      "loss": 0.6909,
+      "step": 1774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019480963653111023,
+      "loss": 0.4355,
+      "step": 1775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019480380492001043,
+      "loss": 0.8197,
+      "step": 1776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019479797012209238,
+      "loss": 0.6341,
+      "step": 1777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001947921321375522,
+      "loss": 0.6213,
+      "step": 1778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001947862909665862,
+      "loss": 0.6906,
+      "step": 1779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019478044660939067,
+      "loss": 0.7516,
+      "step": 1780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019477459906616206,
+      "loss": 0.6935,
+      "step": 1781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019476874833709696,
+      "loss": 0.73,
+      "step": 1782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001947628944223921,
+      "loss": 0.6928,
+      "step": 1783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019475703732224414,
+      "loss": 0.7208,
+      "step": 1784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019475117703685003,
+      "loss": 0.9563,
+      "step": 1785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019474531356640677,
+      "loss": 0.8433,
+      "step": 1786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019473944691111145,
+      "loss": 0.8037,
+      "step": 1787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019473357707116128,
+      "loss": 0.6411,
+      "step": 1788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019472770404675352,
+      "loss": 0.5364,
+      "step": 1789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001947218278380857,
+      "loss": 0.718,
+      "step": 1790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019471594844535525,
+      "loss": 0.7926,
+      "step": 1791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019471006586875987,
+      "loss": 0.6399,
+      "step": 1792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019470418010849723,
+      "loss": 0.8533,
+      "step": 1793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019469829116476524,
+      "loss": 0.8819,
+      "step": 1794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019469239903776187,
+      "loss": 0.7707,
+      "step": 1795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019468650372768512,
+      "loss": 0.8522,
+      "step": 1796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019468060523473318,
+      "loss": 0.5922,
+      "step": 1797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019467470355910438,
+      "loss": 1.0705,
+      "step": 1798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019466879870099705,
+      "loss": 0.7626,
+      "step": 1799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001946628906606097,
+      "loss": 0.4321,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001946569794381409,
+      "loss": 0.6649,
+      "step": 1801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001946510650337894,
+      "loss": 0.7113,
+      "step": 1802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019464514744775398,
+      "loss": 0.4957,
+      "step": 1803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001946392266802336,
+      "loss": 0.5293,
+      "step": 1804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019463330273142718,
+      "loss": 0.5468,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019462737560153398,
+      "loss": 0.9107,
+      "step": 1806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019462144529075318,
+      "loss": 0.7727,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019461551179928414,
+      "loss": 0.5737,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001946095751273263,
+      "loss": 0.6217,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001946036352750792,
+      "loss": 0.671,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001945976922427426,
+      "loss": 0.3624,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019459174603051615,
+      "loss": 0.5212,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001945857966385998,
+      "loss": 0.5466,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019457984406719356,
+      "loss": 0.6131,
+      "step": 1814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019457388831649748,
+      "loss": 0.5887,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019456792938671175,
+      "loss": 0.8818,
+      "step": 1816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019456196727803672,
+      "loss": 0.4944,
+      "step": 1817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019455600199067284,
+      "loss": 0.4947,
+      "step": 1818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019455003352482052,
+      "loss": 0.7363,
+      "step": 1819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019454406188068046,
+      "loss": 0.6157,
+      "step": 1820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001945380870584534,
+      "loss": 0.9085,
+      "step": 1821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019453210905834016,
+      "loss": 0.8505,
+      "step": 1822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001945261278805417,
+      "loss": 0.2908,
+      "step": 1823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001945201435252591,
+      "loss": 0.7441,
+      "step": 1824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019451415599269347,
+      "loss": 0.6181,
+      "step": 1825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019450816528304615,
+      "loss": 0.568,
+      "step": 1826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019450217139651844,
+      "loss": 1.011,
+      "step": 1827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019449617433331187,
+      "loss": 0.5914,
+      "step": 1828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000194490174093628,
+      "loss": 0.6229,
+      "step": 1829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944841706776686,
+      "loss": 0.7363,
+      "step": 1830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944781640856354,
+      "loss": 1.0399,
+      "step": 1831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944721543177303,
+      "loss": 0.7364,
+      "step": 1832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944661413741554,
+      "loss": 0.4435,
+      "step": 1833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019446012525511276,
+      "loss": 0.6564,
+      "step": 1834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019445410596080463,
+      "loss": 0.9276,
+      "step": 1835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019444808349143334,
+      "loss": 0.591,
+      "step": 1836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019444205784720133,
+      "loss": 0.8363,
+      "step": 1837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019443602902831115,
+      "loss": 0.5498,
+      "step": 1838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944299970349655,
+      "loss": 0.5998,
+      "step": 1839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944239618673671,
+      "loss": 0.7988,
+      "step": 1840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019441792352571885,
+      "loss": 0.7365,
+      "step": 1841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001944118820102237,
+      "loss": 0.5801,
+      "step": 1842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019440583732108473,
+      "loss": 0.5022,
+      "step": 1843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019439978945850515,
+      "loss": 0.5719,
+      "step": 1844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001943937384226883,
+      "loss": 0.8041,
+      "step": 1845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019438768421383752,
+      "loss": 0.5985,
+      "step": 1846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019438162683215634,
+      "loss": 0.7639,
+      "step": 1847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019437556627784837,
+      "loss": 0.6939,
+      "step": 1848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019436950255111735,
+      "loss": 0.8658,
+      "step": 1849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019436343565216711,
+      "loss": 0.8132,
+      "step": 1850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019435736558120158,
+      "loss": 0.851,
+      "step": 1851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019435129233842484,
+      "loss": 0.5876,
+      "step": 1852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019434521592404097,
+      "loss": 0.6112,
+      "step": 1853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001943391363382543,
+      "loss": 0.4416,
+      "step": 1854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019433305358126915,
+      "loss": 0.7211,
+      "step": 1855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019432696765329002,
+      "loss": 0.6426,
+      "step": 1856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019432087855452145,
+      "loss": 0.7133,
+      "step": 1857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001943147862851682,
+      "loss": 0.5957,
+      "step": 1858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019430869084543495,
+      "loss": 0.7136,
+      "step": 1859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001943025922355267,
+      "loss": 0.6861,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019429649045564837,
+      "loss": 0.6142,
+      "step": 1861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019429038550600511,
+      "loss": 0.5032,
+      "step": 1862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019428427738680217,
+      "loss": 0.9022,
+      "step": 1863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019427816609824482,
+      "loss": 0.8068,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019427205164053852,
+      "loss": 0.7138,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019426593401388883,
+      "loss": 0.576,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001942598132185013,
+      "loss": 0.9026,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001942536892545818,
+      "loss": 0.8298,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019424756212233608,
+      "loss": 0.6707,
+      "step": 1869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019424143182197016,
+      "loss": 0.7962,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001942352983536901,
+      "loss": 0.9118,
+      "step": 1871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019422916171770208,
+      "loss": 0.8316,
+      "step": 1872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001942230219142124,
+      "loss": 0.6856,
+      "step": 1873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001942168789434274,
+      "loss": 0.7022,
+      "step": 1874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019421073280555362,
+      "loss": 0.7033,
+      "step": 1875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019420458350079763,
+      "loss": 0.5045,
+      "step": 1876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019419843102936617,
+      "loss": 0.5125,
+      "step": 1877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000194192275391466,
+      "loss": 0.6325,
+      "step": 1878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001941861165873041,
+      "loss": 0.6809,
+      "step": 1879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019417995461708747,
+      "loss": 0.6031,
+      "step": 1880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019417378948102324,
+      "loss": 0.5223,
+      "step": 1881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019416762117931868,
+      "loss": 0.7016,
+      "step": 1882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001941614497121811,
+      "loss": 0.6324,
+      "step": 1883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000194155275079818,
+      "loss": 0.7281,
+      "step": 1884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019414909728243687,
+      "loss": 0.7522,
+      "step": 1885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019414291632024542,
+      "loss": 0.5916,
+      "step": 1886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019413673219345143,
+      "loss": 0.8402,
+      "step": 1887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019413054490226276,
+      "loss": 0.698,
+      "step": 1888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001941243544468874,
+      "loss": 0.6496,
+      "step": 1889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019411816082753343,
+      "loss": 0.6336,
+      "step": 1890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019411196404440907,
+      "loss": 0.726,
+      "step": 1891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001941057640977226,
+      "loss": 0.5397,
+      "step": 1892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019409956098768248,
+      "loss": 0.917,
+      "step": 1893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019409335471449714,
+      "loss": 0.792,
+      "step": 1894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019408714527837528,
+      "loss": 0.8016,
+      "step": 1895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001940809326795256,
+      "loss": 0.6631,
+      "step": 1896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019407471691815692,
+      "loss": 0.7642,
+      "step": 1897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019406849799447824,
+      "loss": 0.7656,
+      "step": 1898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019406227590869853,
+      "loss": 1.0302,
+      "step": 1899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019405605066102704,
+      "loss": 0.7065,
+      "step": 1900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019404982225167293,
+      "loss": 0.7552,
+      "step": 1901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001940435906808456,
+      "loss": 0.4441,
+      "step": 1902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019403735594875456,
+      "loss": 0.7264,
+      "step": 1903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019403111805560937,
+      "loss": 0.9226,
+      "step": 1904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001940248770016197,
+      "loss": 0.591,
+      "step": 1905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019401863278699534,
+      "loss": 0.6455,
+      "step": 1906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019401238541194623,
+      "loss": 0.8407,
+      "step": 1907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019400613487668234,
+      "loss": 0.8538,
+      "step": 1908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019399988118141376,
+      "loss": 0.505,
+      "step": 1909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019399362432635076,
+      "loss": 1.1786,
+      "step": 1910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019398736431170365,
+      "loss": 0.5494,
+      "step": 1911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001939811011376828,
+      "loss": 0.5981,
+      "step": 1912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019397483480449885,
+      "loss": 0.6258,
+      "step": 1913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019396856531236236,
+      "loss": 0.812,
+      "step": 1914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001939622926614841,
+      "loss": 0.5628,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019395601685207495,
+      "loss": 0.4042,
+      "step": 1916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019394973788434584,
+      "loss": 0.9777,
+      "step": 1917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001939434557585078,
+      "loss": 0.7962,
+      "step": 1918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001939371704747721,
+      "loss": 0.7051,
+      "step": 1919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019393088203334996,
+      "loss": 0.5944,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019392459043445276,
+      "loss": 0.722,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019391829567829203,
+      "loss": 0.6628,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001939119977650793,
+      "loss": 0.7108,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019390569669502634,
+      "loss": 0.6963,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019389939246834494,
+      "loss": 0.6384,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000193893085085247,
+      "loss": 1.0094,
+      "step": 1926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019388677454594455,
+      "loss": 1.0445,
+      "step": 1927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019388046085064972,
+      "loss": 0.916,
+      "step": 1928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019387414399957476,
+      "loss": 0.861,
+      "step": 1929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019386782399293198,
+      "loss": 0.9206,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019386150083093386,
+      "loss": 0.6764,
+      "step": 1931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019385517451379292,
+      "loss": 0.5935,
+      "step": 1932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019384884504172184,
+      "loss": 0.565,
+      "step": 1933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019384251241493335,
+      "loss": 0.6486,
+      "step": 1934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019383617663364038,
+      "loss": 0.8701,
+      "step": 1935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019382983769805585,
+      "loss": 0.6048,
+      "step": 1936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019382349560839287,
+      "loss": 0.7857,
+      "step": 1937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019381715036486462,
+      "loss": 0.8133,
+      "step": 1938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001938108019676844,
+      "loss": 0.7891,
+      "step": 1939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001938044504170656,
+      "loss": 0.3563,
+      "step": 1940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019379809571322176,
+      "loss": 0.669,
+      "step": 1941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019379173785636646,
+      "loss": 0.5231,
+      "step": 1942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019378537684671342,
+      "loss": 0.6204,
+      "step": 1943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019377901268447648,
+      "loss": 0.7116,
+      "step": 1944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019377264536986957,
+      "loss": 0.5799,
+      "step": 1945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001937662749031067,
+      "loss": 0.663,
+      "step": 1946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019375990128440204,
+      "loss": 0.8527,
+      "step": 1947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019375352451396984,
+      "loss": 0.7379,
+      "step": 1948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019374714459202444,
+      "loss": 0.8716,
+      "step": 1949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001937407615187803,
+      "loss": 0.6544,
+      "step": 1950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019373437529445194,
+      "loss": 0.7412,
+      "step": 1951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019372798591925415,
+      "loss": 0.4997,
+      "step": 1952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019372159339340163,
+      "loss": 0.6834,
+      "step": 1953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019371519771710927,
+      "loss": 0.5936,
+      "step": 1954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019370879889059206,
+      "loss": 0.7766,
+      "step": 1955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001937023969140651,
+      "loss": 0.6623,
+      "step": 1956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001936959917877436,
+      "loss": 0.76,
+      "step": 1957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019368958351184286,
+      "loss": 0.3779,
+      "step": 1958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019368317208657827,
+      "loss": 0.6373,
+      "step": 1959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019367675751216537,
+      "loss": 0.9424,
+      "step": 1960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001936703397888198,
+      "loss": 0.84,
+      "step": 1961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019366391891675728,
+      "loss": 0.7687,
+      "step": 1962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019365749489619365,
+      "loss": 0.6166,
+      "step": 1963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019365106772734487,
+      "loss": 0.5689,
+      "step": 1964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019364463741042694,
+      "loss": 0.8529,
+      "step": 1965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019363820394565604,
+      "loss": 0.875,
+      "step": 1966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019363176733324842,
+      "loss": 0.8309,
+      "step": 1967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019362532757342046,
+      "loss": 0.785,
+      "step": 1968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019361888466638865,
+      "loss": 0.7252,
+      "step": 1969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019361243861236953,
+      "loss": 0.5885,
+      "step": 1970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001936059894115798,
+      "loss": 0.7981,
+      "step": 1971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019359953706423624,
+      "loss": 0.8347,
+      "step": 1972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019359308157055577,
+      "loss": 0.9631,
+      "step": 1973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019358662293075539,
+      "loss": 0.7773,
+      "step": 1974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019358016114505216,
+      "loss": 0.7323,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001935736962136633,
+      "loss": 0.6153,
+      "step": 1976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019356722813680622,
+      "loss": 0.6007,
+      "step": 1977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019356075691469822,
+      "loss": 1.095,
+      "step": 1978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019355428254755688,
+      "loss": 0.9057,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019354780503559986,
+      "loss": 0.6923,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019354132437904487,
+      "loss": 0.6396,
+      "step": 1981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019353484057810976,
+      "loss": 0.4571,
+      "step": 1982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001935283536330125,
+      "loss": 0.7131,
+      "step": 1983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019352186354397116,
+      "loss": 0.5165,
+      "step": 1984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019351537031120383,
+      "loss": 0.7291,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019350887393492885,
+      "loss": 0.7985,
+      "step": 1986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019350237441536456,
+      "loss": 0.7222,
+      "step": 1987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019349587175272948,
+      "loss": 0.8899,
+      "step": 1988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019348936594724213,
+      "loss": 0.8334,
+      "step": 1989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001934828569991213,
+      "loss": 0.808,
+      "step": 1990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019347634490858568,
+      "loss": 0.8873,
+      "step": 1991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019346982967585424,
+      "loss": 0.8158,
+      "step": 1992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019346331130114597,
+      "loss": 1.046,
+      "step": 1993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019345678978467998,
+      "loss": 0.7508,
+      "step": 1994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001934502651266755,
+      "loss": 0.7037,
+      "step": 1995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019344373732735183,
+      "loss": 0.491,
+      "step": 1996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019343720638692846,
+      "loss": 0.5827,
+      "step": 1997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019343067230562485,
+      "loss": 0.6111,
+      "step": 1998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001934241350836607,
+      "loss": 0.7278,
+      "step": 1999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019341759472125576,
+      "loss": 0.6009,
+      "step": 2000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019341105121862984,
+      "loss": 0.7391,
+      "step": 2001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019340450457600294,
+      "loss": 0.7271,
+      "step": 2002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001933979547935951,
+      "loss": 0.7679,
+      "step": 2003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019339140187162648,
+      "loss": 0.7709,
+      "step": 2004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019338484581031738,
+      "loss": 0.6944,
+      "step": 2005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019337828660988818,
+      "loss": 0.7821,
+      "step": 2006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019337172427055934,
+      "loss": 0.4628,
+      "step": 2007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019336515879255148,
+      "loss": 0.7363,
+      "step": 2008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001933585901760853,
+      "loss": 0.6801,
+      "step": 2009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001933520184213816,
+      "loss": 0.8241,
+      "step": 2010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019334544352866127,
+      "loss": 0.6385,
+      "step": 2011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019333886549814533,
+      "loss": 0.8891,
+      "step": 2012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019333228433005491,
+      "loss": 0.6838,
+      "step": 2013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019332570002461124,
+      "loss": 0.7179,
+      "step": 2014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019331911258203564,
+      "loss": 0.7893,
+      "step": 2015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001933125220025495,
+      "loss": 0.7307,
+      "step": 2016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019330592828637448,
+      "loss": 0.5763,
+      "step": 2017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019329933143373212,
+      "loss": 0.7098,
+      "step": 2018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019329273144484422,
+      "loss": 0.5847,
+      "step": 2019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019328612831993262,
+      "loss": 0.4222,
+      "step": 2020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019327952205921932,
+      "loss": 0.8947,
+      "step": 2021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001932729126629263,
+      "loss": 0.7525,
+      "step": 2022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019326630013127583,
+      "loss": 0.7539,
+      "step": 2023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019325968446449015,
+      "loss": 0.5774,
+      "step": 2024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019325306566279164,
+      "loss": 0.6324,
+      "step": 2025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019324644372640276,
+      "loss": 0.932,
+      "step": 2026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019323981865554618,
+      "loss": 0.4624,
+      "step": 2027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001932331904504446,
+      "loss": 0.4145,
+      "step": 2028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019322655911132075,
+      "loss": 0.6737,
+      "step": 2029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019321992463839757,
+      "loss": 0.5147,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019321328703189808,
+      "loss": 0.5939,
+      "step": 2031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019320664629204543,
+      "loss": 0.5839,
+      "step": 2032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019320000241906284,
+      "loss": 0.5542,
+      "step": 2033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019319335541317361,
+      "loss": 0.5383,
+      "step": 2034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019318670527460123,
+      "loss": 0.8337,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001931800520035692,
+      "loss": 0.7797,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019317339560030118,
+      "loss": 0.6198,
+      "step": 2037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001931667360650209,
+      "loss": 0.604,
+      "step": 2038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019316007339795227,
+      "loss": 0.6106,
+      "step": 2039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019315340759931923,
+      "loss": 0.6418,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019314673866934582,
+      "loss": 0.903,
+      "step": 2041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001931400666082563,
+      "loss": 0.6948,
+      "step": 2042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019313339141627486,
+      "loss": 0.8368,
+      "step": 2043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019312671309362593,
+      "loss": 0.6573,
+      "step": 2044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000193120031640534,
+      "loss": 0.8219,
+      "step": 2045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019311334705722364,
+      "loss": 0.4578,
+      "step": 2046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001931066593439196,
+      "loss": 0.5661,
+      "step": 2047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019309996850084662,
+      "loss": 0.6908,
+      "step": 2048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019309327452822966,
+      "loss": 0.8534,
+      "step": 2049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001930865774262937,
+      "loss": 0.7787,
+      "step": 2050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019307987719526394,
+      "loss": 0.5371,
+      "step": 2051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019307317383536555,
+      "loss": 0.8698,
+      "step": 2052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019306646734682388,
+      "loss": 0.3825,
+      "step": 2053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019305975772986432,
+      "loss": 0.6263,
+      "step": 2054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019305304498471246,
+      "loss": 0.7365,
+      "step": 2055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019304632911159393,
+      "loss": 0.8551,
+      "step": 2056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019303961011073447,
+      "loss": 0.7889,
+      "step": 2057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019303288798236002,
+      "loss": 0.6137,
+      "step": 2058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019302616272669643,
+      "loss": 0.7544,
+      "step": 2059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019301943434396986,
+      "loss": 0.6625,
+      "step": 2060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019301270283440643,
+      "loss": 0.6983,
+      "step": 2061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001930059681982324,
+      "loss": 0.5198,
+      "step": 2062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019299923043567422,
+      "loss": 0.6916,
+      "step": 2063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019299248954695838,
+      "loss": 0.5086,
+      "step": 2064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019298574553231138,
+      "loss": 0.7569,
+      "step": 2065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019297899839196002,
+      "loss": 0.676,
+      "step": 2066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019297224812613105,
+      "loss": 0.8624,
+      "step": 2067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019296549473505137,
+      "loss": 0.6105,
+      "step": 2068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019295873821894804,
+      "loss": 0.581,
+      "step": 2069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019295197857804817,
+      "loss": 0.6689,
+      "step": 2070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019294521581257894,
+      "loss": 0.8772,
+      "step": 2071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019293844992276772,
+      "loss": 1.0178,
+      "step": 2072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019293168090884195,
+      "loss": 0.6233,
+      "step": 2073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019292490877102916,
+      "loss": 0.7313,
+      "step": 2074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000192918133509557,
+      "loss": 0.3446,
+      "step": 2075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019291135512465316,
+      "loss": 0.6365,
+      "step": 2076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019290457361654557,
+      "loss": 0.6492,
+      "step": 2077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019289778898546218,
+      "loss": 0.6297,
+      "step": 2078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019289100123163103,
+      "loss": 0.9508,
+      "step": 2079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019288421035528028,
+      "loss": 0.6794,
+      "step": 2080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019287741635663826,
+      "loss": 0.3702,
+      "step": 2081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019287061923593327,
+      "loss": 0.7584,
+      "step": 2082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019286381899339387,
+      "loss": 0.8672,
+      "step": 2083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928570156292486,
+      "loss": 0.2438,
+      "step": 2084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019285020914372618,
+      "loss": 0.6396,
+      "step": 2085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019284339953705537,
+      "loss": 0.6606,
+      "step": 2086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019283658680946514,
+      "loss": 0.7182,
+      "step": 2087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019282977096118445,
+      "loss": 0.7318,
+      "step": 2088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019282295199244242,
+      "loss": 1.0125,
+      "step": 2089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019281612990346828,
+      "loss": 0.7164,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019280930469449135,
+      "loss": 0.6835,
+      "step": 2091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019280247636574104,
+      "loss": 0.7229,
+      "step": 2092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001927956449174469,
+      "loss": 0.7492,
+      "step": 2093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019278881034983858,
+      "loss": 0.6102,
+      "step": 2094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019278197266314582,
+      "loss": 0.6195,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019277513185759844,
+      "loss": 0.7522,
+      "step": 2096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019276828793342646,
+      "loss": 0.7173,
+      "step": 2097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019276144089085982,
+      "loss": 0.7655,
+      "step": 2098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001927545907301288,
+      "loss": 0.35,
+      "step": 2099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001927477374514636,
+      "loss": 0.731,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019274088105509463,
+      "loss": 0.6631,
+      "step": 2101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019273402154125236,
+      "loss": 0.7678,
+      "step": 2102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019272715891016735,
+      "loss": 0.7476,
+      "step": 2103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001927202931620703,
+      "loss": 1.0483,
+      "step": 2104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019271342429719198,
+      "loss": 0.5534,
+      "step": 2105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019270655231576328,
+      "loss": 0.6656,
+      "step": 2106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019269967721801526,
+      "loss": 0.5603,
+      "step": 2107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000192692799004179,
+      "loss": 0.6863,
+      "step": 2108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926859176744857,
+      "loss": 0.7493,
+      "step": 2109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019267903322916667,
+      "loss": 0.6722,
+      "step": 2110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926721456684533,
+      "loss": 0.6723,
+      "step": 2111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926652549925772,
+      "loss": 1.0615,
+      "step": 2112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019265836120176995,
+      "loss": 0.4272,
+      "step": 2113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019265146429626326,
+      "loss": 0.731,
+      "step": 2114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019264456427628898,
+      "loss": 0.817,
+      "step": 2115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926376611420791,
+      "loss": 0.6701,
+      "step": 2116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019263075489386563,
+      "loss": 0.8536,
+      "step": 2117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019262384553188073,
+      "loss": 0.7694,
+      "step": 2118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019261693305635662,
+      "loss": 0.7036,
+      "step": 2119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019261001746752573,
+      "loss": 0.6916,
+      "step": 2120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019260309876562048,
+      "loss": 0.7413,
+      "step": 2121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019259617695087345,
+      "loss": 1.0175,
+      "step": 2122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019258925202351733,
+      "loss": 0.6645,
+      "step": 2123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019258232398378488,
+      "loss": 0.7965,
+      "step": 2124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000192575392831909,
+      "loss": 0.8657,
+      "step": 2125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019256845856812266,
+      "loss": 0.4802,
+      "step": 2126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000192561521192659,
+      "loss": 0.6583,
+      "step": 2127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019255458070575117,
+      "loss": 0.5923,
+      "step": 2128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019254763710763252,
+      "loss": 0.5724,
+      "step": 2129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019254069039853638,
+      "loss": 0.7602,
+      "step": 2130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019253374057869634,
+      "loss": 0.6101,
+      "step": 2131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000192526787648346,
+      "loss": 0.6221,
+      "step": 2132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019251983160771907,
+      "loss": 0.8508,
+      "step": 2133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019251287245704938,
+      "loss": 0.656,
+      "step": 2134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019250591019657085,
+      "loss": 0.571,
+      "step": 2135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019249894482651755,
+      "loss": 0.818,
+      "step": 2136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019249197634712355,
+      "loss": 0.6841,
+      "step": 2137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019248500475862315,
+      "loss": 0.6571,
+      "step": 2138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001924780300612507,
+      "loss": 0.6644,
+      "step": 2139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019247105225524066,
+      "loss": 0.5602,
+      "step": 2140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019246407134082758,
+      "loss": 0.4109,
+      "step": 2141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019245708731824608,
+      "loss": 0.4533,
+      "step": 2142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019245010018773098,
+      "loss": 0.6847,
+      "step": 2143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019244310994951714,
+      "loss": 0.8933,
+      "step": 2144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924361166038395,
+      "loss": 0.714,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019242912015093319,
+      "loss": 0.867,
+      "step": 2146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019242212059103334,
+      "loss": 0.8265,
+      "step": 2147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019241511792437532,
+      "loss": 0.5971,
+      "step": 2148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019240811215119448,
+      "loss": 0.5568,
+      "step": 2149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924011032717263,
+      "loss": 0.6241,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019239409128620638,
+      "loss": 0.5573,
+      "step": 2151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019238707619487045,
+      "loss": 0.5976,
+      "step": 2152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923800579979543,
+      "loss": 0.5651,
+      "step": 2153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923730366956939,
+      "loss": 0.4873,
+      "step": 2154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923660122883252,
+      "loss": 0.8219,
+      "step": 2155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019235898477608439,
+      "loss": 0.8224,
+      "step": 2156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019235195415920765,
+      "loss": 0.6628,
+      "step": 2157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019234492043793133,
+      "loss": 0.7962,
+      "step": 2158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019233788361249182,
+      "loss": 0.6503,
+      "step": 2159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923308436831258,
+      "loss": 0.67,
+      "step": 2160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019232380065006973,
+      "loss": 0.4339,
+      "step": 2161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019231675451356054,
+      "loss": 0.3274,
+      "step": 2162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019230970527383496,
+      "loss": 0.5534,
+      "step": 2163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019230265293113,
+      "loss": 0.9065,
+      "step": 2164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001922955974856827,
+      "loss": 0.4157,
+      "step": 2165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019228853893773026,
+      "loss": 0.8987,
+      "step": 2166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001922814772875099,
+      "loss": 0.5037,
+      "step": 2167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019227441253525906,
+      "loss": 0.8284,
+      "step": 2168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001922673446812152,
+      "loss": 0.6805,
+      "step": 2169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019226027372561588,
+      "loss": 0.8482,
+      "step": 2170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001922531996686988,
+      "loss": 0.583,
+      "step": 2171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019224612251070175,
+      "loss": 0.8976,
+      "step": 2172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019223904225186265,
+      "loss": 0.7564,
+      "step": 2173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001922319588924195,
+      "loss": 0.6729,
+      "step": 2174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019222487243261038,
+      "loss": 0.6185,
+      "step": 2175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019221778287267354,
+      "loss": 0.4661,
+      "step": 2176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019221069021284722,
+      "loss": 0.311,
+      "step": 2177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019220359445336994,
+      "loss": 0.607,
+      "step": 2178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019219649559448017,
+      "loss": 0.6595,
+      "step": 2179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001921893936364165,
+      "loss": 0.8456,
+      "step": 2180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019218228857941774,
+      "loss": 0.8284,
+      "step": 2181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019217518042372267,
+      "loss": 0.4202,
+      "step": 2182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019216806916957024,
+      "loss": 0.8549,
+      "step": 2183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001921609548171995,
+      "loss": 0.5154,
+      "step": 2184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001921538373668496,
+      "loss": 0.7465,
+      "step": 2185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019214671681875979,
+      "loss": 0.6502,
+      "step": 2186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019213959317316946,
+      "loss": 0.554,
+      "step": 2187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000192132466430318,
+      "loss": 0.7512,
+      "step": 2188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019212533659044502,
+      "loss": 0.5658,
+      "step": 2189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001921182036537902,
+      "loss": 0.706,
+      "step": 2190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019211106762059328,
+      "loss": 0.6421,
+      "step": 2191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019210392849109412,
+      "loss": 0.4365,
+      "step": 2192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019209678626553277,
+      "loss": 0.6596,
+      "step": 2193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001920896409441493,
+      "loss": 0.7049,
+      "step": 2194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001920824925271838,
+      "loss": 0.8189,
+      "step": 2195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001920753410148767,
+      "loss": 0.5502,
+      "step": 2196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001920681864074683,
+      "loss": 0.5483,
+      "step": 2197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019206102870519915,
+      "loss": 0.7398,
+      "step": 2198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019205386790830984,
+      "loss": 0.5381,
+      "step": 2199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001920467040170411,
+      "loss": 0.478,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019203953703163368,
+      "loss": 0.8027,
+      "step": 2201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001920323669523286,
+      "loss": 0.768,
+      "step": 2202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001920251937793668,
+      "loss": 0.6782,
+      "step": 2203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019201801751298942,
+      "loss": 0.9104,
+      "step": 2204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001920108381534377,
+      "loss": 0.5319,
+      "step": 2205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019200365570095298,
+      "loss": 0.7457,
+      "step": 2206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001919964701557767,
+      "loss": 0.7626,
+      "step": 2207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019198928151815037,
+      "loss": 0.6432,
+      "step": 2208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019198208978831564,
+      "loss": 0.4365,
+      "step": 2209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001919748949665143,
+      "loss": 0.714,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001919676970529882,
+      "loss": 0.8569,
+      "step": 2211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019196049604797922,
+      "loss": 0.7615,
+      "step": 2212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001919532919517295,
+      "loss": 0.7798,
+      "step": 2213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019194608476448118,
+      "loss": 0.6473,
+      "step": 2214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019193887448647654,
+      "loss": 0.8737,
+      "step": 2215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019193166111795793,
+      "loss": 0.6992,
+      "step": 2216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019192444465916783,
+      "loss": 0.6098,
+      "step": 2217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019191722511034884,
+      "loss": 0.7653,
+      "step": 2218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019191000247174365,
+      "loss": 0.9852,
+      "step": 2219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000191902776743595,
+      "loss": 0.6254,
+      "step": 2220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001918955479261458,
+      "loss": 0.6497,
+      "step": 2221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001918883160196391,
+      "loss": 0.5423,
+      "step": 2222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019188108102431795,
+      "loss": 0.6439,
+      "step": 2223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019187384294042553,
+      "loss": 0.6304,
+      "step": 2224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001918666017682052,
+      "loss": 0.515,
+      "step": 2225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019185935750790035,
+      "loss": 0.4887,
+      "step": 2226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001918521101597545,
+      "loss": 0.3567,
+      "step": 2227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019184485972401126,
+      "loss": 0.8902,
+      "step": 2228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019183760620091435,
+      "loss": 0.6894,
+      "step": 2229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019183034959070763,
+      "loss": 0.4721,
+      "step": 2230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019182308989363497,
+      "loss": 0.6367,
+      "step": 2231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019181582710994043,
+      "loss": 0.5312,
+      "step": 2232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019180856123986816,
+      "loss": 0.5865,
+      "step": 2233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001918012922836624,
+      "loss": 0.5444,
+      "step": 2234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019179402024156752,
+      "loss": 0.8685,
+      "step": 2235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001917867451138279,
+      "loss": 0.7251,
+      "step": 2236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019177946690068816,
+      "loss": 0.7581,
+      "step": 2237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019177218560239293,
+      "loss": 0.7166,
+      "step": 2238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019176490121918693,
+      "loss": 0.4965,
+      "step": 2239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001917576137513151,
+      "loss": 0.715,
+      "step": 2240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019175032319902234,
+      "loss": 0.5933,
+      "step": 2241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019174302956255376,
+      "loss": 0.3772,
+      "step": 2242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001917357328421545,
+      "loss": 0.7305,
+      "step": 2243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019172843303806993,
+      "loss": 0.7104,
+      "step": 2244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019172113015054532,
+      "loss": 0.6177,
+      "step": 2245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001917138241798262,
+      "loss": 0.6817,
+      "step": 2246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019170651512615816,
+      "loss": 0.6906,
+      "step": 2247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001916992029897869,
+      "loss": 0.7652,
+      "step": 2248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001916918877709582,
+      "loss": 0.5904,
+      "step": 2249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019168456946991796,
+      "loss": 0.5297,
+      "step": 2250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019167724808691222,
+      "loss": 0.7152,
+      "step": 2251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019166992362218706,
+      "loss": 0.7471,
+      "step": 2252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019166259607598867,
+      "loss": 0.821,
+      "step": 2253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001916552654485634,
+      "loss": 0.8044,
+      "step": 2254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019164793174015766,
+      "loss": 0.5373,
+      "step": 2255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019164059495101795,
+      "loss": 0.7943,
+      "step": 2256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001916332550813909,
+      "loss": 0.7718,
+      "step": 2257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001916259121315233,
+      "loss": 0.6183,
+      "step": 2258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001916185661016619,
+      "loss": 0.7574,
+      "step": 2259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019161121699205364,
+      "loss": 0.6549,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019160386480294563,
+      "loss": 0.4432,
+      "step": 2261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019159650953458495,
+      "loss": 0.6178,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001915891511872189,
+      "loss": 0.5917,
+      "step": 2263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019158178976109476,
+      "loss": 0.7151,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019157442525646003,
+      "loss": 0.6061,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019156705767356223,
+      "loss": 0.5856,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001915596870126491,
+      "loss": 0.6631,
+      "step": 2267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019155231327396832,
+      "loss": 0.75,
+      "step": 2268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001915449364577678,
+      "loss": 0.5846,
+      "step": 2269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019153755656429552,
+      "loss": 0.5308,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001915301735937995,
+      "loss": 0.685,
+      "step": 2271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019152278754652798,
+      "loss": 0.5495,
+      "step": 2272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019151539842272917,
+      "loss": 0.8509,
+      "step": 2273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019150800622265153,
+      "loss": 0.7653,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001915006109465435,
+      "loss": 0.6509,
+      "step": 2275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001914932125946537,
+      "loss": 0.8043,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001914858111672308,
+      "loss": 0.7904,
+      "step": 2277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019147840666452364,
+      "loss": 0.7709,
+      "step": 2278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001914709990867811,
+      "loss": 0.6044,
+      "step": 2279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019146358843425212,
+      "loss": 0.498,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019145617470718587,
+      "loss": 0.6109,
+      "step": 2281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019144875790583156,
+      "loss": 0.5982,
+      "step": 2282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001914413380304385,
+      "loss": 0.6308,
+      "step": 2283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019143391508125615,
+      "loss": 0.7315,
+      "step": 2284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019142648905853395,
+      "loss": 0.4058,
+      "step": 2285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001914190599625216,
+      "loss": 0.4576,
+      "step": 2286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019141162779346874,
+      "loss": 0.347,
+      "step": 2287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019140419255162528,
+      "loss": 0.715,
+      "step": 2288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019139675423724114,
+      "loss": 0.542,
+      "step": 2289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019138931285056631,
+      "loss": 0.8393,
+      "step": 2290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019138186839185098,
+      "loss": 0.5917,
+      "step": 2291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001913744208613454,
+      "loss": 0.764,
+      "step": 2292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001913669702592999,
+      "loss": 0.8135,
+      "step": 2293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019135951658596497,
+      "loss": 0.5553,
+      "step": 2294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019135205984159107,
+      "loss": 1.0145,
+      "step": 2295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001913446000264289,
+      "loss": 0.5936,
+      "step": 2296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019133713714072928,
+      "loss": 0.8031,
+      "step": 2297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019132967118474302,
+      "loss": 0.6925,
+      "step": 2298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001913222021587211,
+      "loss": 0.7123,
+      "step": 2299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019131473006291457,
+      "loss": 0.5895,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001913072548975746,
+      "loss": 0.4774,
+      "step": 2301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001912997766629525,
+      "loss": 0.6239,
+      "step": 2302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019129229535929968,
+      "loss": 0.7638,
+      "step": 2303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001912848109868675,
+      "loss": 0.8144,
+      "step": 2304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019127732354590768,
+      "loss": 0.5202,
+      "step": 2305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019126983303667186,
+      "loss": 0.3268,
+      "step": 2306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001912623394594118,
+      "loss": 0.6969,
+      "step": 2307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019125484281437945,
+      "loss": 0.7278,
+      "step": 2308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019124734310182678,
+      "loss": 0.8234,
+      "step": 2309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019123984032200586,
+      "loss": 0.88,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019123233447516892,
+      "loss": 0.7717,
+      "step": 2311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019122482556156834,
+      "loss": 0.7942,
+      "step": 2312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019121731358145644,
+      "loss": 0.6343,
+      "step": 2313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019120979853508573,
+      "loss": 0.6891,
+      "step": 2314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001912022804227089,
+      "loss": 0.3771,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001911947592445786,
+      "loss": 0.6301,
+      "step": 2316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001911872350009477,
+      "loss": 0.8297,
+      "step": 2317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019117970769206912,
+      "loss": 0.4532,
+      "step": 2318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019117217731819588,
+      "loss": 0.6368,
+      "step": 2319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001911646438795811,
+      "loss": 0.844,
+      "step": 2320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019115710737647803,
+      "loss": 0.5372,
+      "step": 2321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019114956780914,
+      "loss": 0.6723,
+      "step": 2322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019114202517782046,
+      "loss": 0.6905,
+      "step": 2323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019113447948277296,
+      "loss": 0.6098,
+      "step": 2324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019112693072425115,
+      "loss": 0.7055,
+      "step": 2325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019111937890250874,
+      "loss": 0.884,
+      "step": 2326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019111182401779963,
+      "loss": 0.7674,
+      "step": 2327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001911042660703778,
+      "loss": 0.5684,
+      "step": 2328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001910967050604972,
+      "loss": 0.501,
+      "step": 2329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019108914098841213,
+      "loss": 0.7619,
+      "step": 2330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019108157385437676,
+      "loss": 0.7486,
+      "step": 2331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001910740036586455,
+      "loss": 0.731,
+      "step": 2332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019106643040147278,
+      "loss": 0.824,
+      "step": 2333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001910588540831132,
+      "loss": 0.5241,
+      "step": 2334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019105127470382148,
+      "loss": 0.9085,
+      "step": 2335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019104369226385231,
+      "loss": 0.7107,
+      "step": 2336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019103610676346065,
+      "loss": 0.4442,
+      "step": 2337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019102851820290148,
+      "loss": 0.4021,
+      "step": 2338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019102092658242984,
+      "loss": 0.5099,
+      "step": 2339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001910133319023009,
+      "loss": 0.5181,
+      "step": 2340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019100573416277005,
+      "loss": 0.8171,
+      "step": 2341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019099813336409264,
+      "loss": 0.4872,
+      "step": 2342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001909905295065242,
+      "loss": 0.523,
+      "step": 2343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019098292259032022,
+      "loss": 0.6556,
+      "step": 2344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019097531261573654,
+      "loss": 0.6529,
+      "step": 2345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001909676995830289,
+      "loss": 0.3749,
+      "step": 2346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019096008349245324,
+      "loss": 0.3849,
+      "step": 2347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019095246434426558,
+      "loss": 0.6825,
+      "step": 2348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019094484213872198,
+      "loss": 0.6547,
+      "step": 2349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019093721687607872,
+      "loss": 0.7432,
+      "step": 2350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019092958855659205,
+      "loss": 0.5481,
+      "step": 2351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019092195718051848,
+      "loss": 0.5065,
+      "step": 2352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001909143227481145,
+      "loss": 0.7698,
+      "step": 2353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019090668525963676,
+      "loss": 0.4668,
+      "step": 2354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019089904471534196,
+      "loss": 1.0024,
+      "step": 2355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019089140111548696,
+      "loss": 0.914,
+      "step": 2356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019088375446032867,
+      "loss": 0.7318,
+      "step": 2357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019087610475012416,
+      "loss": 0.6364,
+      "step": 2358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019086845198513057,
+      "loss": 0.8587,
+      "step": 2359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019086079616560515,
+      "loss": 0.8749,
+      "step": 2360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019085313729180523,
+      "loss": 0.3928,
+      "step": 2361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019084547536398828,
+      "loss": 0.6158,
+      "step": 2362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019083781038241183,
+      "loss": 0.6651,
+      "step": 2363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001908301423473336,
+      "loss": 0.5262,
+      "step": 2364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019082247125901125,
+      "loss": 1.1359,
+      "step": 2365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019081479711770272,
+      "loss": 0.4686,
+      "step": 2366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019080711992366595,
+      "loss": 0.5901,
+      "step": 2367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019079943967715903,
+      "loss": 0.6208,
+      "step": 2368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001907917563784401,
+      "loss": 0.9814,
+      "step": 2369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019078407002776743,
+      "loss": 0.7456,
+      "step": 2370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019077638062539945,
+      "loss": 0.5499,
+      "step": 2371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019076868817159457,
+      "loss": 0.5446,
+      "step": 2372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001907609926666114,
+      "loss": 0.5741,
+      "step": 2373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019075329411070863,
+      "loss": 0.3994,
+      "step": 2374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019074559250414506,
+      "loss": 0.8005,
+      "step": 2375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019073788784717955,
+      "loss": 0.4406,
+      "step": 2376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019073018014007106,
+      "loss": 0.5729,
+      "step": 2377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019072246938307876,
+      "loss": 0.3951,
+      "step": 2378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001907147555764618,
+      "loss": 0.6911,
+      "step": 2379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019070703872047952,
+      "loss": 0.4193,
+      "step": 2380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019069931881539125,
+      "loss": 0.6352,
+      "step": 2381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019069159586145652,
+      "loss": 0.7033,
+      "step": 2382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000190683869858935,
+      "loss": 0.4819,
+      "step": 2383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001906761408080863,
+      "loss": 0.8607,
+      "step": 2384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001906684087091703,
+      "loss": 0.8603,
+      "step": 2385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019066067356244691,
+      "loss": 0.762,
+      "step": 2386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019065293536817607,
+      "loss": 0.8319,
+      "step": 2387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019064519412661802,
+      "loss": 0.8874,
+      "step": 2388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019063744983803288,
+      "loss": 0.4494,
+      "step": 2389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019062970250268105,
+      "loss": 0.6903,
+      "step": 2390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001906219521208229,
+      "loss": 0.7018,
+      "step": 2391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019061419869271894,
+      "loss": 0.6152,
+      "step": 2392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019060644221862986,
+      "loss": 0.6605,
+      "step": 2393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001905986826988164,
+      "loss": 0.6501,
+      "step": 2394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001905909201335393,
+      "loss": 0.7657,
+      "step": 2395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001905831545230596,
+      "loss": 0.6322,
+      "step": 2396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019057538586763828,
+      "loss": 0.7813,
+      "step": 2397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019056761416753654,
+      "loss": 0.6136,
+      "step": 2398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019055983942301556,
+      "loss": 0.3766,
+      "step": 2399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019055206163433671,
+      "loss": 0.869,
+      "step": 2400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019054428080176147,
+      "loss": 0.7514,
+      "step": 2401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019053649692555135,
+      "loss": 0.571,
+      "step": 2402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000190528710005968,
+      "loss": 0.4947,
+      "step": 2403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019052092004327324,
+      "loss": 0.641,
+      "step": 2404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019051312703772884,
+      "loss": 0.4425,
+      "step": 2405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019050533098959686,
+      "loss": 0.6432,
+      "step": 2406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001904975318991393,
+      "loss": 0.7477,
+      "step": 2407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001904897297666183,
+      "loss": 0.7926,
+      "step": 2408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001904819245922962,
+      "loss": 0.7437,
+      "step": 2409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019047411637643528,
+      "loss": 0.5447,
+      "step": 2410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001904663051192981,
+      "loss": 0.4789,
+      "step": 2411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019045849082114722,
+      "loss": 0.5038,
+      "step": 2412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019045067348224528,
+      "loss": 0.6178,
+      "step": 2413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019044285310285505,
+      "loss": 0.5041,
+      "step": 2414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019043502968323948,
+      "loss": 0.4404,
+      "step": 2415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019042720322366146,
+      "loss": 0.6951,
+      "step": 2416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019041937372438414,
+      "loss": 0.7574,
+      "step": 2417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019041154118567072,
+      "loss": 0.69,
+      "step": 2418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019040370560778445,
+      "loss": 0.8615,
+      "step": 2419
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019039586699098874,
+      "loss": 0.5888,
+      "step": 2420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019038802533554705,
+      "loss": 0.5286,
+      "step": 2421
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019038018064172302,
+      "loss": 0.6408,
+      "step": 2422
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019037233290978035,
+      "loss": 0.6633,
+      "step": 2423
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019036448213998283,
+      "loss": 0.6944,
+      "step": 2424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019035662833259432,
+      "loss": 0.7388,
+      "step": 2425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019034877148787889,
+      "loss": 0.4617,
+      "step": 2426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019034091160610058,
+      "loss": 0.7327,
+      "step": 2427
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019033304868752365,
+      "loss": 0.8833,
+      "step": 2428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019032518273241243,
+      "loss": 0.3489,
+      "step": 2429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019031731374103127,
+      "loss": 0.8501,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019030944171364474,
+      "loss": 0.6249,
+      "step": 2431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019030156665051745,
+      "loss": 0.4784,
+      "step": 2432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019029368855191407,
+      "loss": 0.7694,
+      "step": 2433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019028580741809945,
+      "loss": 0.7081,
+      "step": 2434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019027792324933853,
+      "loss": 0.5224,
+      "step": 2435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001902700360458963,
+      "loss": 0.6517,
+      "step": 2436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019026214580803794,
+      "loss": 0.7655,
+      "step": 2437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019025425253602865,
+      "loss": 0.4906,
+      "step": 2438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019024635623013375,
+      "loss": 0.764,
+      "step": 2439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019023845689061867,
+      "loss": 0.6412,
+      "step": 2440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000190230554517749,
+      "loss": 0.7835,
+      "step": 2441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019022264911179032,
+      "loss": 0.8066,
+      "step": 2442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019021474067300838,
+      "loss": 0.5895,
+      "step": 2443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019020682920166903,
+      "loss": 0.5307,
+      "step": 2444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019019891469803822,
+      "loss": 0.7458,
+      "step": 2445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019019099716238196,
+      "loss": 0.5982,
+      "step": 2446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019018307659496642,
+      "loss": 0.662,
+      "step": 2447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019017515299605788,
+      "loss": 0.4095,
+      "step": 2448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019016722636592265,
+      "loss": 0.7261,
+      "step": 2449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001901592967048272,
+      "loss": 0.4554,
+      "step": 2450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019015136401303809,
+      "loss": 0.6008,
+      "step": 2451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019014342829082193,
+      "loss": 0.5136,
+      "step": 2452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019013548953844553,
+      "loss": 0.4409,
+      "step": 2453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019012754775617574,
+      "loss": 0.6997,
+      "step": 2454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019011960294427953,
+      "loss": 0.6257,
+      "step": 2455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019011165510302393,
+      "loss": 0.7894,
+      "step": 2456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019010370423267614,
+      "loss": 0.5806,
+      "step": 2457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001900957503335034,
+      "loss": 0.791,
+      "step": 2458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019008779340577313,
+      "loss": 0.829,
+      "step": 2459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019007983344975272,
+      "loss": 0.5613,
+      "step": 2460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001900718704657098,
+      "loss": 0.5659,
+      "step": 2461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019006390445391202,
+      "loss": 0.4052,
+      "step": 2462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001900559354146272,
+      "loss": 0.249,
+      "step": 2463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019004796334812316,
+      "loss": 1.0805,
+      "step": 2464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001900399882546679,
+      "loss": 0.608,
+      "step": 2465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019003201013452953,
+      "loss": 0.6772,
+      "step": 2466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019002402898797623,
+      "loss": 0.6263,
+      "step": 2467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001900160448152762,
+      "loss": 0.8404,
+      "step": 2468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019000805761669795,
+      "loss": 0.7299,
+      "step": 2469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019000006739250993,
+      "loss": 0.9125,
+      "step": 2470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018999207414298067,
+      "loss": 1.0042,
+      "step": 2471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018998407786837893,
+      "loss": 0.6935,
+      "step": 2472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018997607856897347,
+      "loss": 0.7224,
+      "step": 2473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018996807624503318,
+      "loss": 0.4205,
+      "step": 2474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018996007089682707,
+      "loss": 0.671,
+      "step": 2475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018995206252462427,
+      "loss": 0.5156,
+      "step": 2476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018994405112869397,
+      "loss": 0.7946,
+      "step": 2477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001899360367093054,
+      "loss": 0.7485,
+      "step": 2478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018992801926672804,
+      "loss": 0.6784,
+      "step": 2479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018991999880123137,
+      "loss": 0.7096,
+      "step": 2480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000189911975313085,
+      "loss": 0.6671,
+      "step": 2481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018990394880255863,
+      "loss": 0.5786,
+      "step": 2482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001898959192699221,
+      "loss": 0.4005,
+      "step": 2483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018988788671544528,
+      "loss": 0.6938,
+      "step": 2484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018987985113939822,
+      "loss": 0.7079,
+      "step": 2485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000189871812542051,
+      "loss": 0.9803,
+      "step": 2486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018986377092367388,
+      "loss": 0.47,
+      "step": 2487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018985572628453713,
+      "loss": 0.5537,
+      "step": 2488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018984767862491116,
+      "loss": 0.6807,
+      "step": 2489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018983962794506657,
+      "loss": 0.8672,
+      "step": 2490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018983157424527392,
+      "loss": 0.8064,
+      "step": 2491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018982351752580397,
+      "loss": 0.513,
+      "step": 2492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001898154577869275,
+      "loss": 0.8908,
+      "step": 2493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018980739502891546,
+      "loss": 0.6973,
+      "step": 2494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018979932925203893,
+      "loss": 0.8382,
+      "step": 2495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018979126045656894,
+      "loss": 0.612,
+      "step": 2496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018978318864277678,
+      "loss": 0.7141,
+      "step": 2497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018977511381093379,
+      "loss": 0.6614,
+      "step": 2498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018976703596131138,
+      "loss": 0.7341,
+      "step": 2499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001897589550941811,
+      "loss": 0.457,
+      "step": 2500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018975087120981458,
+      "loss": 0.506,
+      "step": 2501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018974278430848357,
+      "loss": 0.7868,
+      "step": 2502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001897346943904599,
+      "loss": 0.4576,
+      "step": 2503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001897266014560155,
+      "loss": 0.65,
+      "step": 2504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018971850550542245,
+      "loss": 0.8178,
+      "step": 2505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018971040653895286,
+      "loss": 0.3928,
+      "step": 2506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018970230455687902,
+      "loss": 0.6091,
+      "step": 2507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001896941995594732,
+      "loss": 0.6297,
+      "step": 2508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001896860915470079,
+      "loss": 0.536,
+      "step": 2509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018967798051975566,
+      "loss": 0.6738,
+      "step": 2510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018966986647798913,
+      "loss": 0.5547,
+      "step": 2511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018966174942198108,
+      "loss": 0.6218,
+      "step": 2512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018965362935200437,
+      "loss": 0.709,
+      "step": 2513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001896455062683319,
+      "loss": 0.7571,
+      "step": 2514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001896373801712368,
+      "loss": 0.6668,
+      "step": 2515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001896292510609922,
+      "loss": 0.604,
+      "step": 2516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018962111893787128,
+      "loss": 0.7201,
+      "step": 2517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018961298380214752,
+      "loss": 0.7945,
+      "step": 2518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001896048456540943,
+      "loss": 0.4643,
+      "step": 2519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018959670449398525,
+      "loss": 0.3462,
+      "step": 2520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018958856032209398,
+      "loss": 0.768,
+      "step": 2521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018958041313869425,
+      "loss": 0.5869,
+      "step": 2522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018957226294406,
+      "loss": 0.5671,
+      "step": 2523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001895641097384651,
+      "loss": 0.4756,
+      "step": 2524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018955595352218368,
+      "loss": 0.5746,
+      "step": 2525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001895477942954899,
+      "loss": 0.6287,
+      "step": 2526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018953963205865803,
+      "loss": 0.5018,
+      "step": 2527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018953146681196246,
+      "loss": 0.7379,
+      "step": 2528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001895232985556776,
+      "loss": 0.8254,
+      "step": 2529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001895151272900781,
+      "loss": 0.8172,
+      "step": 2530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001895069530154386,
+      "loss": 0.6847,
+      "step": 2531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001894987757320339,
+      "loss": 0.7262,
+      "step": 2532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018949059544013882,
+      "loss": 0.8338,
+      "step": 2533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001894824121400284,
+      "loss": 0.471,
+      "step": 2534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001894742258319777,
+      "loss": 0.6028,
+      "step": 2535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018946603651626192,
+      "loss": 0.7083,
+      "step": 2536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001894578441931563,
+      "loss": 0.699,
+      "step": 2537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018944964886293626,
+      "loss": 0.4903,
+      "step": 2538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018944145052587728,
+      "loss": 0.6451,
+      "step": 2539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018943324918225494,
+      "loss": 0.8626,
+      "step": 2540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001894250448323449,
+      "loss": 0.8465,
+      "step": 2541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018941683747642304,
+      "loss": 0.586,
+      "step": 2542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018940862711476513,
+      "loss": 0.3888,
+      "step": 2543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018940041374764722,
+      "loss": 0.4964,
+      "step": 2544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001893921973753454,
+      "loss": 0.6917,
+      "step": 2545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001893839779981359,
+      "loss": 0.7323,
+      "step": 2546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001893757556162949,
+      "loss": 0.5886,
+      "step": 2547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018936753023009891,
+      "loss": 0.7536,
+      "step": 2548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001893593018398244,
+      "loss": 0.7131,
+      "step": 2549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018935107044574793,
+      "loss": 1.1061,
+      "step": 2550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018934283604814622,
+      "loss": 0.7386,
+      "step": 2551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018933459864729606,
+      "loss": 0.5738,
+      "step": 2552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018932635824347432,
+      "loss": 0.4849,
+      "step": 2553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001893181148369581,
+      "loss": 0.655,
+      "step": 2554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001893098684280244,
+      "loss": 0.4981,
+      "step": 2555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018930161901695045,
+      "loss": 0.6052,
+      "step": 2556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018929336660401358,
+      "loss": 0.5083,
+      "step": 2557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018928511118949116,
+      "loss": 0.5355,
+      "step": 2558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018927685277366068,
+      "loss": 0.6587,
+      "step": 2559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018926859135679982,
+      "loss": 0.7711,
+      "step": 2560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018926032693918621,
+      "loss": 0.579,
+      "step": 2561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018925205952109774,
+      "loss": 0.8144,
+      "step": 2562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001892437891028122,
+      "loss": 0.723,
+      "step": 2563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001892355156846077,
+      "loss": 0.4172,
+      "step": 2564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001892272392667623,
+      "loss": 0.8512,
+      "step": 2565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018921895984955423,
+      "loss": 0.3875,
+      "step": 2566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018921067743326182,
+      "loss": 0.5848,
+      "step": 2567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018920239201816342,
+      "loss": 0.6031,
+      "step": 2568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001891941036045376,
+      "loss": 0.6602,
+      "step": 2569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018918581219266298,
+      "loss": 0.6541,
+      "step": 2570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001891775177828182,
+      "loss": 0.6054,
+      "step": 2571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018916922037528217,
+      "loss": 0.6348,
+      "step": 2572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018916091997033376,
+      "loss": 0.7349,
+      "step": 2573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000189152616568252,
+      "loss": 0.4628,
+      "step": 2574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018914431016931596,
+      "loss": 0.6489,
+      "step": 2575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018913600077380492,
+      "loss": 0.6441,
+      "step": 2576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018912768838199817,
+      "loss": 0.5726,
+      "step": 2577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018911937299417514,
+      "loss": 0.6275,
+      "step": 2578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018911105461061535,
+      "loss": 0.535,
+      "step": 2579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018910273323159845,
+      "loss": 0.5617,
+      "step": 2580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018909440885740407,
+      "loss": 0.3724,
+      "step": 2581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018908608148831212,
+      "loss": 0.8643,
+      "step": 2582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018907775112460252,
+      "loss": 0.6425,
+      "step": 2583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018906941776655525,
+      "loss": 0.5271,
+      "step": 2584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018906108141445046,
+      "loss": 0.533,
+      "step": 2585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018905274206856837,
+      "loss": 0.8986,
+      "step": 2586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001890443997291893,
+      "loss": 0.5708,
+      "step": 2587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018903605439659369,
+      "loss": 0.5553,
+      "step": 2588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018902770607106206,
+      "loss": 0.6075,
+      "step": 2589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018901935475287504,
+      "loss": 0.5913,
+      "step": 2590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018901100044231336,
+      "loss": 0.8667,
+      "step": 2591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018900264313965784,
+      "loss": 0.7552,
+      "step": 2592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018899428284518944,
+      "loss": 0.722,
+      "step": 2593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018898591955918915,
+      "loss": 0.5604,
+      "step": 2594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018897755328193814,
+      "loss": 0.5146,
+      "step": 2595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018896918401371758,
+      "loss": 0.538,
+      "step": 2596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018896081175480886,
+      "loss": 0.4455,
+      "step": 2597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018895243650549342,
+      "loss": 0.9768,
+      "step": 2598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018894405826605273,
+      "loss": 0.7535,
+      "step": 2599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018893567703676847,
+      "loss": 0.4691,
+      "step": 2600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018892729281792236,
+      "loss": 0.6216,
+      "step": 2601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018891890560979624,
+      "loss": 0.6663,
+      "step": 2602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018891051541267207,
+      "loss": 0.6174,
+      "step": 2603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001889021222268318,
+      "loss": 0.7843,
+      "step": 2604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001888937260525577,
+      "loss": 0.7269,
+      "step": 2605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018888532689013187,
+      "loss": 0.2889,
+      "step": 2606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018887692473983675,
+      "loss": 0.4875,
+      "step": 2607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018886851960195472,
+      "loss": 0.5396,
+      "step": 2608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018886011147676833,
+      "loss": 0.6522,
+      "step": 2609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018885170036456024,
+      "loss": 0.4801,
+      "step": 2610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018884328626561314,
+      "loss": 0.8112,
+      "step": 2611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001888348691802099,
+      "loss": 0.6433,
+      "step": 2612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018882644910863349,
+      "loss": 0.6572,
+      "step": 2613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001888180260511669,
+      "loss": 0.3503,
+      "step": 2614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018880960000809327,
+      "loss": 0.5344,
+      "step": 2615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018880117097969587,
+      "loss": 0.6858,
+      "step": 2616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018879273896625805,
+      "loss": 0.5646,
+      "step": 2617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001887843039680632,
+      "loss": 0.3731,
+      "step": 2618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001887758659853949,
+      "loss": 0.4026,
+      "step": 2619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001887674250185368,
+      "loss": 0.8209,
+      "step": 2620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018875898106777262,
+      "loss": 0.7285,
+      "step": 2621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018875053413338618,
+      "loss": 0.9846,
+      "step": 2622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018874208421566148,
+      "loss": 0.7708,
+      "step": 2623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018873363131488254,
+      "loss": 0.5741,
+      "step": 2624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018872517543133346,
+      "loss": 0.6058,
+      "step": 2625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018871671656529854,
+      "loss": 0.5254,
+      "step": 2626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018870825471706214,
+      "loss": 0.3993,
+      "step": 2627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018869978988690858,
+      "loss": 0.7526,
+      "step": 2628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018869132207512257,
+      "loss": 0.8859,
+      "step": 2629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018868285128198866,
+      "loss": 0.4549,
+      "step": 2630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018867437750779158,
+      "loss": 0.697,
+      "step": 2631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018866590075281624,
+      "loss": 0.7822,
+      "step": 2632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001886574210173475,
+      "loss": 0.3588,
+      "step": 2633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018864893830167052,
+      "loss": 0.5965,
+      "step": 2634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018864045260607036,
+      "loss": 0.9526,
+      "step": 2635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018863196393083227,
+      "loss": 0.6285,
+      "step": 2636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018862347227624162,
+      "loss": 0.7075,
+      "step": 2637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018861497764258387,
+      "loss": 0.7128,
+      "step": 2638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018860648003014453,
+      "loss": 0.6495,
+      "step": 2639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018859797943920924,
+      "loss": 0.6873,
+      "step": 2640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001885894758700638,
+      "loss": 0.8854,
+      "step": 2641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018858096932299404,
+      "loss": 0.6311,
+      "step": 2642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018857245979828584,
+      "loss": 0.7442,
+      "step": 2643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018856394729622534,
+      "loss": 0.6355,
+      "step": 2644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001885554318170986,
+      "loss": 0.4256,
+      "step": 2645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018854691336119196,
+      "loss": 0.7452,
+      "step": 2646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018853839192879167,
+      "loss": 0.5432,
+      "step": 2647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018852986752018424,
+      "loss": 0.8278,
+      "step": 2648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018852134013565622,
+      "loss": 0.8149,
+      "step": 2649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018851280977549422,
+      "loss": 1.1091,
+      "step": 2650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000188504276439985,
+      "loss": 0.537,
+      "step": 2651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018849574012941544,
+      "loss": 0.3427,
+      "step": 2652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018848720084407242,
+      "loss": 0.6945,
+      "step": 2653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018847865858424308,
+      "loss": 0.5969,
+      "step": 2654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018847011335021449,
+      "loss": 0.5892,
+      "step": 2655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001884615651422739,
+      "loss": 0.673,
+      "step": 2656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018845301396070867,
+      "loss": 0.6856,
+      "step": 2657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001884444598058063,
+      "loss": 0.623,
+      "step": 2658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018843590267785424,
+      "loss": 0.6547,
+      "step": 2659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018842734257714023,
+      "loss": 0.6754,
+      "step": 2660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018841877950395196,
+      "loss": 0.6602,
+      "step": 2661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018841021345857728,
+      "loss": 0.5488,
+      "step": 2662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018840164444130415,
+      "loss": 0.5262,
+      "step": 2663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018839307245242063,
+      "loss": 0.6931,
+      "step": 2664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018838449749221484,
+      "loss": 0.454,
+      "step": 2665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018837591956097504,
+      "loss": 0.5545,
+      "step": 2666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018836733865898957,
+      "loss": 0.6817,
+      "step": 2667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001883587547865469,
+      "loss": 0.5014,
+      "step": 2668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018835016794393553,
+      "loss": 0.5529,
+      "step": 2669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018834157813144413,
+      "loss": 0.7071,
+      "step": 2670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018833298534936144,
+      "loss": 0.6919,
+      "step": 2671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018832438959797632,
+      "loss": 0.648,
+      "step": 2672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018831579087757772,
+      "loss": 0.3869,
+      "step": 2673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018830718918845467,
+      "loss": 0.4261,
+      "step": 2674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001882985845308963,
+      "loss": 0.5841,
+      "step": 2675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018828997690519187,
+      "loss": 0.9207,
+      "step": 2676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018828136631163074,
+      "loss": 0.4453,
+      "step": 2677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018827275275050233,
+      "loss": 0.5062,
+      "step": 2678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001882641362220962,
+      "loss": 0.681,
+      "step": 2679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018825551672670197,
+      "loss": 0.6091,
+      "step": 2680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018824689426460943,
+      "loss": 0.5634,
+      "step": 2681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018823826883610835,
+      "loss": 0.6342,
+      "step": 2682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018822964044148876,
+      "loss": 0.5307,
+      "step": 2683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018822100908104063,
+      "loss": 0.5497,
+      "step": 2684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018821237475505413,
+      "loss": 0.6342,
+      "step": 2685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018820373746381952,
+      "loss": 0.7085,
+      "step": 2686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001881950972076271,
+      "loss": 0.6088,
+      "step": 2687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018818645398676736,
+      "loss": 0.5904,
+      "step": 2688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018817780780153078,
+      "loss": 0.9449,
+      "step": 2689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018816915865220806,
+      "loss": 0.6645,
+      "step": 2690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001881605065390899,
+      "loss": 0.7154,
+      "step": 2691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018815185146246716,
+      "loss": 0.6134,
+      "step": 2692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001881431934226308,
+      "loss": 0.7306,
+      "step": 2693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018813453241987182,
+      "loss": 0.3545,
+      "step": 2694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018812586845448135,
+      "loss": 0.5371,
+      "step": 2695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018811720152675068,
+      "loss": 0.7351,
+      "step": 2696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001881085316369711,
+      "loss": 0.7535,
+      "step": 2697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018809985878543404,
+      "loss": 0.6944,
+      "step": 2698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001880911829724311,
+      "loss": 0.3196,
+      "step": 2699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018808250419825387,
+      "loss": 0.5281,
+      "step": 2700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018807382246319412,
+      "loss": 0.4901,
+      "step": 2701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001880651377675436,
+      "loss": 0.4627,
+      "step": 2702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018805645011159435,
+      "loss": 0.5222,
+      "step": 2703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018804775949563836,
+      "loss": 0.6882,
+      "step": 2704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018803906591996773,
+      "loss": 0.8839,
+      "step": 2705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018803036938487474,
+      "loss": 0.4673,
+      "step": 2706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018802166989065176,
+      "loss": 0.5509,
+      "step": 2707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001880129674375911,
+      "loss": 0.4915,
+      "step": 2708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001880042620259854,
+      "loss": 0.629,
+      "step": 2709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018799555365612726,
+      "loss": 0.6936,
+      "step": 2710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018798684232830938,
+      "loss": 0.9691,
+      "step": 2711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018797812804282463,
+      "loss": 0.6075,
+      "step": 2712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018796941079996592,
+      "loss": 0.7332,
+      "step": 2713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001879606906000263,
+      "loss": 0.7556,
+      "step": 2714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018795196744329887,
+      "loss": 0.4937,
+      "step": 2715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001879432413300769,
+      "loss": 0.6537,
+      "step": 2716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018793451226065366,
+      "loss": 0.4813,
+      "step": 2717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001879257802353226,
+      "loss": 0.4222,
+      "step": 2718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018791704525437727,
+      "loss": 0.7171,
+      "step": 2719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018790830731811126,
+      "loss": 0.6251,
+      "step": 2720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018789956642681834,
+      "loss": 0.7057,
+      "step": 2721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018789082258079226,
+      "loss": 0.9251,
+      "step": 2722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018788207578032703,
+      "loss": 0.7456,
+      "step": 2723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018787332602571662,
+      "loss": 0.5692,
+      "step": 2724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018786457331725512,
+      "loss": 0.7251,
+      "step": 2725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018785581765523685,
+      "loss": 0.6285,
+      "step": 2726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018784705903995605,
+      "loss": 0.5941,
+      "step": 2727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001878382974717072,
+      "loss": 0.8032,
+      "step": 2728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018782953295078475,
+      "loss": 0.2871,
+      "step": 2729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018782076547748332,
+      "loss": 0.563,
+      "step": 2730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001878119950520977,
+      "loss": 0.5974,
+      "step": 2731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018780322167492267,
+      "loss": 0.9282,
+      "step": 2732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018779444534625315,
+      "loss": 0.6948,
+      "step": 2733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018778566606638411,
+      "loss": 0.7003,
+      "step": 2734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018777688383561075,
+      "loss": 0.6598,
+      "step": 2735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001877680986542282,
+      "loss": 0.6211,
+      "step": 2736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001877593105225318,
+      "loss": 0.6723,
+      "step": 2737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018775051944081698,
+      "loss": 0.6361,
+      "step": 2738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018774172540937927,
+      "loss": 0.7056,
+      "step": 2739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018773292842851421,
+      "loss": 0.7037,
+      "step": 2740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018772412849851758,
+      "loss": 0.6759,
+      "step": 2741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018771532561968513,
+      "loss": 0.6389,
+      "step": 2742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001877065197923128,
+      "loss": 0.496,
+      "step": 2743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018769771101669665,
+      "loss": 0.7473,
+      "step": 2744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018768889929313267,
+      "loss": 0.5881,
+      "step": 2745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018768008462191716,
+      "loss": 0.3731,
+      "step": 2746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018767126700334634,
+      "loss": 0.5531,
+      "step": 2747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018766244643771672,
+      "loss": 0.5376,
+      "step": 2748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018765362292532469,
+      "loss": 0.5745,
+      "step": 2749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018764479646646694,
+      "loss": 0.7849,
+      "step": 2750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018763596706144014,
+      "loss": 0.7403,
+      "step": 2751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001876271347105411,
+      "loss": 0.8986,
+      "step": 2752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018761829941406664,
+      "loss": 0.485,
+      "step": 2753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018760946117231386,
+      "loss": 0.7387,
+      "step": 2754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018760061998557984,
+      "loss": 0.739,
+      "step": 2755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001875917758541617,
+      "loss": 0.5627,
+      "step": 2756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018758292877835684,
+      "loss": 0.5933,
+      "step": 2757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001875740787584626,
+      "loss": 0.4961,
+      "step": 2758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018756522579477643,
+      "loss": 0.6681,
+      "step": 2759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000187556369887596,
+      "loss": 0.5742,
+      "step": 2760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018754751103721896,
+      "loss": 0.5347,
+      "step": 2761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018753864924394306,
+      "loss": 0.6316,
+      "step": 2762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018752978450806626,
+      "loss": 0.5094,
+      "step": 2763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018752091682988655,
+      "loss": 0.7573,
+      "step": 2764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018751204620970192,
+      "loss": 0.6411,
+      "step": 2765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001875031726478107,
+      "loss": 0.591,
+      "step": 2766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018749429614451103,
+      "loss": 0.3831,
+      "step": 2767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001874854167001014,
+      "loss": 0.798,
+      "step": 2768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018747653431488022,
+      "loss": 0.7388,
+      "step": 2769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001874676489891461,
+      "loss": 0.7092,
+      "step": 2770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018745876072319773,
+      "loss": 0.8448,
+      "step": 2771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018744986951733385,
+      "loss": 0.7788,
+      "step": 2772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018744097537185336,
+      "loss": 0.3664,
+      "step": 2773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018743207828705524,
+      "loss": 1.0742,
+      "step": 2774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001874231782632386,
+      "loss": 0.7057,
+      "step": 2775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018741427530070254,
+      "loss": 0.5023,
+      "step": 2776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018740536939974639,
+      "loss": 0.5887,
+      "step": 2777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018739646056066946,
+      "loss": 0.5226,
+      "step": 2778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001873875487837713,
+      "loss": 1.0226,
+      "step": 2779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001873786340693514,
+      "loss": 0.5355,
+      "step": 2780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018736971641770947,
+      "loss": 0.7657,
+      "step": 2781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018736079582914528,
+      "loss": 0.7129,
+      "step": 2782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018735187230395868,
+      "loss": 0.6515,
+      "step": 2783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018734294584244962,
+      "loss": 0.4568,
+      "step": 2784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001873340164449182,
+      "loss": 0.9165,
+      "step": 2785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018732508411166453,
+      "loss": 0.4672,
+      "step": 2786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001873161488429889,
+      "loss": 0.7509,
+      "step": 2787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018730721063919172,
+      "loss": 0.7268,
+      "step": 2788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018729826950057336,
+      "loss": 0.7537,
+      "step": 2789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001872893254274344,
+      "loss": 0.5167,
+      "step": 2790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018728037842007548,
+      "loss": 0.9387,
+      "step": 2791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001872714284787974,
+      "loss": 0.5102,
+      "step": 2792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018726247560390099,
+      "loss": 0.4914,
+      "step": 2793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018725351979568717,
+      "loss": 0.4793,
+      "step": 2794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018724456105445703,
+      "loss": 0.583,
+      "step": 2795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001872355993805117,
+      "loss": 0.8116,
+      "step": 2796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001872266347741524,
+      "loss": 0.7699,
+      "step": 2797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018721766723568055,
+      "loss": 0.5813,
+      "step": 2798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001872086967653975,
+      "loss": 0.5206,
+      "step": 2799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018719972336360486,
+      "loss": 0.76,
+      "step": 2800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018719074703060423,
+      "loss": 0.9043,
+      "step": 2801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018718176776669735,
+      "loss": 0.525,
+      "step": 2802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001871727855721861,
+      "loss": 0.6488,
+      "step": 2803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018716380044737235,
+      "loss": 0.5602,
+      "step": 2804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018715481239255818,
+      "loss": 0.8639,
+      "step": 2805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018714582140804572,
+      "loss": 0.6523,
+      "step": 2806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001871368274941372,
+      "loss": 0.6843,
+      "step": 2807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018712783065113494,
+      "loss": 0.9753,
+      "step": 2808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018711883087934133,
+      "loss": 0.4666,
+      "step": 2809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018710982817905894,
+      "loss": 0.503,
+      "step": 2810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018710082255059046,
+      "loss": 0.4861,
+      "step": 2811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018709181399423848,
+      "loss": 0.8685,
+      "step": 2812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018708280251030588,
+      "loss": 0.6017,
+      "step": 2813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001870737880990956,
+      "loss": 0.8084,
+      "step": 2814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018706477076091066,
+      "loss": 0.6381,
+      "step": 2815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018705575049605413,
+      "loss": 0.4953,
+      "step": 2816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018704672730482925,
+      "loss": 0.5839,
+      "step": 2817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018703770118753935,
+      "loss": 0.5643,
+      "step": 2818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018702867214448783,
+      "loss": 0.3615,
+      "step": 2819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001870196401759782,
+      "loss": 0.6465,
+      "step": 2820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018701060528231406,
+      "loss": 0.6948,
+      "step": 2821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018700156746379913,
+      "loss": 0.6218,
+      "step": 2822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001869925267207372,
+      "loss": 0.6175,
+      "step": 2823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001869834830534322,
+      "loss": 0.5105,
+      "step": 2824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001869744364621881,
+      "loss": 0.8315,
+      "step": 2825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018696538694730906,
+      "loss": 0.6496,
+      "step": 2826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018695633450909916,
+      "loss": 0.9544,
+      "step": 2827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018694727914786284,
+      "loss": 0.6098,
+      "step": 2828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001869382208639044,
+      "loss": 0.4053,
+      "step": 2829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018692915965752834,
+      "loss": 0.6772,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018692009552903927,
+      "loss": 0.7208,
+      "step": 2831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018691102847874188,
+      "loss": 0.7024,
+      "step": 2832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018690195850694098,
+      "loss": 0.4908,
+      "step": 2833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001868928856139414,
+      "loss": 0.7519,
+      "step": 2834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001868838098000482,
+      "loss": 0.5069,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001868747310655664,
+      "loss": 0.6597,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018686564941080116,
+      "loss": 0.5389,
+      "step": 2837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018685656483605784,
+      "loss": 0.5973,
+      "step": 2838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018684747734164177,
+      "loss": 0.271,
+      "step": 2839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018683838692785842,
+      "loss": 0.4143,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018682929359501338,
+      "loss": 0.8213,
+      "step": 2841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001868201973434123,
+      "loss": 0.5945,
+      "step": 2842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018681109817336099,
+      "loss": 0.4718,
+      "step": 2843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018680199608516528,
+      "loss": 0.5826,
+      "step": 2844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018679289107913113,
+      "loss": 0.6379,
+      "step": 2845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018678378315556467,
+      "loss": 0.7395,
+      "step": 2846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018677467231477198,
+      "loss": 0.3103,
+      "step": 2847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018676555855705936,
+      "loss": 0.8071,
+      "step": 2848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018675644188273315,
+      "loss": 0.4384,
+      "step": 2849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018674732229209982,
+      "loss": 0.2666,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001867381997854659,
+      "loss": 0.7452,
+      "step": 2851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018672907436313808,
+      "loss": 0.603,
+      "step": 2852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001867199460254231,
+      "loss": 0.7792,
+      "step": 2853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001867108147726278,
+      "loss": 0.6008,
+      "step": 2854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001867016806050591,
+      "loss": 0.6003,
+      "step": 2855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001866925435230241,
+      "loss": 0.6582,
+      "step": 2856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018668340352682989,
+      "loss": 0.7686,
+      "step": 2857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018667426061678375,
+      "loss": 0.7744,
+      "step": 2858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000186665114793193,
+      "loss": 0.6682,
+      "step": 2859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018665596605636507,
+      "loss": 0.4671,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018664681440660747,
+      "loss": 0.5,
+      "step": 2861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018663765984422786,
+      "loss": 0.4548,
+      "step": 2862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000186628502369534,
+      "loss": 0.5151,
+      "step": 2863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018661934198283372,
+      "loss": 0.925,
+      "step": 2864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018661017868443485,
+      "loss": 0.7668,
+      "step": 2865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001866010124746455,
+      "loss": 0.8913,
+      "step": 2866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018659184335377375,
+      "loss": 0.4635,
+      "step": 2867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018658267132212785,
+      "loss": 0.4511,
+      "step": 2868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001865734963800161,
+      "loss": 0.6696,
+      "step": 2869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018656431852774694,
+      "loss": 0.5898,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018655513776562883,
+      "loss": 0.8164,
+      "step": 2871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018654595409397042,
+      "loss": 0.2518,
+      "step": 2872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001865367675130804,
+      "loss": 0.2778,
+      "step": 2873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001865275780232676,
+      "loss": 0.6955,
+      "step": 2874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018651838562484086,
+      "loss": 0.611,
+      "step": 2875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018650919031810927,
+      "loss": 0.5411,
+      "step": 2876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018649999210338186,
+      "loss": 0.7184,
+      "step": 2877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018649079098096788,
+      "loss": 0.5068,
+      "step": 2878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018648158695117658,
+      "loss": 0.709,
+      "step": 2879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001864723800143174,
+      "loss": 0.5861,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001864631701706998,
+      "loss": 0.5259,
+      "step": 2881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018645395742063333,
+      "loss": 0.7487,
+      "step": 2882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018644474176442773,
+      "loss": 0.6948,
+      "step": 2883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018643552320239277,
+      "loss": 0.8571,
+      "step": 2884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018642630173483832,
+      "loss": 0.4749,
+      "step": 2885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018641707736207438,
+      "loss": 0.5531,
+      "step": 2886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018640785008441102,
+      "loss": 0.5278,
+      "step": 2887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001863986199021584,
+      "loss": 0.6209,
+      "step": 2888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018638938681562678,
+      "loss": 0.6784,
+      "step": 2889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018638015082512657,
+      "loss": 0.5864,
+      "step": 2890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018637091193096818,
+      "loss": 0.5171,
+      "step": 2891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018636167013346224,
+      "loss": 0.5238,
+      "step": 2892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018635242543291937,
+      "loss": 0.384,
+      "step": 2893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018634317782965037,
+      "loss": 0.6499,
+      "step": 2894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018633392732396602,
+      "loss": 0.4904,
+      "step": 2895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018632467391617734,
+      "loss": 0.4633,
+      "step": 2896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018631541760659535,
+      "loss": 0.4873,
+      "step": 2897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018630615839553122,
+      "loss": 0.4779,
+      "step": 2898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001862968962832962,
+      "loss": 0.5918,
+      "step": 2899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001862876312702016,
+      "loss": 0.6912,
+      "step": 2900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018627836335655893,
+      "loss": 0.7305,
+      "step": 2901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018626909254267963,
+      "loss": 0.4185,
+      "step": 2902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018625981882887545,
+      "loss": 0.951,
+      "step": 2903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018625054221545803,
+      "loss": 0.4309,
+      "step": 2904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018624126270273928,
+      "loss": 0.5657,
+      "step": 2905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018623198029103105,
+      "loss": 0.9362,
+      "step": 2906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018622269498064543,
+      "loss": 0.4389,
+      "step": 2907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018621340677189453,
+      "loss": 0.5474,
+      "step": 2908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018620411566509052,
+      "loss": 0.5802,
+      "step": 2909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001861948216605458,
+      "loss": 0.3693,
+      "step": 2910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018618552475857273,
+      "loss": 0.7428,
+      "step": 2911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018617622495948388,
+      "loss": 0.7742,
+      "step": 2912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018616692226359184,
+      "loss": 0.6298,
+      "step": 2913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018615761667120927,
+      "loss": 0.5683,
+      "step": 2914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000186148308182649,
+      "loss": 0.818,
+      "step": 2915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018613899679822396,
+      "loss": 0.528,
+      "step": 2916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018612968251824713,
+      "loss": 0.4641,
+      "step": 2917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018612036534303164,
+      "loss": 0.6683,
+      "step": 2918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018611104527289061,
+      "loss": 0.7283,
+      "step": 2919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018610172230813741,
+      "loss": 1.193,
+      "step": 2920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018609239644908543,
+      "loss": 0.8164,
+      "step": 2921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001860830676960481,
+      "loss": 0.6299,
+      "step": 2922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018607373604933905,
+      "loss": 0.4226,
+      "step": 2923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018606440150927196,
+      "loss": 0.8054,
+      "step": 2924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018605506407616058,
+      "loss": 0.6415,
+      "step": 2925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018604572375031882,
+      "loss": 0.6456,
+      "step": 2926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018603638053206063,
+      "loss": 0.5334,
+      "step": 2927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018602703442170007,
+      "loss": 0.6819,
+      "step": 2928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018601768541955135,
+      "loss": 0.5209,
+      "step": 2929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001860083335259287,
+      "loss": 0.4848,
+      "step": 2930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018599897874114652,
+      "loss": 0.6462,
+      "step": 2931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018598962106551926,
+      "loss": 0.5977,
+      "step": 2932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018598026049936143,
+      "loss": 0.5678,
+      "step": 2933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001859708970429877,
+      "loss": 0.5963,
+      "step": 2934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018596153069671284,
+      "loss": 0.955,
+      "step": 2935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018595216146085174,
+      "loss": 0.6277,
+      "step": 2936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001859427893357193,
+      "loss": 0.7662,
+      "step": 2937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018593341432163055,
+      "loss": 0.5355,
+      "step": 2938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018592403641890063,
+      "loss": 0.6485,
+      "step": 2939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018591465562784482,
+      "loss": 0.5772,
+      "step": 2940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018590527194877843,
+      "loss": 0.5456,
+      "step": 2941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018589588538201688,
+      "loss": 0.6313,
+      "step": 2942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018588649592787569,
+      "loss": 0.669,
+      "step": 2943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018587710358667053,
+      "loss": 0.6766,
+      "step": 2944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018586770835871708,
+      "loss": 0.6939,
+      "step": 2945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018585831024433116,
+      "loss": 0.5289,
+      "step": 2946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001858489092438287,
+      "loss": 0.5291,
+      "step": 2947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018583950535752574,
+      "loss": 0.4458,
+      "step": 2948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018583009858573837,
+      "loss": 0.5091,
+      "step": 2949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018582068892878276,
+      "loss": 0.703,
+      "step": 2950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018581127638697526,
+      "loss": 0.519,
+      "step": 2951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018580186096063225,
+      "loss": 0.5329,
+      "step": 2952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018579244265007024,
+      "loss": 0.8018,
+      "step": 2953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018578302145560584,
+      "loss": 0.7861,
+      "step": 2954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001857735973775557,
+      "loss": 0.311,
+      "step": 2955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018576417041623667,
+      "loss": 0.5052,
+      "step": 2956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018575474057196557,
+      "loss": 0.5953,
+      "step": 2957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001857453078450594,
+      "loss": 0.8426,
+      "step": 2958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018573587223583529,
+      "loss": 0.8016,
+      "step": 2959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018572643374461034,
+      "loss": 0.7193,
+      "step": 2960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018571699237170187,
+      "loss": 0.4205,
+      "step": 2961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018570754811742725,
+      "loss": 0.4982,
+      "step": 2962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018569810098210396,
+      "loss": 0.9572,
+      "step": 2963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018568865096604955,
+      "loss": 0.4563,
+      "step": 2964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018567919806958166,
+      "loss": 0.2352,
+      "step": 2965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018566974229301803,
+      "loss": 0.5123,
+      "step": 2966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018566028363667657,
+      "loss": 0.6414,
+      "step": 2967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018565082210087526,
+      "loss": 0.47,
+      "step": 2968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018564135768593204,
+      "loss": 0.8008,
+      "step": 2969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018563189039216512,
+      "loss": 0.4644,
+      "step": 2970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018562242021989274,
+      "loss": 0.7421,
+      "step": 2971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018561294716943328,
+      "loss": 0.7187,
+      "step": 2972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001856034712411051,
+      "loss": 0.8053,
+      "step": 2973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018559399243522676,
+      "loss": 0.534,
+      "step": 2974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001855845107521169,
+      "loss": 0.3996,
+      "step": 2975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001855750261920942,
+      "loss": 0.4562,
+      "step": 2976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018556553875547754,
+      "loss": 0.372,
+      "step": 2977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018555604844258582,
+      "loss": 0.7911,
+      "step": 2978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018554655525373806,
+      "loss": 0.7327,
+      "step": 2979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018553705918925337,
+      "loss": 0.494,
+      "step": 2980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018552756024945094,
+      "loss": 0.7982,
+      "step": 2981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001855180584346501,
+      "loss": 0.5308,
+      "step": 2982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018550855374517027,
+      "loss": 0.523,
+      "step": 2983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018549904618133086,
+      "loss": 0.3059,
+      "step": 2984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018548953574345158,
+      "loss": 0.4549,
+      "step": 2985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018548002243185205,
+      "loss": 0.5437,
+      "step": 2986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018547050624685208,
+      "loss": 0.6373,
+      "step": 2987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018546098718877154,
+      "loss": 0.879,
+      "step": 2988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018545146525793046,
+      "loss": 0.4015,
+      "step": 2989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018544194045464886,
+      "loss": 0.5103,
+      "step": 2990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018543241277924698,
+      "loss": 0.3875,
+      "step": 2991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018542288223204502,
+      "loss": 0.3752,
+      "step": 2992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018541334881336336,
+      "loss": 0.5772,
+      "step": 2993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001854038125235225,
+      "loss": 0.7274,
+      "step": 2994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018539427336284296,
+      "loss": 0.5007,
+      "step": 2995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018538473133164547,
+      "loss": 0.3778,
+      "step": 2996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001853751864302507,
+      "loss": 0.6677,
+      "step": 2997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018536563865897957,
+      "loss": 0.4784,
+      "step": 2998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018535608801815298,
+      "loss": 0.3509,
+      "step": 2999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018534653450809197,
+      "loss": 0.9756,
+      "step": 3000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018533697812911774,
+      "loss": 0.6465,
+      "step": 3001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018532741888155148,
+      "loss": 0.6505,
+      "step": 3002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018531785676571453,
+      "loss": 0.8613,
+      "step": 3003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018530829178192829,
+      "loss": 0.5596,
+      "step": 3004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001852987239305143,
+      "loss": 0.4527,
+      "step": 3005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018528915321179424,
+      "loss": 0.6291,
+      "step": 3006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018527957962608977,
+      "loss": 0.6887,
+      "step": 3007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001852700031737227,
+      "loss": 0.3835,
+      "step": 3008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000185260423855015,
+      "loss": 0.3178,
+      "step": 3009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001852508416702886,
+      "loss": 0.5873,
+      "step": 3010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018524125661986565,
+      "loss": 0.2542,
+      "step": 3011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018523166870406833,
+      "loss": 0.7204,
+      "step": 3012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018522207792321898,
+      "loss": 0.5955,
+      "step": 3013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018521248427763994,
+      "loss": 0.6946,
+      "step": 3014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018520288776765372,
+      "loss": 0.6563,
+      "step": 3015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018519328839358291,
+      "loss": 0.4434,
+      "step": 3016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018518368615575022,
+      "loss": 0.8099,
+      "step": 3017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018517408105447837,
+      "loss": 0.3919,
+      "step": 3018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018516447309009025,
+      "loss": 1.0146,
+      "step": 3019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018515486226290887,
+      "loss": 0.6942,
+      "step": 3020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018514524857325722,
+      "loss": 0.7934,
+      "step": 3021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018513563202145856,
+      "loss": 0.5968,
+      "step": 3022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018512601260783606,
+      "loss": 0.6044,
+      "step": 3023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018511639033271315,
+      "loss": 0.4319,
+      "step": 3024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018510676519641324,
+      "loss": 0.8206,
+      "step": 3025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018509713719925988,
+      "loss": 0.7105,
+      "step": 3026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018508750634157672,
+      "loss": 0.6446,
+      "step": 3027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018507787262368746,
+      "loss": 0.4777,
+      "step": 3028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018506823604591603,
+      "loss": 0.4688,
+      "step": 3029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018505859660858628,
+      "loss": 0.4093,
+      "step": 3030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001850489543120223,
+      "loss": 0.5727,
+      "step": 3031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018503930915654814,
+      "loss": 1.015,
+      "step": 3032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018502966114248808,
+      "loss": 0.486,
+      "step": 3033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018502001027016643,
+      "loss": 0.7269,
+      "step": 3034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018501035653990753,
+      "loss": 0.586,
+      "step": 3035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018500069995203603,
+      "loss": 0.5184,
+      "step": 3036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001849910405068764,
+      "loss": 0.623,
+      "step": 3037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018498137820475345,
+      "loss": 0.4615,
+      "step": 3038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018497171304599188,
+      "loss": 0.7316,
+      "step": 3039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018496204503091663,
+      "loss": 0.6878,
+      "step": 3040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001849523741598527,
+      "loss": 0.6289,
+      "step": 3041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018494270043312513,
+      "loss": 0.5823,
+      "step": 3042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018493302385105918,
+      "loss": 0.6606,
+      "step": 3043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018492334441398006,
+      "loss": 0.2789,
+      "step": 3044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018491366212221316,
+      "loss": 0.7932,
+      "step": 3045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018490397697608395,
+      "loss": 0.5569,
+      "step": 3046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000184894288975918,
+      "loss": 0.5493,
+      "step": 3047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018488459812204092,
+      "loss": 0.3323,
+      "step": 3048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018487490441477858,
+      "loss": 0.5542,
+      "step": 3049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018486520785445672,
+      "loss": 0.5622,
+      "step": 3050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018485550844140139,
+      "loss": 0.3715,
+      "step": 3051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018484580617593853,
+      "loss": 0.5693,
+      "step": 3052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018483610105839434,
+      "loss": 0.8504,
+      "step": 3053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018482639308909504,
+      "loss": 0.5022,
+      "step": 3054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000184816682268367,
+      "loss": 0.3022,
+      "step": 3055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001848069685965366,
+      "loss": 0.3994,
+      "step": 3056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018479725207393036,
+      "loss": 0.672,
+      "step": 3057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018478753270087495,
+      "loss": 0.7411,
+      "step": 3058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018477781047769704,
+      "loss": 0.4368,
+      "step": 3059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018476808540472347,
+      "loss": 0.5304,
+      "step": 3060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018475835748228112,
+      "loss": 0.4432,
+      "step": 3061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018474862671069702,
+      "loss": 0.8684,
+      "step": 3062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018473889309029824,
+      "loss": 0.294,
+      "step": 3063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000184729156621412,
+      "loss": 0.7842,
+      "step": 3064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001847194173043656,
+      "loss": 0.5012,
+      "step": 3065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001847096751394864,
+      "loss": 0.377,
+      "step": 3066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018469993012710186,
+      "loss": 1.0905,
+      "step": 3067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001846901822675396,
+      "loss": 0.4425,
+      "step": 3068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018468043156112728,
+      "loss": 0.442,
+      "step": 3069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001846706780081927,
+      "loss": 0.7888,
+      "step": 3070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018466092160906362,
+      "loss": 0.8038,
+      "step": 3071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018465116236406813,
+      "loss": 0.6237,
+      "step": 3072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018464140027353422,
+      "loss": 0.6651,
+      "step": 3073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018463163533779002,
+      "loss": 0.4797,
+      "step": 3074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018462186755716382,
+      "loss": 0.8343,
+      "step": 3075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018461209693198396,
+      "loss": 0.8316,
+      "step": 3076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001846023234625789,
+      "loss": 0.5759,
+      "step": 3077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018459254714927708,
+      "loss": 0.4604,
+      "step": 3078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018458276799240722,
+      "loss": 0.6435,
+      "step": 3079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018457298599229802,
+      "loss": 0.4954,
+      "step": 3080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001845632011492783,
+      "loss": 0.7885,
+      "step": 3081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018455341346367696,
+      "loss": 0.6524,
+      "step": 3082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018454362293582303,
+      "loss": 0.6407,
+      "step": 3083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018453382956604562,
+      "loss": 0.5748,
+      "step": 3084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018452403335467393,
+      "loss": 0.6585,
+      "step": 3085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018451423430203723,
+      "loss": 0.3931,
+      "step": 3086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018450443240846496,
+      "loss": 0.6272,
+      "step": 3087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018449462767428657,
+      "loss": 0.3939,
+      "step": 3088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018448482009983166,
+      "loss": 0.4881,
+      "step": 3089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018447500968542992,
+      "loss": 0.5358,
+      "step": 3090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018446519643141113,
+      "loss": 0.7008,
+      "step": 3091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018445538033810515,
+      "loss": 0.538,
+      "step": 3092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018444556140584194,
+      "loss": 0.4656,
+      "step": 3093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018443573963495156,
+      "loss": 0.5597,
+      "step": 3094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001844259150257642,
+      "loss": 0.71,
+      "step": 3095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018441608757861007,
+      "loss": 0.5089,
+      "step": 3096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018440625729381952,
+      "loss": 0.3873,
+      "step": 3097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018439642417172306,
+      "loss": 0.5793,
+      "step": 3098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018438658821265115,
+      "loss": 0.7636,
+      "step": 3099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018437674941693447,
+      "loss": 0.7497,
+      "step": 3100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018436690778490374,
+      "loss": 0.8177,
+      "step": 3101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018435706331688975,
+      "loss": 1.0044,
+      "step": 3102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001843472160132235,
+      "loss": 0.9633,
+      "step": 3103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001843373658742359,
+      "loss": 0.7174,
+      "step": 3104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018432751290025813,
+      "loss": 0.5896,
+      "step": 3105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001843176570916214,
+      "loss": 0.5516,
+      "step": 3106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018430779844865703,
+      "loss": 0.5976,
+      "step": 3107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018429793697169634,
+      "loss": 0.8318,
+      "step": 3108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018428807266107085,
+      "loss": 0.6677,
+      "step": 3109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001842782055171122,
+      "loss": 0.6929,
+      "step": 3110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018426833554015202,
+      "loss": 0.4167,
+      "step": 3111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018425846273052214,
+      "loss": 0.5289,
+      "step": 3112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018424858708855437,
+      "loss": 0.4752,
+      "step": 3113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001842387086145807,
+      "loss": 0.8749,
+      "step": 3114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001842288273089332,
+      "loss": 0.3734,
+      "step": 3115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018421894317194405,
+      "loss": 0.4358,
+      "step": 3116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018420905620394548,
+      "loss": 1.0212,
+      "step": 3117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018419916640526984,
+      "loss": 0.6095,
+      "step": 3118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018418927377624958,
+      "loss": 0.6735,
+      "step": 3119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001841793783172172,
+      "loss": 0.5412,
+      "step": 3120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001841694800285054,
+      "loss": 0.4063,
+      "step": 3121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018415957891044688,
+      "loss": 0.6708,
+      "step": 3122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018414967496337446,
+      "loss": 0.7938,
+      "step": 3123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018413976818762106,
+      "loss": 0.7329,
+      "step": 3124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001841298585835197,
+      "loss": 0.4316,
+      "step": 3125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001841199461514035,
+      "loss": 0.5912,
+      "step": 3126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018411003089160565,
+      "loss": 0.4674,
+      "step": 3127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018410011280445946,
+      "loss": 0.6969,
+      "step": 3128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018409019189029834,
+      "loss": 0.7754,
+      "step": 3129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018408026814945572,
+      "loss": 0.531,
+      "step": 3130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018407034158226529,
+      "loss": 0.4116,
+      "step": 3131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001840604121890606,
+      "loss": 0.5706,
+      "step": 3132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018405047997017553,
+      "loss": 0.7216,
+      "step": 3133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001840405449259439,
+      "loss": 0.6176,
+      "step": 3134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018403060705669968,
+      "loss": 0.6341,
+      "step": 3135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018402066636277696,
+      "loss": 0.7557,
+      "step": 3136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018401072284450986,
+      "loss": 0.6165,
+      "step": 3137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018400077650223263,
+      "loss": 0.5731,
+      "step": 3138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018399082733627965,
+      "loss": 0.6173,
+      "step": 3139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018398087534698534,
+      "loss": 0.7017,
+      "step": 3140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018397092053468422,
+      "loss": 0.5402,
+      "step": 3141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018396096289971094,
+      "loss": 0.5789,
+      "step": 3142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018395100244240022,
+      "loss": 0.7655,
+      "step": 3143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018394103916308688,
+      "loss": 0.7102,
+      "step": 3144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018393107306210584,
+      "loss": 0.6228,
+      "step": 3145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001839211041397921,
+      "loss": 0.4564,
+      "step": 3146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018391113239648075,
+      "loss": 0.8699,
+      "step": 3147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000183901157832507,
+      "loss": 0.7047,
+      "step": 3148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018389118044820617,
+      "loss": 0.4498,
+      "step": 3149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001838812002439136,
+      "loss": 0.5782,
+      "step": 3150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018387121721996484,
+      "loss": 0.4742,
+      "step": 3151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001838612313766954,
+      "loss": 0.486,
+      "step": 3152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018385124271444098,
+      "loss": 0.4626,
+      "step": 3153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018384125123353738,
+      "loss": 0.3918,
+      "step": 3154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018383125693432039,
+      "loss": 0.2804,
+      "step": 3155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000183821259817126,
+      "loss": 0.728,
+      "step": 3156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001838112598822903,
+      "loss": 0.6864,
+      "step": 3157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001838012571301494,
+      "loss": 0.5013,
+      "step": 3158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018379125156103957,
+      "loss": 0.5755,
+      "step": 3159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018378124317529709,
+      "loss": 0.6137,
+      "step": 3160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018377123197325842,
+      "loss": 0.47,
+      "step": 3161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018376121795526013,
+      "loss": 0.6491,
+      "step": 3162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018375120112163878,
+      "loss": 0.6589,
+      "step": 3163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001837411814727311,
+      "loss": 0.6711,
+      "step": 3164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001837311590088739,
+      "loss": 0.2401,
+      "step": 3165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018372113373040408,
+      "loss": 0.6108,
+      "step": 3166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018371110563765867,
+      "loss": 0.1792,
+      "step": 3167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018370107473097473,
+      "loss": 0.5689,
+      "step": 3168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018369104101068946,
+      "loss": 0.5751,
+      "step": 3169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018368100447714014,
+      "loss": 0.6261,
+      "step": 3170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018367096513066414,
+      "loss": 0.4749,
+      "step": 3171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018366092297159893,
+      "loss": 0.4511,
+      "step": 3172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001836508780002821,
+      "loss": 0.5512,
+      "step": 3173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018364083021705127,
+      "loss": 0.8177,
+      "step": 3174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018363077962224423,
+      "loss": 0.8374,
+      "step": 3175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018362072621619884,
+      "loss": 0.6767,
+      "step": 3176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018361066999925303,
+      "loss": 0.9313,
+      "step": 3177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018360061097174479,
+      "loss": 0.4365,
+      "step": 3178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018359054913401232,
+      "loss": 0.5408,
+      "step": 3179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001835804844863938,
+      "loss": 1.0847,
+      "step": 3180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018357041702922756,
+      "loss": 0.6003,
+      "step": 3181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018356034676285209,
+      "loss": 0.576,
+      "step": 3182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001835502736876058,
+      "loss": 0.6565,
+      "step": 3183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018354019780382735,
+      "loss": 0.5406,
+      "step": 3184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018353011911185542,
+      "loss": 0.4769,
+      "step": 3185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018352003761202878,
+      "loss": 0.5018,
+      "step": 3186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018350995330468638,
+      "loss": 0.5665,
+      "step": 3187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018349986619016717,
+      "loss": 0.6348,
+      "step": 3188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018348977626881022,
+      "loss": 0.4505,
+      "step": 3189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018347968354095472,
+      "loss": 0.6217,
+      "step": 3190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018346958800693992,
+      "loss": 0.4794,
+      "step": 3191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018345948966710517,
+      "loss": 0.7535,
+      "step": 3192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018344938852178993,
+      "loss": 0.4642,
+      "step": 3193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018343928457133379,
+      "loss": 0.5843,
+      "step": 3194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018342917781607632,
+      "loss": 0.5586,
+      "step": 3195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018341906825635737,
+      "loss": 0.7032,
+      "step": 3196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018340895589251662,
+      "loss": 0.5914,
+      "step": 3197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018339884072489414,
+      "loss": 0.7849,
+      "step": 3198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018338872275382986,
+      "loss": 0.3952,
+      "step": 3199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001833786019796639,
+      "loss": 0.9403,
+      "step": 3200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018336847840273653,
+      "loss": 0.6107,
+      "step": 3201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018335835202338798,
+      "loss": 0.52,
+      "step": 3202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001833482228419587,
+      "loss": 0.3587,
+      "step": 3203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018333809085878912,
+      "loss": 0.439,
+      "step": 3204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001833279560742199,
+      "loss": 0.7017,
+      "step": 3205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001833178184885917,
+      "loss": 0.5705,
+      "step": 3206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018330767810224524,
+      "loss": 0.4018,
+      "step": 3207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018329753491552145,
+      "loss": 0.7098,
+      "step": 3208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018328738892876126,
+      "loss": 0.7171,
+      "step": 3209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018327724014230573,
+      "loss": 0.8076,
+      "step": 3210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018326708855649602,
+      "loss": 0.8818,
+      "step": 3211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018325693417167336,
+      "loss": 0.6176,
+      "step": 3212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018324677698817912,
+      "loss": 0.8201,
+      "step": 3213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018323661700635468,
+      "loss": 0.6031,
+      "step": 3214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018322645422654162,
+      "loss": 0.3803,
+      "step": 3215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018321628864908154,
+      "loss": 0.8352,
+      "step": 3216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018320612027431614,
+      "loss": 0.4271,
+      "step": 3217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018319594910258725,
+      "loss": 0.5098,
+      "step": 3218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018318577513423673,
+      "loss": 0.5945,
+      "step": 3219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018317559836960662,
+      "loss": 0.479,
+      "step": 3220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018316541880903904,
+      "loss": 0.6306,
+      "step": 3221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001831552364528761,
+      "loss": 0.5562,
+      "step": 3222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018314505130146013,
+      "loss": 0.5124,
+      "step": 3223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018313486335513348,
+      "loss": 0.5472,
+      "step": 3224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018312467261423862,
+      "loss": 0.47,
+      "step": 3225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018311447907911813,
+      "loss": 0.5016,
+      "step": 3226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001831042827501146,
+      "loss": 0.5178,
+      "step": 3227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018309408362757087,
+      "loss": 0.4362,
+      "step": 3228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018308388171182972,
+      "loss": 0.6926,
+      "step": 3229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001830736770032341,
+      "loss": 0.7609,
+      "step": 3230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018306346950212708,
+      "loss": 0.549,
+      "step": 3231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018305325920885173,
+      "loss": 0.5648,
+      "step": 3232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018304304612375128,
+      "loss": 0.6352,
+      "step": 3233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018303283024716905,
+      "loss": 0.5307,
+      "step": 3234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018302261157944843,
+      "loss": 0.3545,
+      "step": 3235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018301239012093293,
+      "loss": 0.5581,
+      "step": 3236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018300216587196618,
+      "loss": 0.3323,
+      "step": 3237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001829919388328918,
+      "loss": 1.0304,
+      "step": 3238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018298170900405363,
+      "loss": 0.5703,
+      "step": 3239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018297147638579548,
+      "loss": 0.5144,
+      "step": 3240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001829612409784614,
+      "loss": 0.5631,
+      "step": 3241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001829510027823954,
+      "loss": 0.5157,
+      "step": 3242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001829407617979416,
+      "loss": 0.7015,
+      "step": 3243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018293051802544433,
+      "loss": 0.7317,
+      "step": 3244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001829202714652479,
+      "loss": 0.4976,
+      "step": 3245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001829100221176967,
+      "loss": 0.6683,
+      "step": 3246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018289976998313537,
+      "loss": 0.8802,
+      "step": 3247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018288951506190842,
+      "loss": 0.7653,
+      "step": 3248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001828792573543606,
+      "loss": 0.3407,
+      "step": 3249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018286899686083678,
+      "loss": 0.4615,
+      "step": 3250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018285873358168183,
+      "loss": 0.5909,
+      "step": 3251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001828484675172407,
+      "loss": 0.5417,
+      "step": 3252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018283819866785853,
+      "loss": 0.4688,
+      "step": 3253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018282792703388053,
+      "loss": 0.6619,
+      "step": 3254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018281765261565194,
+      "loss": 0.5102,
+      "step": 3255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018280737541351812,
+      "loss": 0.6088,
+      "step": 3256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018279709542782458,
+      "loss": 0.4739,
+      "step": 3257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018278681265891685,
+      "loss": 0.6817,
+      "step": 3258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001827765271071406,
+      "loss": 0.7125,
+      "step": 3259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018276623877284156,
+      "loss": 0.6295,
+      "step": 3260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001827559476563656,
+      "loss": 0.7031,
+      "step": 3261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018274565375805862,
+      "loss": 0.5305,
+      "step": 3262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018273535707826665,
+      "loss": 0.5627,
+      "step": 3263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018272505761733588,
+      "loss": 0.6267,
+      "step": 3264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018271475537561246,
+      "loss": 0.3968,
+      "step": 3265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018270445035344267,
+      "loss": 0.6749,
+      "step": 3266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000182694142551173,
+      "loss": 0.4917,
+      "step": 3267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018268383196914988,
+      "loss": 0.5519,
+      "step": 3268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001826735186077199,
+      "loss": 0.7159,
+      "step": 3269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018266320246722976,
+      "loss": 0.702,
+      "step": 3270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018265288354802625,
+      "loss": 0.491,
+      "step": 3271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018264256185045622,
+      "loss": 0.4494,
+      "step": 3272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018263223737486665,
+      "loss": 0.4175,
+      "step": 3273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001826219101216046,
+      "loss": 0.5125,
+      "step": 3274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018261158009101716,
+      "loss": 0.4968,
+      "step": 3275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018260124728345162,
+      "loss": 0.6106,
+      "step": 3276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018259091169925534,
+      "loss": 0.9854,
+      "step": 3277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018258057333877568,
+      "loss": 0.5682,
+      "step": 3278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018257023220236024,
+      "loss": 0.3971,
+      "step": 3279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001825598882903566,
+      "loss": 0.4834,
+      "step": 3280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018254954160311243,
+      "loss": 0.6662,
+      "step": 3281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001825391921409756,
+      "loss": 0.4678,
+      "step": 3282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018252883990429398,
+      "loss": 0.6199,
+      "step": 3283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018251848489341557,
+      "loss": 0.7752,
+      "step": 3284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001825081271086884,
+      "loss": 0.3125,
+      "step": 3285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018249776655046072,
+      "loss": 0.5176,
+      "step": 3286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018248740321908077,
+      "loss": 0.6365,
+      "step": 3287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018247703711489686,
+      "loss": 0.6452,
+      "step": 3288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001824666682382575,
+      "loss": 0.5507,
+      "step": 3289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018245629658951125,
+      "loss": 0.6692,
+      "step": 3290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018244592216900673,
+      "loss": 0.4238,
+      "step": 3291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001824355449770927,
+      "loss": 0.5561,
+      "step": 3292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018242516501411793,
+      "loss": 0.4275,
+      "step": 3293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018241478228043137,
+      "loss": 0.738,
+      "step": 3294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018240439677638207,
+      "loss": 0.482,
+      "step": 3295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018239400850231908,
+      "loss": 0.6311,
+      "step": 3296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018238361745859165,
+      "loss": 0.8831,
+      "step": 3297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018237322364554902,
+      "loss": 0.5948,
+      "step": 3298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018236282706354063,
+      "loss": 0.404,
+      "step": 3299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001823524277129159,
+      "loss": 0.6369,
+      "step": 3300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001823420255940245,
+      "loss": 0.6532,
+      "step": 3301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000182331620707216,
+      "loss": 0.7271,
+      "step": 3302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018232121305284023,
+      "loss": 0.7121,
+      "step": 3303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018231080263124698,
+      "loss": 0.4723,
+      "step": 3304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018230038944278622,
+      "loss": 0.4964,
+      "step": 3305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018228997348780797,
+      "loss": 0.5554,
+      "step": 3306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018227955476666244,
+      "loss": 0.5957,
+      "step": 3307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018226913327969974,
+      "loss": 0.5421,
+      "step": 3308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001822587090272703,
+      "loss": 0.4751,
+      "step": 3309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018224828200972445,
+      "loss": 0.5498,
+      "step": 3310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001822378522274127,
+      "loss": 0.5461,
+      "step": 3311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018222741968068563,
+      "loss": 0.6853,
+      "step": 3312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018221698436989402,
+      "loss": 0.6886,
+      "step": 3313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001822065462953886,
+      "loss": 0.6123,
+      "step": 3314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018219610545752016,
+      "loss": 0.6107,
+      "step": 3315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001821856618566398,
+      "loss": 0.5424,
+      "step": 3316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018217521549309854,
+      "loss": 0.7694,
+      "step": 3317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018216476636724748,
+      "loss": 0.5088,
+      "step": 3318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018215431447943792,
+      "loss": 0.502,
+      "step": 3319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018214385983002118,
+      "loss": 0.7465,
+      "step": 3320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001821334024193487,
+      "loss": 0.5999,
+      "step": 3321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018212294224777197,
+      "loss": 0.4251,
+      "step": 3322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018211247931564266,
+      "loss": 0.5153,
+      "step": 3323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018210201362331248,
+      "loss": 0.3962,
+      "step": 3324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018209154517113317,
+      "loss": 0.6184,
+      "step": 3325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018208107395945668,
+      "loss": 0.4952,
+      "step": 3326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018207059998863497,
+      "loss": 0.3927,
+      "step": 3327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018206012325902018,
+      "loss": 0.6335,
+      "step": 3328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001820496437709644,
+      "loss": 0.7651,
+      "step": 3329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018203916152481994,
+      "loss": 0.7163,
+      "step": 3330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001820286765209392,
+      "loss": 0.5594,
+      "step": 3331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001820181887596745,
+      "loss": 0.5674,
+      "step": 3332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018200769824137854,
+      "loss": 0.6126,
+      "step": 3333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018199720496640383,
+      "loss": 0.5254,
+      "step": 3334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018198670893510322,
+      "loss": 0.6369,
+      "step": 3335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018197621014782946,
+      "loss": 0.6441,
+      "step": 3336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018196570860493544,
+      "loss": 0.8338,
+      "step": 3337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018195520430677425,
+      "loss": 0.7609,
+      "step": 3338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018194469725369894,
+      "loss": 0.6068,
+      "step": 3339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018193418744606268,
+      "loss": 0.4825,
+      "step": 3340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001819236748842188,
+      "loss": 0.4825,
+      "step": 3341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001819131595685207,
+      "loss": 0.5018,
+      "step": 3342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018190264149932177,
+      "loss": 0.7122,
+      "step": 3343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018189212067697564,
+      "loss": 0.6635,
+      "step": 3344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018188159710183594,
+      "loss": 0.3931,
+      "step": 3345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001818710707742564,
+      "loss": 0.5508,
+      "step": 3346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018186054169459093,
+      "loss": 0.4793,
+      "step": 3347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001818500098631934,
+      "loss": 0.7195,
+      "step": 3348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018183947528041782,
+      "loss": 0.5249,
+      "step": 3349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018182893794661838,
+      "loss": 0.7435,
+      "step": 3350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018181839786214927,
+      "loss": 0.5069,
+      "step": 3351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018180785502736475,
+      "loss": 0.5309,
+      "step": 3352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018179730944261927,
+      "loss": 0.3874,
+      "step": 3353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018178676110826726,
+      "loss": 0.5512,
+      "step": 3354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018177621002466335,
+      "loss": 0.4377,
+      "step": 3355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001817656561921622,
+      "loss": 0.4705,
+      "step": 3356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018175509961111859,
+      "loss": 0.7159,
+      "step": 3357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018174454028188733,
+      "loss": 0.3393,
+      "step": 3358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001817339782048234,
+      "loss": 0.5755,
+      "step": 3359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018172341338028186,
+      "loss": 0.4368,
+      "step": 3360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018171284580861783,
+      "loss": 0.7167,
+      "step": 3361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018170227549018655,
+      "loss": 0.6822,
+      "step": 3362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018169170242534333,
+      "loss": 0.4504,
+      "step": 3363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018168112661444354,
+      "loss": 0.6864,
+      "step": 3364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018167054805784275,
+      "loss": 0.6668,
+      "step": 3365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018165996675589653,
+      "loss": 0.5451,
+      "step": 3366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018164938270896056,
+      "loss": 0.7796,
+      "step": 3367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018163879591739067,
+      "loss": 0.5909,
+      "step": 3368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018162820638154266,
+      "loss": 0.6204,
+      "step": 3369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018161761410177257,
+      "loss": 0.7396,
+      "step": 3370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001816070190784364,
+      "loss": 0.4208,
+      "step": 3371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018159642131189028,
+      "loss": 0.8027,
+      "step": 3372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018158582080249051,
+      "loss": 0.5919,
+      "step": 3373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001815752175505934,
+      "loss": 0.2698,
+      "step": 3374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018156461155655542,
+      "loss": 0.7779,
+      "step": 3375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018155400282073304,
+      "loss": 0.5434,
+      "step": 3376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001815433913434829,
+      "loss": 0.7057,
+      "step": 3377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018153277712516168,
+      "loss": 0.647,
+      "step": 3378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018152216016612614,
+      "loss": 0.7374,
+      "step": 3379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018151154046673327,
+      "loss": 0.528,
+      "step": 3380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018150091802733999,
+      "loss": 0.6737,
+      "step": 3381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018149029284830333,
+      "loss": 0.5339,
+      "step": 3382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018147966492998053,
+      "loss": 0.4559,
+      "step": 3383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018146903427272883,
+      "loss": 0.3819,
+      "step": 3384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018145840087690552,
+      "loss": 0.4849,
+      "step": 3385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001814477647428681,
+      "loss": 0.6625,
+      "step": 3386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018143712587097413,
+      "loss": 0.7046,
+      "step": 3387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018142648426158113,
+      "loss": 0.6505,
+      "step": 3388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018141583991504692,
+      "loss": 0.4922,
+      "step": 3389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001814051928317292,
+      "loss": 0.6863,
+      "step": 3390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000181394543011986,
+      "loss": 0.6061,
+      "step": 3391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001813838904561752,
+      "loss": 0.4912,
+      "step": 3392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018137323516465497,
+      "loss": 0.348,
+      "step": 3393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001813625771377834,
+      "loss": 0.4857,
+      "step": 3394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018135191637591886,
+      "loss": 0.7223,
+      "step": 3395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018134125287941965,
+      "loss": 0.4752,
+      "step": 3396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001813305866486442,
+      "loss": 0.569,
+      "step": 3397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001813199176839511,
+      "loss": 0.4892,
+      "step": 3398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018130924598569895,
+      "loss": 0.5355,
+      "step": 3399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018129857155424652,
+      "loss": 0.3407,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018128789438995256,
+      "loss": 0.7444,
+      "step": 3401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018127721449317607,
+      "loss": 0.5607,
+      "step": 3402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018126653186427603,
+      "loss": 0.7558,
+      "step": 3403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018125584650361147,
+      "loss": 0.6599,
+      "step": 3404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018124515841154163,
+      "loss": 0.6342,
+      "step": 3405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001812344675884258,
+      "loss": 0.584,
+      "step": 3406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018122377403462334,
+      "loss": 0.5964,
+      "step": 3407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018121307775049368,
+      "loss": 0.6357,
+      "step": 3408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018120237873639643,
+      "loss": 0.3847,
+      "step": 3409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018119167699269116,
+      "loss": 0.5172,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001811809725197377,
+      "loss": 0.5509,
+      "step": 3411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001811702653178958,
+      "loss": 0.3113,
+      "step": 3412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018115955538752543,
+      "loss": 0.7734,
+      "step": 3413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001811488427289866,
+      "loss": 0.6943,
+      "step": 3414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018113812734263938,
+      "loss": 0.6026,
+      "step": 3415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018112740922884398,
+      "loss": 0.9756,
+      "step": 3416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018111668838796072,
+      "loss": 0.6722,
+      "step": 3417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018110596482034995,
+      "loss": 0.6163,
+      "step": 3418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018109523852637214,
+      "loss": 0.5028,
+      "step": 3419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018108450950638788,
+      "loss": 0.5251,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018107377776075778,
+      "loss": 0.5649,
+      "step": 3421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018106304328984263,
+      "loss": 0.7165,
+      "step": 3422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018105230609400324,
+      "loss": 0.4994,
+      "step": 3423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018104156617360055,
+      "loss": 0.5156,
+      "step": 3424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018103082352899556,
+      "loss": 0.5938,
+      "step": 3425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018102007816054943,
+      "loss": 0.748,
+      "step": 3426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018100933006862333,
+      "loss": 0.654,
+      "step": 3427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018099857925357857,
+      "loss": 0.6781,
+      "step": 3428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018098782571577654,
+      "loss": 0.489,
+      "step": 3429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018097706945557866,
+      "loss": 0.3397,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001809663104733466,
+      "loss": 0.7548,
+      "step": 3431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018095554876944198,
+      "loss": 0.7008,
+      "step": 3432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018094478434422647,
+      "loss": 0.3095,
+      "step": 3433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018093401719806207,
+      "loss": 0.5118,
+      "step": 3434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018092324733131059,
+      "loss": 0.4756,
+      "step": 3435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018091247474433413,
+      "loss": 0.6609,
+      "step": 3436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 0.5473,
+      "step": 3437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018089092141115472,
+      "loss": 0.6123,
+      "step": 3438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018088014066567628,
+      "loss": 0.2934,
+      "step": 3439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001808693572014219,
+      "loss": 0.4496,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018085857101875403,
+      "loss": 0.6148,
+      "step": 3441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018084778211803522,
+      "loss": 0.505,
+      "step": 3442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018083699049962815,
+      "loss": 0.3698,
+      "step": 3443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001808261961638956,
+      "loss": 0.6031,
+      "step": 3444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018081539911120038,
+      "loss": 0.8809,
+      "step": 3445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001808045993419055,
+      "loss": 0.7394,
+      "step": 3446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001807937968563739,
+      "loss": 0.5936,
+      "step": 3447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001807829916549688,
+      "loss": 0.8288,
+      "step": 3448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018077218373805333,
+      "loss": 0.3133,
+      "step": 3449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018076137310599089,
+      "loss": 0.6761,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001807505597591448,
+      "loss": 0.3635,
+      "step": 3451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018073974369787855,
+      "loss": 0.4851,
+      "step": 3452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018072892492255576,
+      "loss": 0.4935,
+      "step": 3453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018071810343354012,
+      "loss": 0.5189,
+      "step": 3454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018070727923119532,
+      "loss": 0.7754,
+      "step": 3455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001806964523158853,
+      "loss": 0.5363,
+      "step": 3456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018068562268797392,
+      "loss": 0.5502,
+      "step": 3457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018067479034782527,
+      "loss": 0.507,
+      "step": 3458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018066395529580344,
+      "loss": 0.4564,
+      "step": 3459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018065311753227273,
+      "loss": 0.6196,
+      "step": 3460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018064227705759733,
+      "loss": 0.5628,
+      "step": 3461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018063143387214174,
+      "loss": 0.8097,
+      "step": 3462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001806205879762704,
+      "loss": 0.8264,
+      "step": 3463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018060973937034794,
+      "loss": 0.3866,
+      "step": 3464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018059888805473898,
+      "loss": 0.7708,
+      "step": 3465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001805880340298083,
+      "loss": 0.6057,
+      "step": 3466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018057717729592077,
+      "loss": 0.5777,
+      "step": 3467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018056631785344135,
+      "loss": 0.5045,
+      "step": 3468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001805554557027351,
+      "loss": 0.499,
+      "step": 3469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018054459084416706,
+      "loss": 0.359,
+      "step": 3470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001805337232781025,
+      "loss": 0.4492,
+      "step": 3471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018052285300490677,
+      "loss": 0.3746,
+      "step": 3472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018051198002494524,
+      "loss": 0.8323,
+      "step": 3473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018050110433858339,
+      "loss": 0.5671,
+      "step": 3474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001804902259461868,
+      "loss": 0.5103,
+      "step": 3475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001804793448481212,
+      "loss": 0.6683,
+      "step": 3476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001804684610447523,
+      "loss": 0.5273,
+      "step": 3477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018045757453644598,
+      "loss": 0.6179,
+      "step": 3478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001804466853235682,
+      "loss": 0.73,
+      "step": 3479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018043579340648496,
+      "loss": 0.3626,
+      "step": 3480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018042489878556243,
+      "loss": 0.8064,
+      "step": 3481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018041400146116682,
+      "loss": 0.5244,
+      "step": 3482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018040310143366446,
+      "loss": 0.4343,
+      "step": 3483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001803921987034217,
+      "loss": 0.3044,
+      "step": 3484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001803812932708051,
+      "loss": 0.4298,
+      "step": 3485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018037038513618118,
+      "loss": 0.6636,
+      "step": 3486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018035947429991667,
+      "loss": 0.3452,
+      "step": 3487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001803485607623783,
+      "loss": 0.6137,
+      "step": 3488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018033764452393295,
+      "loss": 0.655,
+      "step": 3489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001803267255849476,
+      "loss": 0.6853,
+      "step": 3490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018031580394578918,
+      "loss": 0.5947,
+      "step": 3491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001803048796068249,
+      "loss": 0.8454,
+      "step": 3492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000180293952568422,
+      "loss": 0.6156,
+      "step": 3493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018028302283094774,
+      "loss": 0.6763,
+      "step": 3494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018027209039476956,
+      "loss": 0.6403,
+      "step": 3495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018026115526025493,
+      "loss": 0.5748,
+      "step": 3496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018025021742777142,
+      "loss": 0.387,
+      "step": 3497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018023927689768673,
+      "loss": 0.7627,
+      "step": 3498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001802283336703686,
+      "loss": 0.7128,
+      "step": 3499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018021738774618492,
+      "loss": 0.5732,
+      "step": 3500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018020643912550362,
+      "loss": 0.2753,
+      "step": 3501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001801954878086927,
+      "loss": 0.619,
+      "step": 3502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018018453379612038,
+      "loss": 0.6847,
+      "step": 3503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018017357708815477,
+      "loss": 0.5708,
+      "step": 3504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018016261768516423,
+      "loss": 0.516,
+      "step": 3505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018015165558751717,
+      "loss": 0.6113,
+      "step": 3506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018014069079558205,
+      "loss": 0.6731,
+      "step": 3507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018012972330972747,
+      "loss": 0.9081,
+      "step": 3508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001801187531303221,
+      "loss": 0.4689,
+      "step": 3509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001801077802577347,
+      "loss": 0.7354,
+      "step": 3510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018009680469233415,
+      "loss": 0.725,
+      "step": 3511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001800858264344893,
+      "loss": 0.5695,
+      "step": 3512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018007484548456928,
+      "loss": 0.764,
+      "step": 3513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018006386184294315,
+      "loss": 0.7626,
+      "step": 3514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018005287550998018,
+      "loss": 0.4068,
+      "step": 3515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018004188648604964,
+      "loss": 0.5218,
+      "step": 3516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018003089477152092,
+      "loss": 0.5837,
+      "step": 3517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018001990036676353,
+      "loss": 0.2811,
+      "step": 3518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018000890327214702,
+      "loss": 0.6065,
+      "step": 3519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017999790348804107,
+      "loss": 0.5949,
+      "step": 3520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017998690101481542,
+      "loss": 0.5171,
+      "step": 3521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017997589585283993,
+      "loss": 0.7452,
+      "step": 3522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017996488800248455,
+      "loss": 0.8171,
+      "step": 3523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001799538774641193,
+      "loss": 0.4216,
+      "step": 3524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017994286423811428,
+      "loss": 0.6041,
+      "step": 3525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017993184832483972,
+      "loss": 0.416,
+      "step": 3526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001799208297246659,
+      "loss": 0.365,
+      "step": 3527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001799098084379632,
+      "loss": 0.5152,
+      "step": 3528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017989878446510215,
+      "loss": 0.6348,
+      "step": 3529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017988775780645327,
+      "loss": 0.8217,
+      "step": 3530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017987672846238724,
+      "loss": 0.5349,
+      "step": 3531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017986569643327482,
+      "loss": 0.8538,
+      "step": 3532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001798546617194868,
+      "loss": 0.4507,
+      "step": 3533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017984362432139418,
+      "loss": 0.5138,
+      "step": 3534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017983258423936795,
+      "loss": 0.5277,
+      "step": 3535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001798215414737792,
+      "loss": 0.7132,
+      "step": 3536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017981049602499915,
+      "loss": 0.5582,
+      "step": 3537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017979944789339912,
+      "loss": 0.446,
+      "step": 3538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017978839707935045,
+      "loss": 0.536,
+      "step": 3539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001797773435832246,
+      "loss": 0.4457,
+      "step": 3540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001797662874053932,
+      "loss": 0.623,
+      "step": 3541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017975522854622784,
+      "loss": 0.6135,
+      "step": 3542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017974416700610027,
+      "loss": 0.4754,
+      "step": 3543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017973310278538235,
+      "loss": 0.4423,
+      "step": 3544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017972203588444597,
+      "loss": 0.5306,
+      "step": 3545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017971096630366314,
+      "loss": 0.4977,
+      "step": 3546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017969989404340603,
+      "loss": 0.4183,
+      "step": 3547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017968881910404675,
+      "loss": 0.5457,
+      "step": 3548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017967774148595762,
+      "loss": 0.417,
+      "step": 3549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017966666118951103,
+      "loss": 0.777,
+      "step": 3550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017965557821507941,
+      "loss": 0.6141,
+      "step": 3551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001796444925630353,
+      "loss": 0.3492,
+      "step": 3552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017963340423375138,
+      "loss": 0.4428,
+      "step": 3553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001796223132276004,
+      "loss": 0.4585,
+      "step": 3554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017961121954495514,
+      "loss": 0.5142,
+      "step": 3555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017960012318618852,
+      "loss": 0.4542,
+      "step": 3556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017958902415167354,
+      "loss": 0.6707,
+      "step": 3557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017957792244178332,
+      "loss": 0.8236,
+      "step": 3558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000179566818056891,
+      "loss": 0.4015,
+      "step": 3559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017955571099736992,
+      "loss": 0.5217,
+      "step": 3560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017954460126359337,
+      "loss": 0.4891,
+      "step": 3561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017953348885593482,
+      "loss": 0.4595,
+      "step": 3562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017952237377476783,
+      "loss": 0.6528,
+      "step": 3563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017951125602046604,
+      "loss": 0.3077,
+      "step": 3564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017950013559340312,
+      "loss": 0.61,
+      "step": 3565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017948901249395296,
+      "loss": 0.3128,
+      "step": 3566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017947788672248942,
+      "loss": 0.4692,
+      "step": 3567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017946675827938647,
+      "loss": 0.5897,
+      "step": 3568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017945562716501824,
+      "loss": 0.5786,
+      "step": 3569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017944449337975885,
+      "loss": 0.5078,
+      "step": 3570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017943335692398259,
+      "loss": 0.4407,
+      "step": 3571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017942221779806384,
+      "loss": 0.677,
+      "step": 3572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017941107600237695,
+      "loss": 0.6877,
+      "step": 3573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017939993153729655,
+      "loss": 0.5451,
+      "step": 3574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001793887844031972,
+      "loss": 0.5405,
+      "step": 3575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017937763460045363,
+      "loss": 0.6618,
+      "step": 3576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017936648212944065,
+      "loss": 0.4492,
+      "step": 3577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001793553269905331,
+      "loss": 0.6147,
+      "step": 3578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017934416918410602,
+      "loss": 0.3932,
+      "step": 3579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017933300871053444,
+      "loss": 1.2031,
+      "step": 3580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017932184557019353,
+      "loss": 0.3982,
+      "step": 3581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017931067976345854,
+      "loss": 0.5257,
+      "step": 3582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017929951129070482,
+      "loss": 0.7261,
+      "step": 3583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017928834015230774,
+      "loss": 0.7843,
+      "step": 3584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017927716634864288,
+      "loss": 0.4553,
+      "step": 3585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017926598988008582,
+      "loss": 0.5829,
+      "step": 3586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017925481074701228,
+      "loss": 0.4147,
+      "step": 3587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000179243628949798,
+      "loss": 0.4532,
+      "step": 3588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001792324444888189,
+      "loss": 0.3885,
+      "step": 3589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001792212573644509,
+      "loss": 0.4052,
+      "step": 3590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001792100675770701,
+      "loss": 0.5301,
+      "step": 3591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017919887512705257,
+      "loss": 0.4643,
+      "step": 3592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017918768001477464,
+      "loss": 0.6417,
+      "step": 3593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017917648224061256,
+      "loss": 0.6228,
+      "step": 3594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017916528180494278,
+      "loss": 0.7417,
+      "step": 3595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017915407870814184,
+      "loss": 0.3572,
+      "step": 3596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001791428729505862,
+      "loss": 0.7369,
+      "step": 3597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001791316645326526,
+      "loss": 0.3067,
+      "step": 3598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001791204534547179,
+      "loss": 0.8207,
+      "step": 3599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001791092397171589,
+      "loss": 0.6418,
+      "step": 3600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017909802332035247,
+      "loss": 0.5194,
+      "step": 3601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017908680426467576,
+      "loss": 0.664,
+      "step": 3602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017907558255050586,
+      "loss": 0.4189,
+      "step": 3603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001790643581782199,
+      "loss": 0.7482,
+      "step": 3604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017905313114819534,
+      "loss": 0.459,
+      "step": 3605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001790419014608095,
+      "loss": 0.562,
+      "step": 3606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017903066911643983,
+      "loss": 0.6288,
+      "step": 3607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017901943411546394,
+      "loss": 0.7771,
+      "step": 3608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001790081964582595,
+      "loss": 0.4969,
+      "step": 3609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017899695614520425,
+      "loss": 0.6166,
+      "step": 3610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017898571317667602,
+      "loss": 0.6918,
+      "step": 3611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001789744675530528,
+      "loss": 0.4173,
+      "step": 3612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001789632192747125,
+      "loss": 0.5328,
+      "step": 3613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017895196834203335,
+      "loss": 0.2916,
+      "step": 3614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017894071475539348,
+      "loss": 0.715,
+      "step": 3615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001789294585151712,
+      "loss": 0.5028,
+      "step": 3616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017891819962174486,
+      "loss": 0.3443,
+      "step": 3617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017890693807549296,
+      "loss": 0.5797,
+      "step": 3618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017889567387679403,
+      "loss": 0.5991,
+      "step": 3619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001788844070260267,
+      "loss": 0.5839,
+      "step": 3620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001788731375235698,
+      "loss": 0.7731,
+      "step": 3621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017886186536980202,
+      "loss": 0.4856,
+      "step": 3622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001788505905651024,
+      "loss": 0.4831,
+      "step": 3623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017883931310984982,
+      "loss": 0.6682,
+      "step": 3624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001788280330044234,
+      "loss": 0.5823,
+      "step": 3625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001788167502492024,
+      "loss": 0.4077,
+      "step": 3626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017880546484456603,
+      "loss": 0.6703,
+      "step": 3627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017879417679089366,
+      "loss": 0.4342,
+      "step": 3628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001787828860885647,
+      "loss": 0.9106,
+      "step": 3629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001787715927379587,
+      "loss": 0.4242,
+      "step": 3630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001787602967394553,
+      "loss": 0.428,
+      "step": 3631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017874899809343427,
+      "loss": 0.5596,
+      "step": 3632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001787376968002753,
+      "loss": 0.6683,
+      "step": 3633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017872639286035835,
+      "loss": 0.296,
+      "step": 3634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017871508627406339,
+      "loss": 0.8207,
+      "step": 3635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017870377704177047,
+      "loss": 0.4487,
+      "step": 3636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017869246516385979,
+      "loss": 0.8665,
+      "step": 3637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017868115064071155,
+      "loss": 0.7216,
+      "step": 3638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017866983347270611,
+      "loss": 0.449,
+      "step": 3639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017865851366022392,
+      "loss": 0.4602,
+      "step": 3640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017864719120364543,
+      "loss": 0.3744,
+      "step": 3641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001786358661033513,
+      "loss": 0.5163,
+      "step": 3642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017862453835972223,
+      "loss": 0.6533,
+      "step": 3643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017861320797313892,
+      "loss": 0.5731,
+      "step": 3644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001786018749439823,
+      "loss": 0.7788,
+      "step": 3645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017859053927263334,
+      "loss": 0.7309,
+      "step": 3646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017857920095947307,
+      "loss": 0.3659,
+      "step": 3647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001785678600048826,
+      "loss": 0.9493,
+      "step": 3648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017855651640924324,
+      "loss": 0.5723,
+      "step": 3649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017854517017293616,
+      "loss": 0.3875,
+      "step": 3650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001785338212963429,
+      "loss": 0.6107,
+      "step": 3651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017852246977984488,
+      "loss": 0.6998,
+      "step": 3652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017851111562382368,
+      "loss": 0.5664,
+      "step": 3653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000178499758828661,
+      "loss": 0.5251,
+      "step": 3654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017848839939473858,
+      "loss": 0.5907,
+      "step": 3655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017847703732243826,
+      "loss": 0.5674,
+      "step": 3656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000178465672612142,
+      "loss": 0.9256,
+      "step": 3657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001784543052642318,
+      "loss": 0.4519,
+      "step": 3658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017844293527908978,
+      "loss": 0.3198,
+      "step": 3659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017843156265709813,
+      "loss": 0.569,
+      "step": 3660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017842018739863913,
+      "loss": 0.4178,
+      "step": 3661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001784088095040952,
+      "loss": 0.6036,
+      "step": 3662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017839742897384877,
+      "loss": 0.601,
+      "step": 3663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001783860458082824,
+      "loss": 0.705,
+      "step": 3664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017837466000777876,
+      "loss": 0.7064,
+      "step": 3665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017836327157272054,
+      "loss": 0.5216,
+      "step": 3666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017835188050349064,
+      "loss": 0.5723,
+      "step": 3667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017834048680047184,
+      "loss": 0.614,
+      "step": 3668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017832909046404724,
+      "loss": 0.5721,
+      "step": 3669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001783176914945999,
+      "loss": 0.37,
+      "step": 3670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000178306289892513,
+      "loss": 0.3891,
+      "step": 3671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017829488565816978,
+      "loss": 0.4006,
+      "step": 3672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017828347879195362,
+      "loss": 0.4391,
+      "step": 3673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017827206929424796,
+      "loss": 0.4576,
+      "step": 3674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001782606571654363,
+      "loss": 0.532,
+      "step": 3675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017824924240590227,
+      "loss": 0.317,
+      "step": 3676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017823782501602957,
+      "loss": 0.7376,
+      "step": 3677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017822640499620205,
+      "loss": 0.7648,
+      "step": 3678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017821498234680348,
+      "loss": 0.3527,
+      "step": 3679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017820355706821796,
+      "loss": 0.6145,
+      "step": 3680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017819212916082947,
+      "loss": 0.4361,
+      "step": 3681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017818069862502215,
+      "loss": 0.5935,
+      "step": 3682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017816926546118027,
+      "loss": 0.3813,
+      "step": 3683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017815782966968817,
+      "loss": 0.737,
+      "step": 3684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001781463912509302,
+      "loss": 0.2868,
+      "step": 3685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017813495020529093,
+      "loss": 0.4603,
+      "step": 3686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001781235065331549,
+      "loss": 0.4737,
+      "step": 3687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017811206023490684,
+      "loss": 0.8914,
+      "step": 3688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017810061131093145,
+      "loss": 0.5737,
+      "step": 3689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017808915976161362,
+      "loss": 0.6336,
+      "step": 3690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001780777055873383,
+      "loss": 0.4475,
+      "step": 3691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017806624878849049,
+      "loss": 0.6869,
+      "step": 3692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017805478936545535,
+      "loss": 0.608,
+      "step": 3693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017804332731861803,
+      "loss": 0.2114,
+      "step": 3694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017803186264836392,
+      "loss": 0.5319,
+      "step": 3695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001780203953550783,
+      "loss": 0.3175,
+      "step": 3696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017800892543914668,
+      "loss": 0.5899,
+      "step": 3697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017799745290095464,
+      "loss": 0.5205,
+      "step": 3698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017798597774088784,
+      "loss": 0.5227,
+      "step": 3699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017797449995933192,
+      "loss": 0.5766,
+      "step": 3700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017796301955667283,
+      "loss": 0.407,
+      "step": 3701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017795153653329642,
+      "loss": 0.4082,
+      "step": 3702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017794005088958868,
+      "loss": 0.4679,
+      "step": 3703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017792856262593572,
+      "loss": 0.4125,
+      "step": 3704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017791707174272372,
+      "loss": 0.554,
+      "step": 3705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017790557824033892,
+      "loss": 0.4317,
+      "step": 3706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017789408211916772,
+      "loss": 0.4276,
+      "step": 3707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001778825833795965,
+      "loss": 0.7462,
+      "step": 3708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017787108202201184,
+      "loss": 0.3703,
+      "step": 3709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001778595780468003,
+      "loss": 0.4427,
+      "step": 3710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017784807145434866,
+      "loss": 0.6135,
+      "step": 3711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017783656224504366,
+      "loss": 0.4729,
+      "step": 3712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017782505041927216,
+      "loss": 0.6945,
+      "step": 3713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017781353597742122,
+      "loss": 0.4973,
+      "step": 3714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001778020189198778,
+      "loss": 0.5207,
+      "step": 3715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001777904992470291,
+      "loss": 0.4798,
+      "step": 3716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001777789769592623,
+      "loss": 0.4401,
+      "step": 3717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001777674520569648,
+      "loss": 0.5485,
+      "step": 3718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017775592454052395,
+      "loss": 0.5099,
+      "step": 3719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017774439441032723,
+      "loss": 0.5881,
+      "step": 3720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017773286166676227,
+      "loss": 0.5694,
+      "step": 3721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017772132631021674,
+      "loss": 0.4512,
+      "step": 3722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001777097883410784,
+      "loss": 0.474,
+      "step": 3723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017769824775973504,
+      "loss": 0.292,
+      "step": 3724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017768670456657466,
+      "loss": 0.4279,
+      "step": 3725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017767515876198527,
+      "loss": 0.1815,
+      "step": 3726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017766361034635494,
+      "loss": 0.3701,
+      "step": 3727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017765205932007195,
+      "loss": 0.5392,
+      "step": 3728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017764050568352447,
+      "loss": 0.4097,
+      "step": 3729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017762894943710097,
+      "loss": 0.473,
+      "step": 3730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001776173905811899,
+      "loss": 0.6183,
+      "step": 3731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017760582911617977,
+      "loss": 0.6898,
+      "step": 3732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017759426504245927,
+      "loss": 0.3299,
+      "step": 3733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017758269836041704,
+      "loss": 0.3885,
+      "step": 3734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000177571129070442,
+      "loss": 0.6328,
+      "step": 3735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017755955717292296,
+      "loss": 0.3948,
+      "step": 3736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017754798266824895,
+      "loss": 0.5964,
+      "step": 3737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017753640555680906,
+      "loss": 0.7417,
+      "step": 3738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001775248258389924,
+      "loss": 0.5598,
+      "step": 3739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017751324351518829,
+      "loss": 0.6836,
+      "step": 3740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017750165858578602,
+      "loss": 0.5811,
+      "step": 3741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017749007105117504,
+      "loss": 0.6222,
+      "step": 3742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017747848091174482,
+      "loss": 0.5704,
+      "step": 3743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000177466888167885,
+      "loss": 0.5709,
+      "step": 3744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001774552928199853,
+      "loss": 0.7108,
+      "step": 3745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017744369486843542,
+      "loss": 0.5765,
+      "step": 3746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017743209431362526,
+      "loss": 0.375,
+      "step": 3747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001774204911559448,
+      "loss": 0.7859,
+      "step": 3748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017740888539578405,
+      "loss": 0.3861,
+      "step": 3749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017739727703353309,
+      "loss": 0.7262,
+      "step": 3750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017738566606958223,
+      "loss": 0.6097,
+      "step": 3751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017737405250432172,
+      "loss": 0.5086,
+      "step": 3752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017736243633814192,
+      "loss": 0.6912,
+      "step": 3753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017735081757143335,
+      "loss": 0.4041,
+      "step": 3754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017733919620458656,
+      "loss": 0.4687,
+      "step": 3755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017732757223799223,
+      "loss": 0.6895,
+      "step": 3756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017731594567204102,
+      "loss": 0.5902,
+      "step": 3757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017730431650712385,
+      "loss": 0.722,
+      "step": 3758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017729268474363154,
+      "loss": 0.5434,
+      "step": 3759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001772810503819552,
+      "loss": 0.3015,
+      "step": 3760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001772694134224858,
+      "loss": 0.753,
+      "step": 3761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017725777386561463,
+      "loss": 0.3997,
+      "step": 3762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017724613171173282,
+      "loss": 0.5477,
+      "step": 3763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017723448696123183,
+      "loss": 0.5838,
+      "step": 3764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017722283961450308,
+      "loss": 0.3476,
+      "step": 3765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017721118967193805,
+      "loss": 0.4934,
+      "step": 3766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017719953713392838,
+      "loss": 0.6281,
+      "step": 3767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017718788200086574,
+      "loss": 0.4262,
+      "step": 3768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017717622427314197,
+      "loss": 0.5718,
+      "step": 3769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001771645639511489,
+      "loss": 0.9732,
+      "step": 3770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001771529010352785,
+      "loss": 0.3588,
+      "step": 3771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017714123552592287,
+      "loss": 0.5888,
+      "step": 3772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017712956742347403,
+      "loss": 0.7675,
+      "step": 3773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001771178967283243,
+      "loss": 0.7179,
+      "step": 3774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017710622344086594,
+      "loss": 0.6775,
+      "step": 3775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017709454756149138,
+      "loss": 0.3833,
+      "step": 3776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017708286909059307,
+      "loss": 0.6548,
+      "step": 3777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017707118802856358,
+      "loss": 0.5269,
+      "step": 3778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001770595043757956,
+      "loss": 0.5514,
+      "step": 3779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017704781813268188,
+      "loss": 0.6158,
+      "step": 3780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017703612929961522,
+      "loss": 0.5604,
+      "step": 3781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001770244378769885,
+      "loss": 0.7613,
+      "step": 3782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017701274386519483,
+      "loss": 0.5293,
+      "step": 3783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017700104726462725,
+      "loss": 0.3411,
+      "step": 3784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017698934807567888,
+      "loss": 0.5594,
+      "step": 3785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001769776462987431,
+      "loss": 0.6599,
+      "step": 3786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017696594193421317,
+      "loss": 0.487,
+      "step": 3787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001769542349824826,
+      "loss": 0.947,
+      "step": 3788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017694252544394487,
+      "loss": 0.5261,
+      "step": 3789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001769308133189936,
+      "loss": 0.8622,
+      "step": 3790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001769190986080225,
+      "loss": 0.3222,
+      "step": 3791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017690738131142538,
+      "loss": 0.4873,
+      "step": 3792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001768956614295961,
+      "loss": 0.4941,
+      "step": 3793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001768839389629286,
+      "loss": 0.7855,
+      "step": 3794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017687221391181695,
+      "loss": 0.5045,
+      "step": 3795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001768604862766553,
+      "loss": 0.5774,
+      "step": 3796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017684875605783782,
+      "loss": 0.4884,
+      "step": 3797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001768370232557589,
+      "loss": 0.4791,
+      "step": 3798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017682528787081286,
+      "loss": 0.4904,
+      "step": 3799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017681354990339423,
+      "loss": 0.6882,
+      "step": 3800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017680180935389754,
+      "loss": 0.5201,
+      "step": 3801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001767900662227175,
+      "loss": 0.7752,
+      "step": 3802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017677832051024884,
+      "loss": 0.8386,
+      "step": 3803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017676657221688636,
+      "loss": 0.5293,
+      "step": 3804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000176754821343025,
+      "loss": 0.738,
+      "step": 3805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017674306788905976,
+      "loss": 0.323,
+      "step": 3806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017673131185538574,
+      "loss": 0.6137,
+      "step": 3807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017671955324239811,
+      "loss": 0.6411,
+      "step": 3808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017670779205049213,
+      "loss": 0.4493,
+      "step": 3809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017669602828006316,
+      "loss": 0.3529,
+      "step": 3810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001766842619315066,
+      "loss": 0.3498,
+      "step": 3811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000176672493005218,
+      "loss": 0.4005,
+      "step": 3812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017666072150159304,
+      "loss": 0.3291,
+      "step": 3813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017664894742102733,
+      "loss": 0.5661,
+      "step": 3814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017663717076391665,
+      "loss": 0.2129,
+      "step": 3815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001766253915306569,
+      "loss": 0.5523,
+      "step": 3816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017661360972164408,
+      "loss": 0.6601,
+      "step": 3817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001766018253372742,
+      "loss": 0.3219,
+      "step": 3818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017659003837794334,
+      "loss": 0.6655,
+      "step": 3819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017657824884404776,
+      "loss": 0.4658,
+      "step": 3820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017656645673598377,
+      "loss": 0.5526,
+      "step": 3821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017655466205414774,
+      "loss": 0.5814,
+      "step": 3822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017654286479893618,
+      "loss": 0.6794,
+      "step": 3823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017653106497074564,
+      "loss": 0.3238,
+      "step": 3824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017651926256997274,
+      "loss": 0.6936,
+      "step": 3825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017650745759701427,
+      "loss": 0.5576,
+      "step": 3826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017649565005226702,
+      "loss": 0.5384,
+      "step": 3827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001764838399361279,
+      "loss": 0.5526,
+      "step": 3828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001764720272489939,
+      "loss": 0.6246,
+      "step": 3829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017646021199126208,
+      "loss": 0.5099,
+      "step": 3830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017644839416332966,
+      "loss": 0.4231,
+      "step": 3831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017643657376559388,
+      "loss": 0.3985,
+      "step": 3832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017642475079845209,
+      "loss": 0.753,
+      "step": 3833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017641292526230168,
+      "loss": 0.6186,
+      "step": 3834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017640109715754017,
+      "loss": 0.6047,
+      "step": 3835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017638926648456518,
+      "loss": 0.2429,
+      "step": 3836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001763774332437744,
+      "loss": 0.46,
+      "step": 3837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001763655974355656,
+      "loss": 0.5133,
+      "step": 3838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001763537590603366,
+      "loss": 0.5066,
+      "step": 3839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017634191811848545,
+      "loss": 0.3524,
+      "step": 3840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017633007461041007,
+      "loss": 0.7363,
+      "step": 3841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001763182285365086,
+      "loss": 0.6876,
+      "step": 3842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001763063798971793,
+      "loss": 0.6555,
+      "step": 3843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017629452869282042,
+      "loss": 0.4278,
+      "step": 3844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017628267492383033,
+      "loss": 0.593,
+      "step": 3845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001762708185906075,
+      "loss": 0.3892,
+      "step": 3846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017625895969355048,
+      "loss": 0.6329,
+      "step": 3847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017624709823305793,
+      "loss": 0.2681,
+      "step": 3848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017623523420952854,
+      "loss": 0.6349,
+      "step": 3849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017622336762336112,
+      "loss": 0.4838,
+      "step": 3850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017621149847495458,
+      "loss": 0.4388,
+      "step": 3851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017619962676470788,
+      "loss": 0.66,
+      "step": 3852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001761877524930201,
+      "loss": 1.0327,
+      "step": 3853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001761758756602904,
+      "loss": 0.5428,
+      "step": 3854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017616399626691802,
+      "loss": 0.5126,
+      "step": 3855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017615211431330226,
+      "loss": 0.5709,
+      "step": 3856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017614022979984254,
+      "loss": 0.3176,
+      "step": 3857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017612834272693836,
+      "loss": 0.7759,
+      "step": 3858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001761164530949893,
+      "loss": 0.2861,
+      "step": 3859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000176104560904395,
+      "loss": 0.4995,
+      "step": 3860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001760926661555553,
+      "loss": 0.3697,
+      "step": 3861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017608076884886997,
+      "loss": 0.7005,
+      "step": 3862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017606886898473897,
+      "loss": 0.5915,
+      "step": 3863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017605696656356224,
+      "loss": 0.4951,
+      "step": 3864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017604506158573998,
+      "loss": 0.3,
+      "step": 3865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001760331540516723,
+      "loss": 0.5855,
+      "step": 3866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001760212439617595,
+      "loss": 0.494,
+      "step": 3867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017600933131640197,
+      "loss": 0.3607,
+      "step": 3868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017599741611600006,
+      "loss": 0.4799,
+      "step": 3869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001759854983609544,
+      "loss": 0.6935,
+      "step": 3870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017597357805166554,
+      "loss": 0.3403,
+      "step": 3871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017596165518853418,
+      "loss": 0.6352,
+      "step": 3872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017594972977196116,
+      "loss": 0.6975,
+      "step": 3873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001759378018023473,
+      "loss": 0.5454,
+      "step": 3874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017592587128009357,
+      "loss": 0.416,
+      "step": 3875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017591393820560097,
+      "loss": 0.5682,
+      "step": 3876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017590200257927075,
+      "loss": 0.6196,
+      "step": 3877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017589006440150398,
+      "loss": 0.5057,
+      "step": 3878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017587812367270207,
+      "loss": 0.4063,
+      "step": 3879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017586618039326638,
+      "loss": 0.3549,
+      "step": 3880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017585423456359834,
+      "loss": 0.5661,
+      "step": 3881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017584228618409953,
+      "loss": 0.3757,
+      "step": 3882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001758303352551716,
+      "loss": 0.4362,
+      "step": 3883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001758183817772163,
+      "loss": 0.3827,
+      "step": 3884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017580642575063538,
+      "loss": 0.3564,
+      "step": 3885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017579446717583082,
+      "loss": 0.7434,
+      "step": 3886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017578250605320455,
+      "loss": 0.5327,
+      "step": 3887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017577054238315866,
+      "loss": 0.4509,
+      "step": 3888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017575857616609534,
+      "loss": 0.6464,
+      "step": 3889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017574660740241673,
+      "loss": 0.304,
+      "step": 3890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017573463609252529,
+      "loss": 0.5131,
+      "step": 3891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017572266223682333,
+      "loss": 0.5313,
+      "step": 3892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017571068583571343,
+      "loss": 0.7028,
+      "step": 3893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001756987068895981,
+      "loss": 0.5307,
+      "step": 3894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017568672539888005,
+      "loss": 0.4021,
+      "step": 3895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017567474136396207,
+      "loss": 0.6455,
+      "step": 3896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017566275478524693,
+      "loss": 0.6282,
+      "step": 3897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017565076566313762,
+      "loss": 0.463,
+      "step": 3898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001756387739980371,
+      "loss": 0.4872,
+      "step": 3899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017562677979034852,
+      "loss": 0.6975,
+      "step": 3900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017561478304047503,
+      "loss": 0.6789,
+      "step": 3901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017560278374881988,
+      "loss": 1.1504,
+      "step": 3902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017559078191578648,
+      "loss": 0.3396,
+      "step": 3903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017557877754177824,
+      "loss": 0.5949,
+      "step": 3904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001755667706271987,
+      "loss": 0.6047,
+      "step": 3905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017555476117245145,
+      "loss": 0.7797,
+      "step": 3906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001755427491779402,
+      "loss": 0.4371,
+      "step": 3907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017553073464406872,
+      "loss": 0.6692,
+      "step": 3908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017551871757124087,
+      "loss": 0.2952,
+      "step": 3909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017550669795986064,
+      "loss": 0.4952,
+      "step": 3910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000175494675810332,
+      "loss": 0.5138,
+      "step": 3911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017548265112305917,
+      "loss": 0.437,
+      "step": 3912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001754706238984463,
+      "loss": 0.4215,
+      "step": 3913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017545859413689767,
+      "loss": 0.6282,
+      "step": 3914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017544656183881768,
+      "loss": 0.3544,
+      "step": 3915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001754345270046108,
+      "loss": 0.2571,
+      "step": 3916
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017542248963468157,
+      "loss": 0.5229,
+      "step": 3917
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017541044972943466,
+      "loss": 0.4534,
+      "step": 3918
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001753984072892747,
+      "loss": 0.56,
+      "step": 3919
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001753863623146066,
+      "loss": 0.8143,
+      "step": 3920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017537431480583516,
+      "loss": 0.4898,
+      "step": 3921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017536226476336543,
+      "loss": 0.5868,
+      "step": 3922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001753502121876024,
+      "loss": 0.7179,
+      "step": 3923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017533815707895128,
+      "loss": 0.5088,
+      "step": 3924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017532609943781728,
+      "loss": 0.7349,
+      "step": 3925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017531403926460574,
+      "loss": 0.2918,
+      "step": 3926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017530197655972197,
+      "loss": 0.8673,
+      "step": 3927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017528991132357156,
+      "loss": 0.6797,
+      "step": 3928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017527784355656002,
+      "loss": 0.441,
+      "step": 3929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017526577325909304,
+      "loss": 0.258,
+      "step": 3930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017525370043157637,
+      "loss": 0.7387,
+      "step": 3931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017524162507441578,
+      "loss": 0.7794,
+      "step": 3932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001752295471880172,
+      "loss": 0.6586,
+      "step": 3933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001752174667727867,
+      "loss": 0.5071,
+      "step": 3934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017520538382913026,
+      "loss": 0.5969,
+      "step": 3935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017519329835745411,
+      "loss": 0.5838,
+      "step": 3936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017518121035816448,
+      "loss": 0.4001,
+      "step": 3937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001751691198316677,
+      "loss": 0.4904,
+      "step": 3938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017515702677837022,
+      "loss": 0.5573,
+      "step": 3939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001751449311986785,
+      "loss": 0.7113,
+      "step": 3940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001751328330929992,
+      "loss": 0.5882,
+      "step": 3941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017512073246173896,
+      "loss": 0.5347,
+      "step": 3942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001751086293053045,
+      "loss": 0.3097,
+      "step": 3943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017509652362410273,
+      "loss": 0.4739,
+      "step": 3944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017508441541854052,
+      "loss": 0.4743,
+      "step": 3945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017507230468902493,
+      "loss": 0.6581,
+      "step": 3946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017506019143596308,
+      "loss": 0.5023,
+      "step": 3947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017504807565976211,
+      "loss": 0.6958,
+      "step": 3948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001750359573608293,
+      "loss": 0.3402,
+      "step": 3949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000175023836539572,
+      "loss": 0.523,
+      "step": 3950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001750117131963977,
+      "loss": 0.3048,
+      "step": 3951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017499958733171385,
+      "loss": 0.6615,
+      "step": 3952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017498745894592812,
+      "loss": 0.53,
+      "step": 3953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017497532803944816,
+      "loss": 0.4997,
+      "step": 3954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017496319461268178,
+      "loss": 0.313,
+      "step": 3955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017495105866603683,
+      "loss": 0.4748,
+      "step": 3956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017493892019992128,
+      "loss": 0.4688,
+      "step": 3957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017492677921474313,
+      "loss": 0.5804,
+      "step": 3958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001749146357109105,
+      "loss": 0.5248,
+      "step": 3959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001749024896888316,
+      "loss": 0.3075,
+      "step": 3960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017489034114891473,
+      "loss": 0.5667,
+      "step": 3961
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017487819009156827,
+      "loss": 0.5634,
+      "step": 3962
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017486603651720067,
+      "loss": 0.3821,
+      "step": 3963
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017485388042622041,
+      "loss": 0.2183,
+      "step": 3964
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001748417218190362,
+      "loss": 0.6385,
+      "step": 3965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017482956069605668,
+      "loss": 0.6947,
+      "step": 3966
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017481739705769073,
+      "loss": 0.4473,
+      "step": 3967
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017480523090434715,
+      "loss": 0.6286,
+      "step": 3968
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017479306223643494,
+      "loss": 0.6628,
+      "step": 3969
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017478089105436314,
+      "loss": 0.4563,
+      "step": 3970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017476871735854087,
+      "loss": 0.232,
+      "step": 3971
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001747565411493774,
+      "loss": 0.7617,
+      "step": 3972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017474436242728199,
+      "loss": 0.5872,
+      "step": 3973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000174732181192664,
+      "loss": 0.4623,
+      "step": 3974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017471999744593292,
+      "loss": 0.385,
+      "step": 3975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017470781118749832,
+      "loss": 0.4572,
+      "step": 3976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017469562241776988,
+      "loss": 0.6659,
+      "step": 3977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017468343113715725,
+      "loss": 0.4544,
+      "step": 3978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017467123734607026,
+      "loss": 0.498,
+      "step": 3979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017465904104491884,
+      "loss": 0.6648,
+      "step": 3980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001746468422341129,
+      "loss": 0.4645,
+      "step": 3981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017463464091406255,
+      "loss": 0.4925,
+      "step": 3982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017462243708517793,
+      "loss": 0.6781,
+      "step": 3983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017461023074786925,
+      "loss": 0.6065,
+      "step": 3984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017459802190254687,
+      "loss": 0.9325,
+      "step": 3985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017458581054962115,
+      "loss": 0.4364,
+      "step": 3986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017457359668950257,
+      "loss": 0.6361,
+      "step": 3987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001745613803226017,
+      "loss": 0.6443,
+      "step": 3988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017454916144932922,
+      "loss": 0.4352,
+      "step": 3989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017453694007009583,
+      "loss": 0.5239,
+      "step": 3990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017452471618531237,
+      "loss": 0.5173,
+      "step": 3991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017451248979538975,
+      "loss": 0.4691,
+      "step": 3992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017450026090073893,
+      "loss": 0.8281,
+      "step": 3993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017448802950177102,
+      "loss": 0.6507,
+      "step": 3994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017447579559889715,
+      "loss": 0.6952,
+      "step": 3995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001744635591925286,
+      "loss": 0.7055,
+      "step": 3996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017445132028307665,
+      "loss": 0.5671,
+      "step": 3997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001744390788709527,
+      "loss": 0.3439,
+      "step": 3998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001744268349565683,
+      "loss": 0.6509,
+      "step": 3999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000174414588540335,
+      "loss": 0.4927,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017440233962266442,
+      "loss": 0.7053,
+      "step": 4001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017439008820396837,
+      "loss": 0.5398,
+      "step": 4002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017437783428465867,
+      "loss": 0.6384,
+      "step": 4003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017436557786514717,
+      "loss": 0.7271,
+      "step": 4004
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017435331894584595,
+      "loss": 0.3345,
+      "step": 4005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017434105752716706,
+      "loss": 0.7177,
+      "step": 4006
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017432879360952265,
+      "loss": 0.5091,
+      "step": 4007
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017431652719332498,
+      "loss": 0.533,
+      "step": 4008
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001743042582789864,
+      "loss": 0.4636,
+      "step": 4009
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017429198686691932,
+      "loss": 0.8748,
+      "step": 4010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017427971295753625,
+      "loss": 0.5006,
+      "step": 4011
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017426743655124974,
+      "loss": 0.4981,
+      "step": 4012
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017425515764847251,
+      "loss": 0.3002,
+      "step": 4013
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017424287624961726,
+      "loss": 0.5084,
+      "step": 4014
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017423059235509688,
+      "loss": 0.4131,
+      "step": 4015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017421830596532422,
+      "loss": 0.6723,
+      "step": 4016
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017420601708071238,
+      "loss": 0.5372,
+      "step": 4017
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001741937257016744,
+      "loss": 0.3925,
+      "step": 4018
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017418143182862343,
+      "loss": 0.3489,
+      "step": 4019
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017416913546197278,
+      "loss": 0.4557,
+      "step": 4020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017415683660213573,
+      "loss": 0.5713,
+      "step": 4021
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017414453524952576,
+      "loss": 0.5042,
+      "step": 4022
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017413223140455637,
+      "loss": 0.5329,
+      "step": 4023
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017411992506764112,
+      "loss": 0.5872,
+      "step": 4024
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017410761623919369,
+      "loss": 0.6557,
+      "step": 4025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017409530491962788,
+      "loss": 0.5315,
+      "step": 4026
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001740829911093575,
+      "loss": 0.4025,
+      "step": 4027
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017407067480879648,
+      "loss": 0.4773,
+      "step": 4028
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017405835601835886,
+      "loss": 0.526,
+      "step": 4029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017404603473845866,
+      "loss": 0.4262,
+      "step": 4030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017403371096951019,
+      "loss": 0.5747,
+      "step": 4031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017402138471192755,
+      "loss": 0.4769,
+      "step": 4032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001740090559661252,
+      "loss": 0.4319,
+      "step": 4033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017399672473251753,
+      "loss": 0.6485,
+      "step": 4034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017398439101151905,
+      "loss": 0.5549,
+      "step": 4035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001739720548035444,
+      "loss": 0.7401,
+      "step": 4036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001739597161090082,
+      "loss": 0.7305,
+      "step": 4037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017394737492832524,
+      "loss": 0.8129,
+      "step": 4038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017393503126191036,
+      "loss": 0.6337,
+      "step": 4039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017392268511017852,
+      "loss": 0.5538,
+      "step": 4040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017391033647354469,
+      "loss": 0.6983,
+      "step": 4041
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017389798535242398,
+      "loss": 0.3269,
+      "step": 4042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017388563174723163,
+      "loss": 0.6457,
+      "step": 4043
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017387327565838282,
+      "loss": 0.419,
+      "step": 4044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017386091708629294,
+      "loss": 0.6233,
+      "step": 4045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001738485560313774,
+      "loss": 0.5382,
+      "step": 4046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017383619249405175,
+      "loss": 0.5149,
+      "step": 4047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017382382647473157,
+      "loss": 0.2752,
+      "step": 4048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017381145797383253,
+      "loss": 0.4287,
+      "step": 4049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001737990869917704,
+      "loss": 0.5222,
+      "step": 4050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001737867135289611,
+      "loss": 0.5264,
+      "step": 4051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017377433758582043,
+      "loss": 0.4968,
+      "step": 4052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017376195916276446,
+      "loss": 0.3654,
+      "step": 4053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017374957826020936,
+      "loss": 0.5728,
+      "step": 4054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017373719487857124,
+      "loss": 0.5699,
+      "step": 4055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017372480901826638,
+      "loss": 0.5846,
+      "step": 4056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017371242067971114,
+      "loss": 0.4907,
+      "step": 4057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017370002986332193,
+      "loss": 0.5171,
+      "step": 4058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017368763656951529,
+      "loss": 0.2969,
+      "step": 4059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001736752407987078,
+      "loss": 0.5527,
+      "step": 4060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017366284255131617,
+      "loss": 0.4855,
+      "step": 4061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017365044182775713,
+      "loss": 0.4437,
+      "step": 4062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017363803862844757,
+      "loss": 0.5055,
+      "step": 4063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017362563295380436,
+      "loss": 0.5555,
+      "step": 4064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001736132248042446,
+      "loss": 0.334,
+      "step": 4065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001736008141801853,
+      "loss": 0.4436,
+      "step": 4066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001735884010820437,
+      "loss": 0.3149,
+      "step": 4067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017357598551023706,
+      "loss": 0.458,
+      "step": 4068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017356356746518268,
+      "loss": 0.3038,
+      "step": 4069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017355114694729806,
+      "loss": 0.5199,
+      "step": 4070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017353872395700068,
+      "loss": 0.4608,
+      "step": 4071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017352629849470814,
+      "loss": 0.5749,
+      "step": 4072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017351387056083808,
+      "loss": 0.4868,
+      "step": 4073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017350144015580834,
+      "loss": 0.4038,
+      "step": 4074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001734890072800367,
+      "loss": 0.4157,
+      "step": 4075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017347657193394117,
+      "loss": 0.5594,
+      "step": 4076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017346413411793967,
+      "loss": 0.6438,
+      "step": 4077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017345169383245035,
+      "loss": 0.4793,
+      "step": 4078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017343925107789137,
+      "loss": 0.4933,
+      "step": 4079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000173426805854681,
+      "loss": 0.4137,
+      "step": 4080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017341435816323756,
+      "loss": 0.691,
+      "step": 4081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017340190800397954,
+      "loss": 0.6923,
+      "step": 4082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001733894553773254,
+      "loss": 0.7407,
+      "step": 4083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017337700028369373,
+      "loss": 0.4448,
+      "step": 4084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001733645427235032,
+      "loss": 0.4548,
+      "step": 4085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001733520826971726,
+      "loss": 0.7254,
+      "step": 4086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017333962020512077,
+      "loss": 0.4101,
+      "step": 4087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017332715524776666,
+      "loss": 0.2976,
+      "step": 4088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017331468782552921,
+      "loss": 0.4507,
+      "step": 4089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017330221793882754,
+      "loss": 0.3867,
+      "step": 4090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017328974558808086,
+      "loss": 0.5055,
+      "step": 4091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017327727077370836,
+      "loss": 0.511,
+      "step": 4092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017326479349612944,
+      "loss": 0.6558,
+      "step": 4093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017325231375576346,
+      "loss": 0.4018,
+      "step": 4094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017323983155303002,
+      "loss": 0.2558,
+      "step": 4095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001732273468883486,
+      "loss": 0.5057,
+      "step": 4096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017321485976213894,
+      "loss": 0.5804,
+      "step": 4097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001732023701748208,
+      "loss": 0.4737,
+      "step": 4098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017318987812681396,
+      "loss": 0.4895,
+      "step": 4099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017317738361853836,
+      "loss": 0.3919,
+      "step": 4100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017316488665041402,
+      "loss": 0.6057,
+      "step": 4101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000173152387222861,
+      "loss": 0.4885,
+      "step": 4102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001731398853362995,
+      "loss": 0.7612,
+      "step": 4103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017312738099114973,
+      "loss": 0.4147,
+      "step": 4104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017311487418783208,
+      "loss": 0.5421,
+      "step": 4105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001731023649267669,
+      "loss": 0.5262,
+      "step": 4106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017308985320837472,
+      "loss": 0.6736,
+      "step": 4107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017307733903307608,
+      "loss": 0.7018,
+      "step": 4108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017306482240129172,
+      "loss": 0.3364,
+      "step": 4109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017305230331344232,
+      "loss": 0.4936,
+      "step": 4110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017303978176994874,
+      "loss": 0.7142,
+      "step": 4111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017302725777123187,
+      "loss": 0.7161,
+      "step": 4112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017301473131771272,
+      "loss": 0.4191,
+      "step": 4113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017300220240981236,
+      "loss": 0.4423,
+      "step": 4114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017298967104795193,
+      "loss": 0.7368,
+      "step": 4115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017297713723255268,
+      "loss": 0.6189,
+      "step": 4116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017296460096403595,
+      "loss": 0.4992,
+      "step": 4117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017295206224282315,
+      "loss": 0.6096,
+      "step": 4118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017293952106933573,
+      "loss": 0.462,
+      "step": 4119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017292697744399527,
+      "loss": 0.3175,
+      "step": 4120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017291443136722344,
+      "loss": 0.555,
+      "step": 4121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017290188283944198,
+      "loss": 0.4422,
+      "step": 4122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017288933186107266,
+      "loss": 0.9117,
+      "step": 4123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017287677843253745,
+      "loss": 0.3588,
+      "step": 4124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017286422255425828,
+      "loss": 0.7714,
+      "step": 4125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017285166422665723,
+      "loss": 0.653,
+      "step": 4126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017283910345015647,
+      "loss": 0.4177,
+      "step": 4127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017282654022517818,
+      "loss": 0.5482,
+      "step": 4128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001728139745521447,
+      "loss": 0.3982,
+      "step": 4129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001728014064314784,
+      "loss": 0.6213,
+      "step": 4130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001727888358636018,
+      "loss": 0.4786,
+      "step": 4131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017277626284893747,
+      "loss": 0.4012,
+      "step": 4132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017276368738790795,
+      "loss": 0.3934,
+      "step": 4133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017275110948093608,
+      "loss": 0.3436,
+      "step": 4134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017273852912844456,
+      "loss": 0.3884,
+      "step": 4135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017272594633085638,
+      "loss": 0.498,
+      "step": 4136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017271336108859442,
+      "loss": 0.4152,
+      "step": 4137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001727007734020818,
+      "loss": 0.7777,
+      "step": 4138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001726881832717416,
+      "loss": 0.4028,
+      "step": 4139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017267559069799706,
+      "loss": 0.5955,
+      "step": 4140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017266299568127148,
+      "loss": 0.4092,
+      "step": 4141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017265039822198826,
+      "loss": 0.7032,
+      "step": 4142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017263779832057082,
+      "loss": 0.5848,
+      "step": 4143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017262519597744272,
+      "loss": 0.6451,
+      "step": 4144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001726125911930276,
+      "loss": 0.7644,
+      "step": 4145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017259998396774914,
+      "loss": 0.5344,
+      "step": 4146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017258737430203116,
+      "loss": 0.5994,
+      "step": 4147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017257476219629752,
+      "loss": 0.3519,
+      "step": 4148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017256214765097217,
+      "loss": 0.5392,
+      "step": 4149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017254953066647913,
+      "loss": 0.5116,
+      "step": 4150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017253691124324258,
+      "loss": 0.3873,
+      "step": 4151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017252428938168665,
+      "loss": 0.4403,
+      "step": 4152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017251166508223568,
+      "loss": 0.5942,
+      "step": 4153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017249903834531401,
+      "loss": 0.2802,
+      "step": 4154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017248640917134604,
+      "loss": 0.8224,
+      "step": 4155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017247377756075637,
+      "loss": 0.285,
+      "step": 4156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001724611435139696,
+      "loss": 0.4563,
+      "step": 4157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017244850703141036,
+      "loss": 0.485,
+      "step": 4158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017243586811350347,
+      "loss": 0.4476,
+      "step": 4159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001724232267606738,
+      "loss": 0.4116,
+      "step": 4160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017241058297334626,
+      "loss": 0.4592,
+      "step": 4161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001723979367519459,
+      "loss": 0.5267,
+      "step": 4162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001723852880968978,
+      "loss": 0.463,
+      "step": 4163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017237263700862712,
+      "loss": 0.7968,
+      "step": 4164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017235998348755915,
+      "loss": 0.5574,
+      "step": 4165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001723473275341192,
+      "loss": 0.5925,
+      "step": 4166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001723346691487328,
+      "loss": 0.725,
+      "step": 4167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017232200833182536,
+      "loss": 0.4262,
+      "step": 4168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001723093450838225,
+      "loss": 0.4898,
+      "step": 4169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017229667940514993,
+      "loss": 0.4487,
+      "step": 4170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017228401129623335,
+      "loss": 0.6049,
+      "step": 4171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001722713407574986,
+      "loss": 0.3044,
+      "step": 4172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017225866778937165,
+      "loss": 0.6958,
+      "step": 4173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017224599239227845,
+      "loss": 0.4376,
+      "step": 4174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001722333145666451,
+      "loss": 0.8738,
+      "step": 4175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017222063431289778,
+      "loss": 0.6074,
+      "step": 4176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017220795163146272,
+      "loss": 0.4252,
+      "step": 4177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017219526652276622,
+      "loss": 0.3295,
+      "step": 4178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017218257898723476,
+      "loss": 0.6525,
+      "step": 4179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017216988902529477,
+      "loss": 0.4938,
+      "step": 4180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017215719663737282,
+      "loss": 0.4155,
+      "step": 4181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017214450182389559,
+      "loss": 0.4958,
+      "step": 4182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001721318045852898,
+      "loss": 0.2726,
+      "step": 4183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017211910492198227,
+      "loss": 0.7718,
+      "step": 4184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001721064028343999,
+      "loss": 0.2731,
+      "step": 4185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017209369832296963,
+      "loss": 0.5933,
+      "step": 4186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017208099138811862,
+      "loss": 0.436,
+      "step": 4187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001720682820302739,
+      "loss": 0.6003,
+      "step": 4188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001720555702498628,
+      "loss": 0.2546,
+      "step": 4189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001720428560473125,
+      "loss": 0.5442,
+      "step": 4190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017203013942305052,
+      "loss": 0.4547,
+      "step": 4191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001720174203775042,
+      "loss": 0.6737,
+      "step": 4192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017200469891110118,
+      "loss": 0.52,
+      "step": 4193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017199197502426906,
+      "loss": 0.6496,
+      "step": 4194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017197924871743556,
+      "loss": 0.5196,
+      "step": 4195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001719665199910285,
+      "loss": 0.5268,
+      "step": 4196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017195378884547565,
+      "loss": 0.7429,
+      "step": 4197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017194105528120507,
+      "loss": 0.3823,
+      "step": 4198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017192831929864477,
+      "loss": 0.4697,
+      "step": 4199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017191558089822286,
+      "loss": 0.9468,
+      "step": 4200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017190284008036756,
+      "loss": 0.4393,
+      "step": 4201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001718900968455071,
+      "loss": 0.3324,
+      "step": 4202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017187735119406989,
+      "loss": 0.5397,
+      "step": 4203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017186460312648436,
+      "loss": 0.421,
+      "step": 4204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017185185264317903,
+      "loss": 0.4117,
+      "step": 4205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001718390997445825,
+      "loss": 0.5177,
+      "step": 4206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001718263444311235,
+      "loss": 0.4609,
+      "step": 4207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017181358670323075,
+      "loss": 0.5722,
+      "step": 4208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001718008265613331,
+      "loss": 0.484,
+      "step": 4209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001717880640058595,
+      "loss": 0.6043,
+      "step": 4210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000171775299037239,
+      "loss": 0.3448,
+      "step": 4211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017176253165590062,
+      "loss": 0.8906,
+      "step": 4212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017174976186227356,
+      "loss": 0.4808,
+      "step": 4213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017173698965678707,
+      "loss": 0.5206,
+      "step": 4214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001717242150398705,
+      "loss": 0.5835,
+      "step": 4215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001717114380119533,
+      "loss": 0.3055,
+      "step": 4216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017169865857346488,
+      "loss": 0.2091,
+      "step": 4217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017168587672483488,
+      "loss": 0.7059,
+      "step": 4218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017167309246649297,
+      "loss": 0.2018,
+      "step": 4219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017166030579886886,
+      "loss": 0.3135,
+      "step": 4220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017164751672239236,
+      "loss": 0.2247,
+      "step": 4221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017163472523749345,
+      "loss": 0.2949,
+      "step": 4222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017162193134460203,
+      "loss": 0.5865,
+      "step": 4223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001716091350441482,
+      "loss": 0.4105,
+      "step": 4224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001715963363365621,
+      "loss": 0.3892,
+      "step": 4225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017158353522227394,
+      "loss": 0.2977,
+      "step": 4226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017157073170171407,
+      "loss": 0.6317,
+      "step": 4227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017155792577531288,
+      "loss": 0.478,
+      "step": 4228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017154511744350078,
+      "loss": 0.5972,
+      "step": 4229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017153230670670838,
+      "loss": 0.4968,
+      "step": 4230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017151949356536628,
+      "loss": 0.6936,
+      "step": 4231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017150667801990517,
+      "loss": 0.402,
+      "step": 4232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001714938600707559,
+      "loss": 0.4368,
+      "step": 4233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017148103971834932,
+      "loss": 0.4407,
+      "step": 4234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017146821696311637,
+      "loss": 0.8883,
+      "step": 4235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001714553918054881,
+      "loss": 0.3379,
+      "step": 4236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017144256424589562,
+      "loss": 0.4783,
+      "step": 4237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017142973428477016,
+      "loss": 0.4274,
+      "step": 4238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017141690192254294,
+      "loss": 0.4659,
+      "step": 4239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017140406715964533,
+      "loss": 0.518,
+      "step": 4240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001713912299965088,
+      "loss": 0.4731,
+      "step": 4241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017137839043356484,
+      "loss": 0.2379,
+      "step": 4242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017136554847124507,
+      "loss": 0.3708,
+      "step": 4243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001713527041099812,
+      "loss": 0.3657,
+      "step": 4244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017133985735020492,
+      "loss": 0.2811,
+      "step": 4245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017132700819234812,
+      "loss": 0.7337,
+      "step": 4246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001713141566368427,
+      "loss": 0.4403,
+      "step": 4247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017130130268412067,
+      "loss": 0.364,
+      "step": 4248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017128844633461417,
+      "loss": 0.9474,
+      "step": 4249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017127558758875527,
+      "loss": 0.7563,
+      "step": 4250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017126272644697626,
+      "loss": 0.4911,
+      "step": 4251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017124986290970948,
+      "loss": 0.3608,
+      "step": 4252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017123699697738732,
+      "loss": 0.4704,
+      "step": 4253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017122412865044225,
+      "loss": 0.5772,
+      "step": 4254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017121125792930684,
+      "loss": 0.3913,
+      "step": 4255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017119838481441382,
+      "loss": 0.5539,
+      "step": 4256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017118550930619582,
+      "loss": 0.418,
+      "step": 4257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017117263140508567,
+      "loss": 0.9109,
+      "step": 4258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017115975111151626,
+      "loss": 0.655,
+      "step": 4259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001711468684259206,
+      "loss": 0.8312,
+      "step": 4260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017113398334873164,
+      "loss": 0.3956,
+      "step": 4261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017112109588038264,
+      "loss": 0.5072,
+      "step": 4262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017110820602130673,
+      "loss": 0.5665,
+      "step": 4263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017109531377193724,
+      "loss": 0.266,
+      "step": 4264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001710824191327075,
+      "loss": 0.4667,
+      "step": 4265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017106952210405098,
+      "loss": 0.4463,
+      "step": 4266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017105662268640122,
+      "loss": 0.3269,
+      "step": 4267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017104372088019184,
+      "loss": 0.4295,
+      "step": 4268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017103081668585647,
+      "loss": 0.5532,
+      "step": 4269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000171017910103829,
+      "loss": 0.5199,
+      "step": 4270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017100500113454318,
+      "loss": 0.6525,
+      "step": 4271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000170992089778433,
+      "loss": 0.5043,
+      "step": 4272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017097917603593244,
+      "loss": 0.4649,
+      "step": 4273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001709662599074756,
+      "loss": 0.439,
+      "step": 4274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017095334139349667,
+      "loss": 0.6896,
+      "step": 4275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001709404204944299,
+      "loss": 0.7092,
+      "step": 4276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001709274972107096,
+      "loss": 0.6454,
+      "step": 4277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017091457154277023,
+      "loss": 0.4095,
+      "step": 4278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017090164349104625,
+      "loss": 0.2882,
+      "step": 4279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017088871305597224,
+      "loss": 0.3701,
+      "step": 4280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017087578023798286,
+      "loss": 0.513,
+      "step": 4281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017086284503751284,
+      "loss": 0.5144,
+      "step": 4282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000170849907454997,
+      "loss": 0.5321,
+      "step": 4283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017083696749087027,
+      "loss": 0.5523,
+      "step": 4284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017082402514556754,
+      "loss": 0.5041,
+      "step": 4285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017081108041952395,
+      "loss": 0.5892,
+      "step": 4286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017079813331317455,
+      "loss": 0.361,
+      "step": 4287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017078518382695465,
+      "loss": 0.3934,
+      "step": 4288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017077223196129948,
+      "loss": 0.4187,
+      "step": 4289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017075927771664445,
+      "loss": 0.6612,
+      "step": 4290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000170746321093425,
+      "loss": 0.399,
+      "step": 4291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017073336209207665,
+      "loss": 0.6979,
+      "step": 4292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017072040071303507,
+      "loss": 0.3361,
+      "step": 4293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001707074369567359,
+      "loss": 0.3979,
+      "step": 4294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017069447082361495,
+      "loss": 0.3422,
+      "step": 4295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017068150231410803,
+      "loss": 0.51,
+      "step": 4296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017066853142865113,
+      "loss": 0.3769,
+      "step": 4297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017065555816768018,
+      "loss": 0.729,
+      "step": 4298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001706425825316314,
+      "loss": 0.3241,
+      "step": 4299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017062960452094084,
+      "loss": 0.643,
+      "step": 4300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017061662413604486,
+      "loss": 0.4659,
+      "step": 4301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001706036413773797,
+      "loss": 0.5192,
+      "step": 4302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017059065624538185,
+      "loss": 0.3769,
+      "step": 4303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017057766874048778,
+      "loss": 0.4346,
+      "step": 4304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000170564678863134,
+      "loss": 0.6151,
+      "step": 4305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017055168661375727,
+      "loss": 0.4633,
+      "step": 4306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017053869199279426,
+      "loss": 0.4662,
+      "step": 4307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017052569500068177,
+      "loss": 0.2696,
+      "step": 4308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017051269563785675,
+      "loss": 0.6806,
+      "step": 4309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017049969390475615,
+      "loss": 0.2546,
+      "step": 4310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017048668980181698,
+      "loss": 0.3917,
+      "step": 4311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001704736833294764,
+      "loss": 0.2064,
+      "step": 4312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017046067448817163,
+      "loss": 0.3334,
+      "step": 4313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017044766327833994,
+      "loss": 0.3341,
+      "step": 4314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017043464970041872,
+      "loss": 0.6336,
+      "step": 4315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017042163375484542,
+      "loss": 0.3386,
+      "step": 4316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017040861544205757,
+      "loss": 0.3862,
+      "step": 4317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017039559476249275,
+      "loss": 0.6004,
+      "step": 4318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017038257171658869,
+      "loss": 0.4483,
+      "step": 4319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017036954630478312,
+      "loss": 0.4931,
+      "step": 4320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001703565185275139,
+      "loss": 0.5975,
+      "step": 4321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017034348838521898,
+      "loss": 0.7319,
+      "step": 4322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001703304558783364,
+      "loss": 0.7478,
+      "step": 4323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001703174210073041,
+      "loss": 0.3859,
+      "step": 4324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017030438377256036,
+      "loss": 0.2061,
+      "step": 4325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017029134417454348,
+      "loss": 0.3014,
+      "step": 4326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017027830221369162,
+      "loss": 0.3928,
+      "step": 4327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017026525789044333,
+      "loss": 0.4797,
+      "step": 4328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017025221120523702,
+      "loss": 0.3545,
+      "step": 4329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001702391621585113,
+      "loss": 0.2914,
+      "step": 4330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017022611075070474,
+      "loss": 0.4979,
+      "step": 4331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001702130569822561,
+      "loss": 0.6248,
+      "step": 4332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001702000008536042,
+      "loss": 0.5184,
+      "step": 4333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001701869423651879,
+      "loss": 0.3902,
+      "step": 4334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017017388151744617,
+      "loss": 0.4815,
+      "step": 4335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017016081831081807,
+      "loss": 0.4466,
+      "step": 4336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001701477527457427,
+      "loss": 0.5838,
+      "step": 4337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017013468482265917,
+      "loss": 0.4643,
+      "step": 4338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017012161454200688,
+      "loss": 0.3802,
+      "step": 4339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017010854190422517,
+      "loss": 0.4568,
+      "step": 4340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017009546690975336,
+      "loss": 0.3209,
+      "step": 4341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017008238955903114,
+      "loss": 0.2684,
+      "step": 4342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000170069309852498,
+      "loss": 0.4572,
+      "step": 4343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001700562277905936,
+      "loss": 0.5314,
+      "step": 4344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017004314337375773,
+      "loss": 0.4509,
+      "step": 4345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001700300566024302,
+      "loss": 0.5582,
+      "step": 4346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017001696747705094,
+      "loss": 0.3435,
+      "step": 4347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001700038759980599,
+      "loss": 0.3812,
+      "step": 4348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016999078216589725,
+      "loss": 0.2645,
+      "step": 4349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016997768598100304,
+      "loss": 0.4485,
+      "step": 4350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016996458744381748,
+      "loss": 0.5399,
+      "step": 4351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016995148655478092,
+      "loss": 0.382,
+      "step": 4352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001699383833143338,
+      "loss": 0.4748,
+      "step": 4353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016992527772291647,
+      "loss": 0.8375,
+      "step": 4354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016991216978096954,
+      "loss": 0.5489,
+      "step": 4355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016989905948893366,
+      "loss": 0.613,
+      "step": 4356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016988594684724947,
+      "loss": 0.3336,
+      "step": 4357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016987283185635775,
+      "loss": 0.3147,
+      "step": 4358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001698597145166994,
+      "loss": 0.3272,
+      "step": 4359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001698465948287153,
+      "loss": 0.5095,
+      "step": 4360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016983347279284655,
+      "loss": 0.4044,
+      "step": 4361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001698203484095342,
+      "loss": 0.4147,
+      "step": 4362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016980722167921942,
+      "loss": 0.5088,
+      "step": 4363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016979409260234346,
+      "loss": 0.3907,
+      "step": 4364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016978096117934767,
+      "loss": 0.5102,
+      "step": 4365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016976782741067343,
+      "loss": 0.7753,
+      "step": 4366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016975469129676228,
+      "loss": 0.3866,
+      "step": 4367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016974155283805574,
+      "loss": 0.5056,
+      "step": 4368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001697284120349955,
+      "loss": 0.5531,
+      "step": 4369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016971526888802325,
+      "loss": 0.7404,
+      "step": 4370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001697021233975808,
+      "loss": 0.4936,
+      "step": 4371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016968897556411005,
+      "loss": 0.7887,
+      "step": 4372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016967582538805298,
+      "loss": 0.4005,
+      "step": 4373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001696626728698516,
+      "loss": 0.4426,
+      "step": 4374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016964951800994805,
+      "loss": 0.9455,
+      "step": 4375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001696363608087845,
+      "loss": 0.4587,
+      "step": 4376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016962320126680323,
+      "loss": 0.6862,
+      "step": 4377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016961003938444663,
+      "loss": 0.5151,
+      "step": 4378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001695968751621571,
+      "loss": 0.3312,
+      "step": 4379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016958370860037717,
+      "loss": 0.3025,
+      "step": 4380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016957053969954947,
+      "loss": 0.3808,
+      "step": 4381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016955736846011661,
+      "loss": 0.5416,
+      "step": 4382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016954419488252138,
+      "loss": 0.6885,
+      "step": 4383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016953101896720658,
+      "loss": 0.7631,
+      "step": 4384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016951784071461514,
+      "loss": 0.6551,
+      "step": 4385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016950466012519,
+      "loss": 0.3486,
+      "step": 4386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016949147719937428,
+      "loss": 0.7829,
+      "step": 4387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016947829193761114,
+      "loss": 0.2064,
+      "step": 4388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001694651043403437,
+      "loss": 0.8761,
+      "step": 4389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016945191440801533,
+      "loss": 0.3382,
+      "step": 4390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001694387221410694,
+      "loss": 0.4808,
+      "step": 4391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001694255275399494,
+      "loss": 0.3126,
+      "step": 4392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001694123306050988,
+      "loss": 0.6299,
+      "step": 4393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016939913133696125,
+      "loss": 0.2929,
+      "step": 4394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016938592973598046,
+      "loss": 0.4652,
+      "step": 4395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016937272580260014,
+      "loss": 0.3583,
+      "step": 4396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016935951953726418,
+      "loss": 0.4008,
+      "step": 4397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016934631094041652,
+      "loss": 0.5052,
+      "step": 4398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016933310001250113,
+      "loss": 0.3594,
+      "step": 4399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001693198867539621,
+      "loss": 0.4154,
+      "step": 4400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016930667116524364,
+      "loss": 0.3086,
+      "step": 4401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016929345324678995,
+      "loss": 0.3992,
+      "step": 4402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016928023299904533,
+      "loss": 0.533,
+      "step": 4403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016926701042245418,
+      "loss": 0.4688,
+      "step": 4404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016925378551746101,
+      "loss": 0.3725,
+      "step": 4405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016924055828451034,
+      "loss": 0.4442,
+      "step": 4406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016922732872404686,
+      "loss": 0.402,
+      "step": 4407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001692140968365152,
+      "loss": 0.3777,
+      "step": 4408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001692008626223602,
+      "loss": 0.4752,
+      "step": 4409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016918762608202675,
+      "loss": 0.4638,
+      "step": 4410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001691743872159597,
+      "loss": 0.4744,
+      "step": 4411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016916114602460417,
+      "loss": 0.6765,
+      "step": 4412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001691479025084052,
+      "loss": 0.3183,
+      "step": 4413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000169134656667808,
+      "loss": 0.5612,
+      "step": 4414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016912140850325783,
+      "loss": 0.4742,
+      "step": 4415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001691081580152,
+      "loss": 0.5213,
+      "step": 4416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016909490520407996,
+      "loss": 0.799,
+      "step": 4417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016908165007034317,
+      "loss": 0.6377,
+      "step": 4418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016906839261443519,
+      "loss": 0.4483,
+      "step": 4419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001690551328368017,
+      "loss": 0.623,
+      "step": 4420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001690418707378884,
+      "loss": 0.4794,
+      "step": 4421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016902860631814112,
+      "loss": 0.5446,
+      "step": 4422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016901533957800572,
+      "loss": 0.3312,
+      "step": 4423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016900207051792819,
+      "loss": 0.5311,
+      "step": 4424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016898879913835452,
+      "loss": 0.4339,
+      "step": 4425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016897552543973084,
+      "loss": 0.4577,
+      "step": 4426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001689622494225034,
+      "loss": 0.379,
+      "step": 4427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001689489710871184,
+      "loss": 0.2819,
+      "step": 4428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001689356904340222,
+      "loss": 0.5248,
+      "step": 4429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001689224074636613,
+      "loss": 0.5696,
+      "step": 4430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016890912217648206,
+      "loss": 0.4867,
+      "step": 4431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016889583457293123,
+      "loss": 0.3729,
+      "step": 4432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001688825446534554,
+      "loss": 0.5107,
+      "step": 4433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016886925241850123,
+      "loss": 0.7426,
+      "step": 4434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016885595786851566,
+      "loss": 0.6858,
+      "step": 4435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016884266100394553,
+      "loss": 0.7134,
+      "step": 4436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001688293618252378,
+      "loss": 0.5606,
+      "step": 4437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016881606033283957,
+      "loss": 0.4226,
+      "step": 4438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016880275652719792,
+      "loss": 0.4894,
+      "step": 4439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016878945040876007,
+      "loss": 0.3539,
+      "step": 4440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001687761419779733,
+      "loss": 0.4562,
+      "step": 4441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000168762831235285,
+      "loss": 0.586,
+      "step": 4442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016874951818114255,
+      "loss": 0.828,
+      "step": 4443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016873620281599352,
+      "loss": 0.5264,
+      "step": 4444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016872288514028548,
+      "loss": 0.371,
+      "step": 4445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016870956515446613,
+      "loss": 0.7311,
+      "step": 4446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016869624285898317,
+      "loss": 0.5284,
+      "step": 4447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016868291825428443,
+      "loss": 0.4375,
+      "step": 4448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001686695913408179,
+      "loss": 0.3333,
+      "step": 4449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016865626211903147,
+      "loss": 0.5177,
+      "step": 4450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016864293058937324,
+      "loss": 0.2987,
+      "step": 4451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016862959675229133,
+      "loss": 0.6676,
+      "step": 4452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016861626060823398,
+      "loss": 0.4163,
+      "step": 4453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016860292215764946,
+      "loss": 0.5816,
+      "step": 4454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001685895814009862,
+      "loss": 0.411,
+      "step": 4455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016857623833869248,
+      "loss": 0.2684,
+      "step": 4456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016856289297121705,
+      "loss": 0.4306,
+      "step": 4457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016854954529900836,
+      "loss": 0.6684,
+      "step": 4458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016853619532251513,
+      "loss": 0.53,
+      "step": 4459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016852284304218613,
+      "loss": 0.4103,
+      "step": 4460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016850948845847021,
+      "loss": 0.5584,
+      "step": 4461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016849613157181623,
+      "loss": 0.4753,
+      "step": 4462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016848277238267322,
+      "loss": 0.6903,
+      "step": 4463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016846941089149024,
+      "loss": 0.4568,
+      "step": 4464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016845604709871646,
+      "loss": 0.5438,
+      "step": 4465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016844268100480106,
+      "loss": 0.555,
+      "step": 4466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016842931261019334,
+      "loss": 0.2921,
+      "step": 4467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001684159419153427,
+      "loss": 0.468,
+      "step": 4468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001684025689206986,
+      "loss": 0.38,
+      "step": 4469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016838919362671052,
+      "loss": 0.5316,
+      "step": 4470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016837581603382814,
+      "loss": 0.7758,
+      "step": 4471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016836243614250113,
+      "loss": 0.5226,
+      "step": 4472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001683490539531792,
+      "loss": 0.5137,
+      "step": 4473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016833566946631225,
+      "loss": 0.541,
+      "step": 4474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016832228268235018,
+      "loss": 0.5154,
+      "step": 4475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016830889360174296,
+      "loss": 0.5448,
+      "step": 4476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016829550222494067,
+      "loss": 0.4527,
+      "step": 4477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016828210855239351,
+      "loss": 0.3307,
+      "step": 4478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016826871258455165,
+      "loss": 0.4838,
+      "step": 4479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016825531432186543,
+      "loss": 0.4816,
+      "step": 4480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016824191376478518,
+      "loss": 0.3683,
+      "step": 4481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016822851091376141,
+      "loss": 0.3586,
+      "step": 4482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016821510576924464,
+      "loss": 0.253,
+      "step": 4483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016820169833168546,
+      "loss": 0.5931,
+      "step": 4484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016818828860153462,
+      "loss": 0.4383,
+      "step": 4485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001681748765792428,
+      "loss": 0.3093,
+      "step": 4486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016816146226526093,
+      "loss": 0.5829,
+      "step": 4487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016814804566003987,
+      "loss": 0.409,
+      "step": 4488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001681346267640306,
+      "loss": 0.4573,
+      "step": 4489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016812120557768427,
+      "loss": 0.4754,
+      "step": 4490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000168107782101452,
+      "loss": 0.5025,
+      "step": 4491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000168094356335785,
+      "loss": 0.4191,
+      "step": 4492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016808092828113458,
+      "loss": 0.2928,
+      "step": 4493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016806749793795214,
+      "loss": 0.5908,
+      "step": 4494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001680540653066891,
+      "loss": 0.4257,
+      "step": 4495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016804063038779705,
+      "loss": 0.5983,
+      "step": 4496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016802719318172757,
+      "loss": 0.3843,
+      "step": 4497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016801375368893237,
+      "loss": 0.4748,
+      "step": 4498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016800031190986322,
+      "loss": 0.379,
+      "step": 4499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001679868678449719,
+      "loss": 0.3403,
+      "step": 4500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001679734214947104,
+      "loss": 0.4231,
+      "step": 4501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001679599728595307,
+      "loss": 0.3898,
+      "step": 4502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016794652193988485,
+      "loss": 0.3018,
+      "step": 4503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016793306873622507,
+      "loss": 0.441,
+      "step": 4504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001679196132490035,
+      "loss": 0.5501,
+      "step": 4505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001679061554786725,
+      "loss": 0.2906,
+      "step": 4506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016789269542568443,
+      "loss": 0.6712,
+      "step": 4507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016787923309049174,
+      "loss": 0.2871,
+      "step": 4508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000167865768473547,
+      "loss": 0.5529,
+      "step": 4509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016785230157530278,
+      "loss": 0.5537,
+      "step": 4510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016783883239621181,
+      "loss": 0.6263,
+      "step": 4511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001678253609367268,
+      "loss": 0.6445,
+      "step": 4512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016781188719730062,
+      "loss": 0.427,
+      "step": 4513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016779841117838624,
+      "loss": 0.6862,
+      "step": 4514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001677849328804366,
+      "loss": 0.7025,
+      "step": 4515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016777145230390474,
+      "loss": 0.774,
+      "step": 4516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016775796944924388,
+      "loss": 0.8186,
+      "step": 4517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001677444843169072,
+      "loss": 0.6576,
+      "step": 4518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016773099690734798,
+      "loss": 0.6053,
+      "step": 4519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016771750722101967,
+      "loss": 0.7344,
+      "step": 4520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016770401525837566,
+      "loss": 0.5732,
+      "step": 4521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016769052101986952,
+      "loss": 0.553,
+      "step": 4522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016767702450595483,
+      "loss": 0.4307,
+      "step": 4523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016766352571708529,
+      "loss": 0.6179,
+      "step": 4524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016765002465371465,
+      "loss": 0.4358,
+      "step": 4525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016763652131629675,
+      "loss": 0.5695,
+      "step": 4526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001676230157052855,
+      "loss": 0.3777,
+      "step": 4527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001676095078211349,
+      "loss": 0.4108,
+      "step": 4528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016759599766429903,
+      "loss": 0.3873,
+      "step": 4529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016758248523523198,
+      "loss": 0.5168,
+      "step": 4530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016756897053438804,
+      "loss": 0.1831,
+      "step": 4531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016755545356222142,
+      "loss": 0.5954,
+      "step": 4532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016754193431918655,
+      "loss": 0.4543,
+      "step": 4533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016752841280573787,
+      "loss": 0.5334,
+      "step": 4534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001675148890223299,
+      "loss": 0.4292,
+      "step": 4535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016750136296941724,
+      "loss": 0.5805,
+      "step": 4536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016748783464745458,
+      "loss": 0.5675,
+      "step": 4537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016747430405689663,
+      "loss": 0.3869,
+      "step": 4538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016746077119819826,
+      "loss": 0.3954,
+      "step": 4539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016744723607181437,
+      "loss": 0.3548,
+      "step": 4540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001674336986781999,
+      "loss": 0.4646,
+      "step": 4541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016742015901780998,
+      "loss": 0.3445,
+      "step": 4542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016740661709109973,
+      "loss": 0.3556,
+      "step": 4543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001673930728985243,
+      "loss": 0.5664,
+      "step": 4544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000167379526440539,
+      "loss": 0.3132,
+      "step": 4545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016736597771759923,
+      "loss": 0.5048,
+      "step": 4546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001673524267301604,
+      "loss": 0.3328,
+      "step": 4547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000167338873478678,
+      "loss": 0.41,
+      "step": 4548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016732531796360772,
+      "loss": 0.4781,
+      "step": 4549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001673117601854051,
+      "loss": 0.4372,
+      "step": 4550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016729820014452592,
+      "loss": 0.4042,
+      "step": 4551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016728463784142607,
+      "loss": 0.3484,
+      "step": 4552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016727107327656136,
+      "loss": 0.4203,
+      "step": 4553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001672575064503878,
+      "loss": 0.6414,
+      "step": 4554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016724393736336146,
+      "loss": 0.607,
+      "step": 4555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016723036601593843,
+      "loss": 0.3887,
+      "step": 4556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016721679240857488,
+      "loss": 0.4977,
+      "step": 4557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016720321654172717,
+      "loss": 0.2818,
+      "step": 4558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016718963841585156,
+      "loss": 0.4517,
+      "step": 4559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016717605803140452,
+      "loss": 0.4748,
+      "step": 4560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016716247538884255,
+      "loss": 0.5123,
+      "step": 4561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016714889048862229,
+      "loss": 0.7183,
+      "step": 4562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016713530333120028,
+      "loss": 0.281,
+      "step": 4563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001671217139170333,
+      "loss": 0.4064,
+      "step": 4564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016710812224657816,
+      "loss": 0.4212,
+      "step": 4565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016709452832029177,
+      "loss": 0.5476,
+      "step": 4566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016708093213863102,
+      "loss": 0.7132,
+      "step": 4567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016706733370205304,
+      "loss": 0.3694,
+      "step": 4568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016705373301101488,
+      "loss": 0.2863,
+      "step": 4569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001670401300659737,
+      "loss": 0.3732,
+      "step": 4570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001670265248673868,
+      "loss": 0.4865,
+      "step": 4571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016701291741571153,
+      "loss": 0.3538,
+      "step": 4572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016699930771140527,
+      "loss": 0.3559,
+      "step": 4573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016698569575492554,
+      "loss": 0.764,
+      "step": 4574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016697208154672986,
+      "loss": 0.3831,
+      "step": 4575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016695846508727588,
+      "loss": 0.5337,
+      "step": 4576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016694484637702134,
+      "loss": 0.3925,
+      "step": 4577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016693122541642404,
+      "loss": 0.4906,
+      "step": 4578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016691760220594184,
+      "loss": 0.3425,
+      "step": 4579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016690397674603267,
+      "loss": 0.3244,
+      "step": 4580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016689034903715454,
+      "loss": 0.39,
+      "step": 4581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016687671907976553,
+      "loss": 0.5644,
+      "step": 4582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016686308687432385,
+      "loss": 0.6016,
+      "step": 4583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016684945242128774,
+      "loss": 0.8977,
+      "step": 4584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016683581572111552,
+      "loss": 0.4773,
+      "step": 4585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016682217677426554,
+      "loss": 0.272,
+      "step": 4586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016680853558119632,
+      "loss": 0.496,
+      "step": 4587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016679489214236641,
+      "loss": 0.6668,
+      "step": 4588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001667812464582344,
+      "loss": 0.4122,
+      "step": 4589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000166767598529259,
+      "loss": 0.8311,
+      "step": 4590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016675394835589897,
+      "loss": 0.5241,
+      "step": 4591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001667402959386132,
+      "loss": 0.3202,
+      "step": 4592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016672664127786058,
+      "loss": 0.2361,
+      "step": 4593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016671298437410015,
+      "loss": 0.3622,
+      "step": 4594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016669932522779093,
+      "loss": 0.2585,
+      "step": 4595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016668566383939211,
+      "loss": 0.5145,
+      "step": 4596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016667200020936286,
+      "loss": 0.3235,
+      "step": 4597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016665833433816255,
+      "loss": 0.2751,
+      "step": 4598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016664466622625052,
+      "loss": 0.9193,
+      "step": 4599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016663099587408623,
+      "loss": 0.6608,
+      "step": 4600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016661732328212924,
+      "loss": 0.3625,
+      "step": 4601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016660364845083912,
+      "loss": 0.2599,
+      "step": 4602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016658997138067552,
+      "loss": 0.5616,
+      "step": 4603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001665762920720982,
+      "loss": 0.5409,
+      "step": 4604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016656261052556704,
+      "loss": 0.5538,
+      "step": 4605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001665489267415419,
+      "loss": 0.4901,
+      "step": 4606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016653524072048284,
+      "loss": 0.7893,
+      "step": 4607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001665215524628498,
+      "loss": 0.4056,
+      "step": 4608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016650786196910293,
+      "loss": 0.2933,
+      "step": 4609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001664941692397025,
+      "loss": 0.7838,
+      "step": 4610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001664804742751087,
+      "loss": 0.5839,
+      "step": 4611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016646677707578196,
+      "loss": 0.5052,
+      "step": 4612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016645307764218267,
+      "loss": 0.4013,
+      "step": 4613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016643937597477136,
+      "loss": 0.3608,
+      "step": 4614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016642567207400863,
+      "loss": 0.3665,
+      "step": 4615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016641196594035506,
+      "loss": 0.3197,
+      "step": 4616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016639825757427143,
+      "loss": 0.7503,
+      "step": 4617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016638454697621855,
+      "loss": 0.4092,
+      "step": 4618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016637083414665726,
+      "loss": 0.8758,
+      "step": 4619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001663571190860486,
+      "loss": 0.4732,
+      "step": 4620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001663434017948535,
+      "loss": 0.4755,
+      "step": 4621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001663296822735331,
+      "loss": 0.2951,
+      "step": 4622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016631596052254862,
+      "loss": 0.8377,
+      "step": 4623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016630223654236123,
+      "loss": 0.4698,
+      "step": 4624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016628851033343236,
+      "loss": 0.3919,
+      "step": 4625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016627478189622336,
+      "loss": 0.3853,
+      "step": 4626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001662610512311957,
+      "loss": 0.3687,
+      "step": 4627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016624731833881097,
+      "loss": 0.338,
+      "step": 4628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016623358321953078,
+      "loss": 0.4867,
+      "step": 4629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016621984587381682,
+      "loss": 0.5045,
+      "step": 4630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001662061063021309,
+      "loss": 0.8256,
+      "step": 4631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016619236450493485,
+      "loss": 0.7719,
+      "step": 4632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016617862048269065,
+      "loss": 0.5576,
+      "step": 4633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016616487423586022,
+      "loss": 0.5879,
+      "step": 4634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016615112576490567,
+      "loss": 0.3816,
+      "step": 4635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016613737507028916,
+      "loss": 0.4566,
+      "step": 4636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016612362215247294,
+      "loss": 0.3948,
+      "step": 4637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001661098670119193,
+      "loss": 0.5093,
+      "step": 4638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016609610964909063,
+      "loss": 0.4661,
+      "step": 4639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001660823500644493,
+      "loss": 0.3725,
+      "step": 4640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016606858825845796,
+      "loss": 0.2942,
+      "step": 4641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016605482423157915,
+      "loss": 0.3811,
+      "step": 4642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016604105798427552,
+      "loss": 0.4877,
+      "step": 4643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016602728951700985,
+      "loss": 0.3913,
+      "step": 4644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016601351883024497,
+      "loss": 0.5754,
+      "step": 4645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016599974592444377,
+      "loss": 0.4449,
+      "step": 4646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016598597080006924,
+      "loss": 0.3766,
+      "step": 4647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001659721934575844,
+      "loss": 0.668,
+      "step": 4648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001659584138974524,
+      "loss": 0.2578,
+      "step": 4649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016594463212013644,
+      "loss": 0.4044,
+      "step": 4650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016593084812609974,
+      "loss": 0.4577,
+      "step": 4651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016591706191580572,
+      "loss": 0.443,
+      "step": 4652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016590327348971776,
+      "loss": 0.6486,
+      "step": 4653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016588948284829939,
+      "loss": 0.8328,
+      "step": 4654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016587568999201413,
+      "loss": 0.3781,
+      "step": 4655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016586189492132566,
+      "loss": 0.3671,
+      "step": 4656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001658480976366977,
+      "loss": 0.3673,
+      "step": 4657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016583429813859404,
+      "loss": 0.4467,
+      "step": 4658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016582049642747853,
+      "loss": 0.4061,
+      "step": 4659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016580669250381512,
+      "loss": 0.3941,
+      "step": 4660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016579288636806787,
+      "loss": 0.4866,
+      "step": 4661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016577907802070083,
+      "loss": 0.556,
+      "step": 4662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016576526746217812,
+      "loss": 0.327,
+      "step": 4663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016575145469296409,
+      "loss": 0.3142,
+      "step": 4664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000165737639713523,
+      "loss": 0.6599,
+      "step": 4665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001657238225243192,
+      "loss": 0.2892,
+      "step": 4666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001657100031258172,
+      "loss": 0.2506,
+      "step": 4667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016569618151848151,
+      "loss": 0.3527,
+      "step": 4668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001656823577027768,
+      "loss": 0.2815,
+      "step": 4669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001656685316791677,
+      "loss": 0.4874,
+      "step": 4670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016565470344811895,
+      "loss": 0.3235,
+      "step": 4671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016564087301009544,
+      "loss": 0.3513,
+      "step": 4672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016562704036556208,
+      "loss": 0.3399,
+      "step": 4673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001656132055149838,
+      "loss": 0.5955,
+      "step": 4674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016559936845882568,
+      "loss": 0.4352,
+      "step": 4675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001655855291975529,
+      "loss": 0.6802,
+      "step": 4676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001655716877316306,
+      "loss": 0.5597,
+      "step": 4677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016555784406152406,
+      "loss": 0.3507,
+      "step": 4678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001655439981876987,
+      "loss": 0.5793,
+      "step": 4679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016553015011061984,
+      "loss": 0.4175,
+      "step": 4680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016551629983075307,
+      "loss": 0.5686,
+      "step": 4681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016550244734856397,
+      "loss": 0.4781,
+      "step": 4682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016548859266451811,
+      "loss": 0.6195,
+      "step": 4683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001654747357790813,
+      "loss": 0.591,
+      "step": 4684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001654608766927193,
+      "loss": 0.2825,
+      "step": 4685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016544701540589796,
+      "loss": 0.3609,
+      "step": 4686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016543315191908323,
+      "loss": 0.6153,
+      "step": 4687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016541928623274119,
+      "loss": 0.3571,
+      "step": 4688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016540541834733787,
+      "loss": 0.5638,
+      "step": 4689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016539154826333943,
+      "loss": 0.3793,
+      "step": 4690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001653776759812122,
+      "loss": 0.3541,
+      "step": 4691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016536380150142236,
+      "loss": 0.2519,
+      "step": 4692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001653499248244364,
+      "loss": 0.2767,
+      "step": 4693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016533604595072077,
+      "loss": 0.2168,
+      "step": 4694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016532216488074198,
+      "loss": 0.5493,
+      "step": 4695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016530828161496664,
+      "loss": 0.5792,
+      "step": 4696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016529439615386147,
+      "loss": 0.6088,
+      "step": 4697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001652805084978932,
+      "loss": 0.4178,
+      "step": 4698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016526661864752864,
+      "loss": 0.4618,
+      "step": 4699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016525272660323473,
+      "loss": 0.2943,
+      "step": 4700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016523883236547847,
+      "loss": 0.4058,
+      "step": 4701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016522493593472683,
+      "loss": 0.3179,
+      "step": 4702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016521103731144702,
+      "loss": 0.4038,
+      "step": 4703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016519713649610623,
+      "loss": 0.4144,
+      "step": 4704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001651832334891717,
+      "loss": 0.6491,
+      "step": 4705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016516932829111077,
+      "loss": 0.3525,
+      "step": 4706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001651554209023909,
+      "loss": 0.5877,
+      "step": 4707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001651415113234796,
+      "loss": 0.4642,
+      "step": 4708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016512759955484435,
+      "loss": 0.7647,
+      "step": 4709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001651136855969529,
+      "loss": 0.4895,
+      "step": 4710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001650997694502729,
+      "loss": 0.4042,
+      "step": 4711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016508585111527216,
+      "loss": 0.7353,
+      "step": 4712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016507193059241853,
+      "loss": 0.4143,
+      "step": 4713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016505800788217995,
+      "loss": 0.4512,
+      "step": 4714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016504408298502444,
+      "loss": 0.2943,
+      "step": 4715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001650301559014201,
+      "loss": 0.2923,
+      "step": 4716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016501622663183508,
+      "loss": 0.4658,
+      "step": 4717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016500229517673755,
+      "loss": 0.4932,
+      "step": 4718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016498836153659588,
+      "loss": 0.4858,
+      "step": 4719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016497442571187845,
+      "loss": 0.54,
+      "step": 4720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001649604877030537,
+      "loss": 0.4138,
+      "step": 4721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016494654751059008,
+      "loss": 0.4296,
+      "step": 4722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016493260513495632,
+      "loss": 0.4409,
+      "step": 4723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000164918660576621,
+      "loss": 0.4238,
+      "step": 4724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016490471383605288,
+      "loss": 0.5018,
+      "step": 4725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001648907649137208,
+      "loss": 0.3215,
+      "step": 4726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001648768138100936,
+      "loss": 0.7406,
+      "step": 4727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016486286052564032,
+      "loss": 0.3079,
+      "step": 4728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016484890506082996,
+      "loss": 0.516,
+      "step": 4729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016483494741613165,
+      "loss": 0.5329,
+      "step": 4730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001648209875920145,
+      "loss": 0.386,
+      "step": 4731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016480702558894787,
+      "loss": 0.312,
+      "step": 4732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016479306140740102,
+      "loss": 0.4277,
+      "step": 4733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001647790950478434,
+      "loss": 0.6862,
+      "step": 4734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016476512651074447,
+      "loss": 0.414,
+      "step": 4735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016475115579657376,
+      "loss": 0.5129,
+      "step": 4736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016473718290580092,
+      "loss": 0.5693,
+      "step": 4737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016472320783889565,
+      "loss": 0.6045,
+      "step": 4738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016470923059632768,
+      "loss": 0.5586,
+      "step": 4739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001646952511785669,
+      "loss": 0.3284,
+      "step": 4740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016468126958608318,
+      "loss": 0.2898,
+      "step": 4741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016466728581934658,
+      "loss": 0.308,
+      "step": 4742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001646532998788271,
+      "loss": 0.3968,
+      "step": 4743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016463931176499493,
+      "loss": 0.5335,
+      "step": 4744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001646253214783202,
+      "loss": 0.4883,
+      "step": 4745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016461132901927322,
+      "loss": 0.512,
+      "step": 4746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016459733438832443,
+      "loss": 0.3444,
+      "step": 4747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016458333758594414,
+      "loss": 0.5059,
+      "step": 4748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001645693386126029,
+      "loss": 0.4215,
+      "step": 4749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001645553374687713,
+      "loss": 0.3269,
+      "step": 4750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016454133415491997,
+      "loss": 0.4654,
+      "step": 4751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016452732867151964,
+      "loss": 0.5035,
+      "step": 4752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016451332101904108,
+      "loss": 0.4003,
+      "step": 4753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016449931119795515,
+      "loss": 0.4814,
+      "step": 4754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016448529920873282,
+      "loss": 0.7285,
+      "step": 4755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001644712850518451,
+      "loss": 0.6403,
+      "step": 4756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016445726872776307,
+      "loss": 0.2181,
+      "step": 4757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016444325023695785,
+      "loss": 0.4805,
+      "step": 4758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001644292295799007,
+      "loss": 0.418,
+      "step": 4759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016441520675706292,
+      "loss": 0.1738,
+      "step": 4760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001644011817689159,
+      "loss": 0.4439,
+      "step": 4761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001643871546159311,
+      "loss": 0.539,
+      "step": 4762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016437312529858,
+      "loss": 0.5421,
+      "step": 4763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016435909381733418,
+      "loss": 0.3707,
+      "step": 4764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016434506017266537,
+      "loss": 0.5464,
+      "step": 4765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016433102436504526,
+      "loss": 0.3499,
+      "step": 4766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001643169863949457,
+      "loss": 0.3417,
+      "step": 4767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016430294626283853,
+      "loss": 0.4279,
+      "step": 4768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001642889039691957,
+      "loss": 0.3172,
+      "step": 4769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016427485951448932,
+      "loss": 0.419,
+      "step": 4770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016426081289919143,
+      "loss": 0.5078,
+      "step": 4771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001642467641237742,
+      "loss": 0.4127,
+      "step": 4772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016423271318870993,
+      "loss": 0.6582,
+      "step": 4773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016421866009447088,
+      "loss": 0.3598,
+      "step": 4774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001642046048415295,
+      "loss": 0.2649,
+      "step": 4775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016419054743035816,
+      "loss": 0.3901,
+      "step": 4776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001641764878614295,
+      "loss": 0.6382,
+      "step": 4777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001641624261352161,
+      "loss": 0.7291,
+      "step": 4778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016414836225219065,
+      "loss": 0.4117,
+      "step": 4779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016413429621282589,
+      "loss": 0.4833,
+      "step": 4780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016412022801759463,
+      "loss": 0.4415,
+      "step": 4781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001641061576669698,
+      "loss": 0.5999,
+      "step": 4782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016409208516142437,
+      "loss": 0.4285,
+      "step": 4783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016407801050143133,
+      "loss": 0.6397,
+      "step": 4784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016406393368746392,
+      "loss": 0.4466,
+      "step": 4785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001640498547199952,
+      "loss": 0.3424,
+      "step": 4786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016403577359949854,
+      "loss": 0.3306,
+      "step": 4787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001640216903264472,
+      "loss": 0.2799,
+      "step": 4788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016400760490131462,
+      "loss": 0.6056,
+      "step": 4789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016399351732457426,
+      "loss": 0.5018,
+      "step": 4790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001639794275966997,
+      "loss": 0.2535,
+      "step": 4791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016396533571816452,
+      "loss": 0.3762,
+      "step": 4792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016395124168944248,
+      "loss": 0.3954,
+      "step": 4793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016393714551100734,
+      "loss": 0.6117,
+      "step": 4794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016392304718333285,
+      "loss": 0.5108,
+      "step": 4795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016390894670689302,
+      "loss": 0.5903,
+      "step": 4796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016389484408216182,
+      "loss": 0.5004,
+      "step": 4797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001638807393096133,
+      "loss": 0.4612,
+      "step": 4798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016386663238972154,
+      "loss": 0.3703,
+      "step": 4799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016385252332296084,
+      "loss": 0.4801,
+      "step": 4800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016383841210980543,
+      "loss": 0.5978,
+      "step": 4801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001638242987507296,
+      "loss": 0.5337,
+      "step": 4802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016381018324620783,
+      "loss": 0.4592,
+      "step": 4803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001637960655967146,
+      "loss": 0.4408,
+      "step": 4804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016378194580272447,
+      "loss": 0.2214,
+      "step": 4805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016376782386471207,
+      "loss": 0.3873,
+      "step": 4806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016375369978315215,
+      "loss": 0.4039,
+      "step": 4807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001637395735585194,
+      "loss": 0.327,
+      "step": 4808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016372544519128876,
+      "loss": 0.1976,
+      "step": 4809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001637113146819351,
+      "loss": 0.3201,
+      "step": 4810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016369718203093343,
+      "loss": 0.7264,
+      "step": 4811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016368304723875886,
+      "loss": 0.4979,
+      "step": 4812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016366891030588642,
+      "loss": 0.3538,
+      "step": 4813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016365477123279147,
+      "loss": 0.5016,
+      "step": 4814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016364063001994916,
+      "loss": 0.2577,
+      "step": 4815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016362648666783488,
+      "loss": 0.736,
+      "step": 4816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016361234117692413,
+      "loss": 0.3981,
+      "step": 4817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016359819354769232,
+      "loss": 0.4437,
+      "step": 4818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016358404378061508,
+      "loss": 0.3193,
+      "step": 4819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016356989187616802,
+      "loss": 0.4651,
+      "step": 4820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016355573783482686,
+      "loss": 0.5289,
+      "step": 4821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016354158165706739,
+      "loss": 0.8073,
+      "step": 4822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016352742334336547,
+      "loss": 0.352,
+      "step": 4823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000163513262894197,
+      "loss": 0.5254,
+      "step": 4824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016349910031003803,
+      "loss": 0.4346,
+      "step": 4825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016348493559136464,
+      "loss": 0.2845,
+      "step": 4826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016347076873865293,
+      "loss": 0.5973,
+      "step": 4827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016345659975237914,
+      "loss": 0.4386,
+      "step": 4828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001634424286330195,
+      "loss": 0.7901,
+      "step": 4829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016342825538105048,
+      "loss": 0.4353,
+      "step": 4830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016341407999694846,
+      "loss": 0.5089,
+      "step": 4831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001633999024811899,
+      "loss": 0.2544,
+      "step": 4832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016338572283425146,
+      "loss": 0.3229,
+      "step": 4833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016337154105660972,
+      "loss": 0.6391,
+      "step": 4834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001633573571487414,
+      "loss": 0.4415,
+      "step": 4835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016334317111112333,
+      "loss": 0.5095,
+      "step": 4836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016332898294423235,
+      "loss": 0.3564,
+      "step": 4837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001633147926485454,
+      "loss": 0.4602,
+      "step": 4838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001633006002245395,
+      "loss": 0.6474,
+      "step": 4839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001632864056726917,
+      "loss": 0.3593,
+      "step": 4840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016327220899347916,
+      "loss": 0.4528,
+      "step": 4841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016325801018737907,
+      "loss": 0.568,
+      "step": 4842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016324380925486875,
+      "loss": 0.5765,
+      "step": 4843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016322960619642558,
+      "loss": 0.2467,
+      "step": 4844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016321540101252698,
+      "loss": 0.7064,
+      "step": 4845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016320119370365042,
+      "loss": 0.415,
+      "step": 4846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016318698427027353,
+      "loss": 0.2118,
+      "step": 4847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001631727727128739,
+      "loss": 0.9657,
+      "step": 4848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016315855903192934,
+      "loss": 0.5527,
+      "step": 4849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016314434322791757,
+      "loss": 0.4324,
+      "step": 4850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016313012530131643,
+      "loss": 0.6407,
+      "step": 4851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016311590525260394,
+      "loss": 0.5227,
+      "step": 4852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016310168308225804,
+      "loss": 0.4865,
+      "step": 4853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016308745879075684,
+      "loss": 0.3639,
+      "step": 4854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016307323237857845,
+      "loss": 0.4287,
+      "step": 4855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001630590038462011,
+      "loss": 0.3523,
+      "step": 4856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016304477319410308,
+      "loss": 0.5972,
+      "step": 4857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001630305404227628,
+      "loss": 0.1573,
+      "step": 4858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016301630553265865,
+      "loss": 0.2953,
+      "step": 4859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016300206852426917,
+      "loss": 0.5575,
+      "step": 4860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016298782939807283,
+      "loss": 0.5314,
+      "step": 4861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001629735881545484,
+      "loss": 0.297,
+      "step": 4862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016295934479417453,
+      "loss": 0.3338,
+      "step": 4863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016294509931743005,
+      "loss": 0.7755,
+      "step": 4864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001629308517247938,
+      "loss": 0.2888,
+      "step": 4865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016291660201674468,
+      "loss": 0.3669,
+      "step": 4866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016290235019376175,
+      "loss": 0.2502,
+      "step": 4867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016288809625632401,
+      "loss": 0.4146,
+      "step": 4868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016287384020491068,
+      "loss": 0.2867,
+      "step": 4869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016285958204000095,
+      "loss": 0.3066,
+      "step": 4870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016284532176207407,
+      "loss": 0.2284,
+      "step": 4871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016283105937160946,
+      "loss": 0.2928,
+      "step": 4872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016281679486908652,
+      "loss": 0.937,
+      "step": 4873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001628025282549847,
+      "loss": 0.5379,
+      "step": 4874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001627882595297837,
+      "loss": 0.2733,
+      "step": 4875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000162773988693963,
+      "loss": 0.2208,
+      "step": 4876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001627597157480024,
+      "loss": 0.5054,
+      "step": 4877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001627454406923817,
+      "loss": 0.43,
+      "step": 4878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016273116352758074,
+      "loss": 0.4242,
+      "step": 4879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016271688425407941,
+      "loss": 0.3178,
+      "step": 4880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016270260287235772,
+      "loss": 0.7239,
+      "step": 4881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016268831938289578,
+      "loss": 0.3447,
+      "step": 4882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016267403378617364,
+      "loss": 0.4014,
+      "step": 4883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001626597460826716,
+      "loss": 0.5915,
+      "step": 4884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001626454562728699,
+      "loss": 0.5586,
+      "step": 4885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001626311643572489,
+      "loss": 0.3411,
+      "step": 4886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016261687033628902,
+      "loss": 0.3045,
+      "step": 4887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001626025742104707,
+      "loss": 0.5089,
+      "step": 4888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016258827598027454,
+      "loss": 0.5332,
+      "step": 4889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016257397564618125,
+      "loss": 0.626,
+      "step": 4890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016255967320867143,
+      "loss": 0.3121,
+      "step": 4891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001625453686682259,
+      "loss": 0.4699,
+      "step": 4892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016253106202532546,
+      "loss": 0.1643,
+      "step": 4893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001625167532804511,
+      "loss": 0.2397,
+      "step": 4894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016250244243408375,
+      "loss": 0.5545,
+      "step": 4895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016248812948670445,
+      "loss": 0.3045,
+      "step": 4896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001624738144387944,
+      "loss": 0.3766,
+      "step": 4897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016245949729083478,
+      "loss": 0.3441,
+      "step": 4898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016244517804330682,
+      "loss": 0.302,
+      "step": 4899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016243085669669186,
+      "loss": 0.512,
+      "step": 4900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016241653325147138,
+      "loss": 0.7952,
+      "step": 4901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016240220770812676,
+      "loss": 0.5681,
+      "step": 4902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001623878800671396,
+      "loss": 0.4084,
+      "step": 4903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016237355032899155,
+      "loss": 0.4435,
+      "step": 4904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001623592184941643,
+      "loss": 0.2967,
+      "step": 4905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016234488456313954,
+      "loss": 0.6133,
+      "step": 4906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001623305485363992,
+      "loss": 0.544,
+      "step": 4907
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016231621041442508,
+      "loss": 0.4029,
+      "step": 4908
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016230187019769928,
+      "loss": 0.4593,
+      "step": 4909
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016228752788670372,
+      "loss": 0.1935,
+      "step": 4910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016227318348192057,
+      "loss": 0.5252,
+      "step": 4911
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016225883698383205,
+      "loss": 0.4192,
+      "step": 4912
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016224448839292037,
+      "loss": 0.5177,
+      "step": 4913
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016223013770966786,
+      "loss": 0.3994,
+      "step": 4914
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016221578493455695,
+      "loss": 0.403,
+      "step": 4915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016220143006807005,
+      "loss": 0.2579,
+      "step": 4916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016218707311068973,
+      "loss": 0.3191,
+      "step": 4917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016217271406289863,
+      "loss": 0.3803,
+      "step": 4918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016215835292517934,
+      "loss": 0.2963,
+      "step": 4919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016214398969801469,
+      "loss": 0.2412,
+      "step": 4920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016212962438188745,
+      "loss": 0.5106,
+      "step": 4921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016211525697728055,
+      "loss": 0.6074,
+      "step": 4922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016210088748467696,
+      "loss": 0.6088,
+      "step": 4923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001620865159045596,
+      "loss": 0.2718,
+      "step": 4924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016207214223741168,
+      "loss": 0.4244,
+      "step": 4925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016205776648371635,
+      "loss": 0.4706,
+      "step": 4926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016204338864395684,
+      "loss": 0.5544,
+      "step": 4927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016202900871861638,
+      "loss": 0.2661,
+      "step": 4928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016201462670817847,
+      "loss": 0.4542,
+      "step": 4929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016200024261312654,
+      "loss": 0.2797,
+      "step": 4930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016198585643394404,
+      "loss": 0.3913,
+      "step": 4931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001619714681711146,
+      "loss": 0.2785,
+      "step": 4932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016195707782512187,
+      "loss": 0.2189,
+      "step": 4933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001619426853964496,
+      "loss": 0.3498,
+      "step": 4934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016192829088558158,
+      "loss": 0.3458,
+      "step": 4935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001619138942930017,
+      "loss": 0.3721,
+      "step": 4936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016189949561919387,
+      "loss": 0.6477,
+      "step": 4937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016188509486464206,
+      "loss": 0.7114,
+      "step": 4938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016187069202983046,
+      "loss": 0.4134,
+      "step": 4939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001618562871152431,
+      "loss": 0.572,
+      "step": 4940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016184188012136432,
+      "loss": 0.3046,
+      "step": 4941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001618274710486783,
+      "loss": 0.5256,
+      "step": 4942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016181305989766947,
+      "loss": 0.5422,
+      "step": 4943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001617986466688222,
+      "loss": 0.4267,
+      "step": 4944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016178423136262103,
+      "loss": 0.2912,
+      "step": 4945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016176981397955057,
+      "loss": 0.6476,
+      "step": 4946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016175539452009536,
+      "loss": 0.4512,
+      "step": 4947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016174097298474015,
+      "loss": 0.5566,
+      "step": 4948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001617265493739698,
+      "loss": 0.2642,
+      "step": 4949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000161712123688269,
+      "loss": 0.4048,
+      "step": 4950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001616976959281228,
+      "loss": 0.4415,
+      "step": 4951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016168326609401613,
+      "loss": 0.3333,
+      "step": 4952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016166883418643402,
+      "loss": 0.4197,
+      "step": 4953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016165440020586168,
+      "loss": 0.4299,
+      "step": 4954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016163996415278424,
+      "loss": 0.3387,
+      "step": 4955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016162552602768702,
+      "loss": 0.2819,
+      "step": 4956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016161108583105527,
+      "loss": 0.3181,
+      "step": 4957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016159664356337446,
+      "loss": 0.3744,
+      "step": 4958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016158219922513006,
+      "loss": 0.2799,
+      "step": 4959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016156775281680764,
+      "loss": 0.6849,
+      "step": 4960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016155330433889273,
+      "loss": 0.5743,
+      "step": 4961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016153885379187108,
+      "loss": 0.4829,
+      "step": 4962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016152440117622843,
+      "loss": 0.37,
+      "step": 4963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016150994649245058,
+      "loss": 0.3327,
+      "step": 4964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016149548974102348,
+      "loss": 0.1542,
+      "step": 4965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016148103092243307,
+      "loss": 0.7714,
+      "step": 4966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001614665700371653,
+      "loss": 0.6266,
+      "step": 4967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001614521070857064,
+      "loss": 0.3421,
+      "step": 4968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016143764206854244,
+      "loss": 0.4522,
+      "step": 4969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016142317498615973,
+      "loss": 0.2445,
+      "step": 4970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016140870583904453,
+      "loss": 0.305,
+      "step": 4971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016139423462768323,
+      "loss": 0.6615,
+      "step": 4972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016137976135256228,
+      "loss": 0.3091,
+      "step": 4973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001613652860141682,
+      "loss": 0.4326,
+      "step": 4974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016135080861298762,
+      "loss": 0.3052,
+      "step": 4975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016133632914950712,
+      "loss": 0.318,
+      "step": 4976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016132184762421346,
+      "loss": 0.3949,
+      "step": 4977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001613073640375934,
+      "loss": 0.4898,
+      "step": 4978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001612928783901339,
+      "loss": 0.4888,
+      "step": 4979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001612783906823218,
+      "loss": 0.3711,
+      "step": 4980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016126390091464413,
+      "loss": 0.629,
+      "step": 4981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000161249409087588,
+      "loss": 0.3417,
+      "step": 4982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016123491520164047,
+      "loss": 0.4477,
+      "step": 4983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001612204192572888,
+      "loss": 0.4519,
+      "step": 4984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001612059212550203,
+      "loss": 0.376,
+      "step": 4985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016119142119532225,
+      "loss": 0.3819,
+      "step": 4986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016117691907868213,
+      "loss": 0.3643,
+      "step": 4987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016116241490558738,
+      "loss": 0.4815,
+      "step": 4988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001611479086765256,
+      "loss": 0.4692,
+      "step": 4989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016113340039198437,
+      "loss": 0.4406,
+      "step": 4990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001611188900524514,
+      "loss": 0.3951,
+      "step": 4991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016110437765841446,
+      "loss": 0.2055,
+      "step": 4992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001610898632103614,
+      "loss": 0.3941,
+      "step": 4993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001610753467087801,
+      "loss": 0.2899,
+      "step": 4994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001610608281541585,
+      "loss": 0.3094,
+      "step": 4995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001610463075469847,
+      "loss": 0.4985,
+      "step": 4996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016103178488774677,
+      "loss": 0.2204,
+      "step": 4997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001610172601769329,
+      "loss": 0.3805,
+      "step": 4998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001610027334150313,
+      "loss": 0.4286,
+      "step": 4999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016098820460253038,
+      "loss": 0.5966,
+      "step": 5000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016097367373991842,
+      "loss": 0.5014,
+      "step": 5001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016095914082768392,
+      "loss": 0.2774,
+      "step": 5002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016094460586631542,
+      "loss": 0.4499,
+      "step": 5003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016093006885630149,
+      "loss": 0.3358,
+      "step": 5004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016091552979813077,
+      "loss": 0.4966,
+      "step": 5005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000160900988692292,
+      "loss": 0.4328,
+      "step": 5006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016088644553927402,
+      "loss": 0.2858,
+      "step": 5007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001608719003395656,
+      "loss": 0.4242,
+      "step": 5008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001608573530936558,
+      "loss": 0.402,
+      "step": 5009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016084280380203355,
+      "loss": 0.4007,
+      "step": 5010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001608282524651879,
+      "loss": 0.5957,
+      "step": 5011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016081369908360803,
+      "loss": 0.2502,
+      "step": 5012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016079914365778314,
+      "loss": 0.4582,
+      "step": 5013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016078458618820255,
+      "loss": 0.185,
+      "step": 5014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016077002667535558,
+      "loss": 0.2509,
+      "step": 5015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016075546511973158,
+      "loss": 0.7247,
+      "step": 5016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016074090152182012,
+      "loss": 0.3327,
+      "step": 5017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016072633588211073,
+      "loss": 0.4383,
+      "step": 5018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016071176820109303,
+      "loss": 0.4358,
+      "step": 5019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001606971984792567,
+      "loss": 0.3069,
+      "step": 5020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016068262671709152,
+      "loss": 0.2965,
+      "step": 5021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016066805291508728,
+      "loss": 0.4041,
+      "step": 5022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016065347707373392,
+      "loss": 0.3503,
+      "step": 5023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001606388991935214,
+      "loss": 0.4306,
+      "step": 5024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016062431927493974,
+      "loss": 0.3445,
+      "step": 5025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016060973731847904,
+      "loss": 0.4345,
+      "step": 5026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016059515332462944,
+      "loss": 0.3726,
+      "step": 5027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016058056729388123,
+      "loss": 0.6297,
+      "step": 5028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016056597922672474,
+      "loss": 0.3155,
+      "step": 5029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016055138912365025,
+      "loss": 0.2155,
+      "step": 5030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001605367969851483,
+      "loss": 0.8241,
+      "step": 5031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016052220281170937,
+      "loss": 0.4181,
+      "step": 5032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016050760660382403,
+      "loss": 0.3689,
+      "step": 5033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016049300836198292,
+      "loss": 0.5454,
+      "step": 5034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001604784080866768,
+      "loss": 0.3962,
+      "step": 5035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016046380577839638,
+      "loss": 0.3566,
+      "step": 5036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016044920143763261,
+      "loss": 0.4139,
+      "step": 5037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016043459506487635,
+      "loss": 0.2101,
+      "step": 5038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016041998666061863,
+      "loss": 0.3404,
+      "step": 5039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016040537622535045,
+      "loss": 0.3836,
+      "step": 5040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000160390763759563,
+      "loss": 0.5151,
+      "step": 5041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016037614926374746,
+      "loss": 0.394,
+      "step": 5042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016036153273839507,
+      "loss": 0.4257,
+      "step": 5043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016034691418399718,
+      "loss": 0.2846,
+      "step": 5044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016033229360104523,
+      "loss": 0.8102,
+      "step": 5045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016031767099003064,
+      "loss": 0.1876,
+      "step": 5046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016030304635144494,
+      "loss": 0.1702,
+      "step": 5047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016028841968577975,
+      "loss": 0.548,
+      "step": 5048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016027379099352676,
+      "loss": 0.5472,
+      "step": 5049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016025916027517768,
+      "loss": 0.2539,
+      "step": 5050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016024452753122433,
+      "loss": 0.4278,
+      "step": 5051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016022989276215865,
+      "loss": 0.5745,
+      "step": 5052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016021525596847248,
+      "loss": 0.3386,
+      "step": 5053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001602006171506579,
+      "loss": 0.2737,
+      "step": 5054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016018597630920698,
+      "loss": 0.5886,
+      "step": 5055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016017133344461187,
+      "loss": 0.4514,
+      "step": 5056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016015668855736477,
+      "loss": 0.5846,
+      "step": 5057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016014204164795803,
+      "loss": 0.4954,
+      "step": 5058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016012739271688395,
+      "loss": 0.3333,
+      "step": 5059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016011274176463492,
+      "loss": 0.4771,
+      "step": 5060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016009808879170347,
+      "loss": 0.3259,
+      "step": 5061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016008343379858217,
+      "loss": 0.37,
+      "step": 5062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016006877678576364,
+      "loss": 0.2901,
+      "step": 5063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016005411775374054,
+      "loss": 0.6371,
+      "step": 5064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001600394567030057,
+      "loss": 0.7345,
+      "step": 5065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001600247936340519,
+      "loss": 0.639,
+      "step": 5066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000160010128547372,
+      "loss": 0.4811,
+      "step": 5067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015999546144345906,
+      "loss": 0.1938,
+      "step": 5068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015998079232280602,
+      "loss": 0.5075,
+      "step": 5069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015996612118590603,
+      "loss": 0.2214,
+      "step": 5070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001599514480332523,
+      "loss": 0.5589,
+      "step": 5071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015993677286533798,
+      "loss": 0.3493,
+      "step": 5072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001599220956826564,
+      "loss": 0.4696,
+      "step": 5073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015990741648570095,
+      "loss": 0.3838,
+      "step": 5074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015989273527496507,
+      "loss": 0.5071,
+      "step": 5075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015987805205094227,
+      "loss": 0.476,
+      "step": 5076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001598633668141261,
+      "loss": 0.4721,
+      "step": 5077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015984867956501023,
+      "loss": 0.3187,
+      "step": 5078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015983399030408834,
+      "loss": 0.4344,
+      "step": 5079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015981929903185425,
+      "loss": 0.5472,
+      "step": 5080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015980460574880175,
+      "loss": 0.389,
+      "step": 5081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001597899104554248,
+      "loss": 0.2774,
+      "step": 5082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015977521315221736,
+      "loss": 0.4783,
+      "step": 5083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015976051383967345,
+      "loss": 0.4237,
+      "step": 5084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015974581251828725,
+      "loss": 0.3092,
+      "step": 5085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015973110918855295,
+      "loss": 0.4857,
+      "step": 5086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001597164038509647,
+      "loss": 0.402,
+      "step": 5087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015970169650601686,
+      "loss": 0.2483,
+      "step": 5088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001596869871542039,
+      "loss": 0.8146,
+      "step": 5089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015967227579602016,
+      "loss": 0.4542,
+      "step": 5090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015965756243196025,
+      "loss": 0.4088,
+      "step": 5091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001596428470625187,
+      "loss": 0.5548,
+      "step": 5092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015962812968819016,
+      "loss": 0.5118,
+      "step": 5093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001596134103094694,
+      "loss": 0.3249,
+      "step": 5094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015959868892685114,
+      "loss": 0.735,
+      "step": 5095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015958396554083032,
+      "loss": 0.536,
+      "step": 5096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015956924015190184,
+      "loss": 0.3405,
+      "step": 5097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015955451276056067,
+      "loss": 0.5324,
+      "step": 5098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015953978336730184,
+      "loss": 0.3175,
+      "step": 5099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015952505197262056,
+      "loss": 0.3844,
+      "step": 5100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015951031857701193,
+      "loss": 0.5095,
+      "step": 5101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001594955831809713,
+      "loss": 0.2708,
+      "step": 5102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015948084578499395,
+      "loss": 0.304,
+      "step": 5103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001594661063895753,
+      "loss": 0.2932,
+      "step": 5104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015945136499521077,
+      "loss": 0.5032,
+      "step": 5105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001594366216023959,
+      "loss": 0.4064,
+      "step": 5106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015942187621162637,
+      "loss": 0.3401,
+      "step": 5107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015940712882339772,
+      "loss": 0.3648,
+      "step": 5108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015939237943820577,
+      "loss": 0.23,
+      "step": 5109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001593776280565463,
+      "loss": 0.4952,
+      "step": 5110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015936287467891512,
+      "loss": 0.5487,
+      "step": 5111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015934811930580825,
+      "loss": 0.3961,
+      "step": 5112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015933336193772163,
+      "loss": 0.4329,
+      "step": 5113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015931860257515138,
+      "loss": 0.4196,
+      "step": 5114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00015930384121859355,
+      "loss": 0.5827,
+      "step": 5115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001592890778685444,
+      "loss": 0.2365,
+      "step": 5116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001592743125255002,
+      "loss": 0.1639,
+      "step": 5117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015925954518995728,
+      "loss": 0.2813,
+      "step": 5118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015924477586241203,
+      "loss": 0.3179,
+      "step": 5119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015923000454336092,
+      "loss": 0.3929,
+      "step": 5120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015921523123330048,
+      "loss": 0.2432,
+      "step": 5121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015920045593272736,
+      "loss": 0.1885,
+      "step": 5122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001591856786421382,
+      "loss": 0.4164,
+      "step": 5123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001591708993620297,
+      "loss": 0.3915,
+      "step": 5124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001591561180928987,
+      "loss": 0.3838,
+      "step": 5125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001591413348352421,
+      "loss": 0.2354,
+      "step": 5126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015912654958955675,
+      "loss": 0.2948,
+      "step": 5127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001591117623563397,
+      "loss": 0.5091,
+      "step": 5128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015909697313608808,
+      "loss": 0.2612,
+      "step": 5129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015908218192929897,
+      "loss": 0.3213,
+      "step": 5130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015906738873646957,
+      "loss": 0.485,
+      "step": 5131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015905259355809717,
+      "loss": 0.4248,
+      "step": 5132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015903779639467907,
+      "loss": 0.5807,
+      "step": 5133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015902299724671274,
+      "loss": 0.7883,
+      "step": 5134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001590081961146956,
+      "loss": 0.5269,
+      "step": 5135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001589933929991252,
+      "loss": 0.4385,
+      "step": 5136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015897858790049917,
+      "loss": 0.5564,
+      "step": 5137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015896378081931513,
+      "loss": 0.2749,
+      "step": 5138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015894897175607086,
+      "loss": 0.4569,
+      "step": 5139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015893416071126418,
+      "loss": 0.3004,
+      "step": 5140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001589193476853929,
+      "loss": 0.6057,
+      "step": 5141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000158904532678955,
+      "loss": 0.4918,
+      "step": 5142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001588897156924485,
+      "loss": 0.4822,
+      "step": 5143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015887489672637145,
+      "loss": 0.4521,
+      "step": 5144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015886007578122196,
+      "loss": 0.6897,
+      "step": 5145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015884525285749827,
+      "loss": 0.3872,
+      "step": 5146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015883042795569865,
+      "loss": 0.2845,
+      "step": 5147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015881560107632142,
+      "loss": 0.3137,
+      "step": 5148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015880077221986502,
+      "loss": 0.2569,
+      "step": 5149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015878594138682788,
+      "loss": 0.581,
+      "step": 5150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015877110857770855,
+      "loss": 0.3603,
+      "step": 5151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015875627379300565,
+      "loss": 0.3955,
+      "step": 5152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015874143703321778,
+      "loss": 0.3881,
+      "step": 5153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015872659829884377,
+      "loss": 0.4713,
+      "step": 5154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015871175759038235,
+      "loss": 0.5354,
+      "step": 5155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015869691490833247,
+      "loss": 0.3275,
+      "step": 5156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000158682070253193,
+      "loss": 0.5844,
+      "step": 5157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015866722362546293,
+      "loss": 0.5095,
+      "step": 5158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015865237502564136,
+      "loss": 0.4162,
+      "step": 5159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001586375244542274,
+      "loss": 0.4026,
+      "step": 5160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001586226719117203,
+      "loss": 0.3012,
+      "step": 5161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015860781739861928,
+      "loss": 0.352,
+      "step": 5162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015859296091542372,
+      "loss": 0.3056,
+      "step": 5163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015857810246263293,
+      "loss": 0.3586,
+      "step": 5164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015856324204074643,
+      "loss": 0.2676,
+      "step": 5165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015854837965026378,
+      "loss": 0.3401,
+      "step": 5166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015853351529168457,
+      "loss": 0.3873,
+      "step": 5167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015851864896550838,
+      "loss": 0.3523,
+      "step": 5168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015850378067223505,
+      "loss": 0.4708,
+      "step": 5169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015848891041236432,
+      "loss": 0.2164,
+      "step": 5170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015847403818639605,
+      "loss": 0.3546,
+      "step": 5171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00015845916399483018,
+      "loss": 0.4016,
+      "step": 5172
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015844428783816673,
+      "loss": 0.4525,
+      "step": 5173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015842940971690566,
+      "loss": 0.4204,
+      "step": 5174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015841452963154726,
+      "loss": 0.5637,
+      "step": 5175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015839964758259157,
+      "loss": 0.5207,
+      "step": 5176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015838476357053893,
+      "loss": 0.3963,
+      "step": 5177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015836987759588965,
+      "loss": 0.5823,
+      "step": 5178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015835498965914409,
+      "loss": 0.5224,
+      "step": 5179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015834009976080273,
+      "loss": 0.3497,
+      "step": 5180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001583252079013661,
+      "loss": 0.3517,
+      "step": 5181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001583103140813348,
+      "loss": 0.3357,
+      "step": 5182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015829541830120945,
+      "loss": 0.5248,
+      "step": 5183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015828052056149076,
+      "loss": 0.4215,
+      "step": 5184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015826562086267956,
+      "loss": 0.3559,
+      "step": 5185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015825071920527668,
+      "loss": 0.394,
+      "step": 5186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015823581558978305,
+      "loss": 0.3759,
+      "step": 5187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015822091001669962,
+      "loss": 0.1987,
+      "step": 5188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015820600248652746,
+      "loss": 0.5946,
+      "step": 5189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015819109299976768,
+      "loss": 0.382,
+      "step": 5190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001581761815569215,
+      "loss": 0.2936,
+      "step": 5191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001581612681584901,
+      "loss": 0.3873,
+      "step": 5192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015814635280497484,
+      "loss": 0.3912,
+      "step": 5193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015813143549687708,
+      "loss": 0.2728,
+      "step": 5194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015811651623469825,
+      "loss": 0.5065,
+      "step": 5195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015810159501893992,
+      "loss": 0.3342,
+      "step": 5196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015808667185010356,
+      "loss": 0.4627,
+      "step": 5197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001580717467286909,
+      "loss": 0.3833,
+      "step": 5198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001580568196552036,
+      "loss": 0.3044,
+      "step": 5199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015804189063014346,
+      "loss": 0.4444,
+      "step": 5200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001580269596540123,
+      "loss": 0.2862,
+      "step": 5201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015801202672731203,
+      "loss": 0.2823,
+      "step": 5202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015799709185054464,
+      "loss": 0.4143,
+      "step": 5203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001579821550242121,
+      "loss": 0.3671,
+      "step": 5204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015796721624881656,
+      "loss": 0.2891,
+      "step": 5205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001579522755248602,
+      "loss": 0.286,
+      "step": 5206
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001579373328528452,
+      "loss": 0.3767,
+      "step": 5207
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015792238823327388,
+      "loss": 0.4055,
+      "step": 5208
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001579074416666486,
+      "loss": 0.3397,
+      "step": 5209
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001578924931534718,
+      "loss": 0.4849,
+      "step": 5210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000157877542694246,
+      "loss": 0.1705,
+      "step": 5211
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015786259028947365,
+      "loss": 0.6817,
+      "step": 5212
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001578476359396575,
+      "loss": 0.3552,
+      "step": 5213
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015783267964530017,
+      "loss": 0.4055,
+      "step": 5214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001578177214069044,
+      "loss": 0.6074,
+      "step": 5215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015780276122497305,
+      "loss": 0.3541,
+      "step": 5216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000157787799100009,
+      "loss": 0.2419,
+      "step": 5217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001577728350325152,
+      "loss": 0.5085,
+      "step": 5218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015775786902299464,
+      "loss": 0.4914,
+      "step": 5219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001577429010719504,
+      "loss": 0.5954,
+      "step": 5220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015772793117988567,
+      "loss": 0.5109,
+      "step": 5221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015771295934730365,
+      "loss": 0.547,
+      "step": 5222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015769798557470753,
+      "loss": 0.4134,
+      "step": 5223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015768300986260077,
+      "loss": 0.3818,
+      "step": 5224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015766803221148673,
+      "loss": 0.2879,
+      "step": 5225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015765305262186886,
+      "loss": 0.5704,
+      "step": 5226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015763807109425075,
+      "loss": 0.2204,
+      "step": 5227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00015762308762913593,
+      "loss": 0.6628,
+      "step": 5228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001576081022270281,
+      "loss": 0.38,
+      "step": 5229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015759311488843105,
+      "loss": 0.3845,
+      "step": 5230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001575781256138485,
+      "loss": 0.4064,
+      "step": 5231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015756313440378432,
+      "loss": 0.4699,
+      "step": 5232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015754814125874247,
+      "loss": 0.5259,
+      "step": 5233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001575331461792269,
+      "loss": 0.531,
+      "step": 5234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015751814916574176,
+      "loss": 0.6211,
+      "step": 5235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015750315021879107,
+      "loss": 0.3145,
+      "step": 5236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015748814933887908,
+      "loss": 0.4139,
+      "step": 5237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015747314652651,
+      "loss": 0.4188,
+      "step": 5238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015745814178218817,
+      "loss": 0.1635,
+      "step": 5239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015744313510641797,
+      "loss": 0.3936,
+      "step": 5240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015742812649970387,
+      "loss": 0.3805,
+      "step": 5241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015741311596255032,
+      "loss": 0.3263,
+      "step": 5242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015739810349546196,
+      "loss": 0.3765,
+      "step": 5243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015738308909894343,
+      "loss": 0.6431,
+      "step": 5244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015736807277349937,
+      "loss": 0.2976,
+      "step": 5245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015735305451963468,
+      "loss": 0.5244,
+      "step": 5246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015733803433785404,
+      "loss": 0.3252,
+      "step": 5247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015732301222866246,
+      "loss": 0.376,
+      "step": 5248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015730798819256484,
+      "loss": 0.1685,
+      "step": 5249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015729296223006628,
+      "loss": 0.3245,
+      "step": 5250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015727793434167184,
+      "loss": 0.3273,
+      "step": 5251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015726290452788665,
+      "loss": 0.4205,
+      "step": 5252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015724787278921602,
+      "loss": 0.282,
+      "step": 5253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015723283912616513,
+      "loss": 0.4692,
+      "step": 5254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015721780353923943,
+      "loss": 0.4473,
+      "step": 5255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015720276602894427,
+      "loss": 0.301,
+      "step": 5256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001571877265957852,
+      "loss": 0.3306,
+      "step": 5257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001571726852402677,
+      "loss": 0.4263,
+      "step": 5258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015715764196289747,
+      "loss": 0.4982,
+      "step": 5259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015714259676418007,
+      "loss": 0.3065,
+      "step": 5260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015712754964462136,
+      "loss": 0.3014,
+      "step": 5261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015711250060472707,
+      "loss": 0.326,
+      "step": 5262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001570974496450031,
+      "loss": 0.3998,
+      "step": 5263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001570823967659554,
+      "loss": 0.2035,
+      "step": 5264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015706734196808993,
+      "loss": 0.4116,
+      "step": 5265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015705228525191278,
+      "loss": 0.5331,
+      "step": 5266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015703722661793005,
+      "loss": 0.4456,
+      "step": 5267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000157022166066648,
+      "loss": 0.3633,
+      "step": 5268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015700710359857286,
+      "loss": 0.4031,
+      "step": 5269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001569920392142109,
+      "loss": 0.3569,
+      "step": 5270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015697697291406859,
+      "loss": 0.3476,
+      "step": 5271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015696190469865234,
+      "loss": 0.2978,
+      "step": 5272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015694683456846862,
+      "loss": 0.4098,
+      "step": 5273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015693176252402414,
+      "loss": 0.1979,
+      "step": 5274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001569166885658254,
+      "loss": 0.4667,
+      "step": 5275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001569016126943792,
+      "loss": 0.2821,
+      "step": 5276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001568865349101923,
+      "loss": 0.4505,
+      "step": 5277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001568714552137715,
+      "loss": 0.5256,
+      "step": 5278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015685637360562376,
+      "loss": 0.27,
+      "step": 5279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015684129008625597,
+      "loss": 0.275,
+      "step": 5280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015682620465617525,
+      "loss": 0.3524,
+      "step": 5281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001568111173158886,
+      "loss": 0.4276,
+      "step": 5282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015679602806590326,
+      "loss": 0.5111,
+      "step": 5283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001567809369067264,
+      "loss": 0.6322,
+      "step": 5284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015676584383886535,
+      "loss": 0.3163,
+      "step": 5285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00015675074886282745,
+      "loss": 0.3067,
+      "step": 5286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001567356519791201,
+      "loss": 0.3183,
+      "step": 5287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015672055318825077,
+      "loss": 0.4283,
+      "step": 5288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015670545249072703,
+      "loss": 0.4266,
+      "step": 5289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001566903498870565,
+      "loss": 0.2491,
+      "step": 5290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001566752453777468,
+      "loss": 0.2747,
+      "step": 5291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001566601389633057,
+      "loss": 0.3948,
+      "step": 5292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015664503064424099,
+      "loss": 0.5334,
+      "step": 5293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015662992042106058,
+      "loss": 0.6233,
+      "step": 5294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001566148082942723,
+      "loss": 0.3624,
+      "step": 5295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015659969426438423,
+      "loss": 0.2868,
+      "step": 5296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015658457833190442,
+      "loss": 0.4343,
+      "step": 5297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015656946049734095,
+      "loss": 0.3171,
+      "step": 5298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015655434076120204,
+      "loss": 0.4856,
+      "step": 5299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015653921912399589,
+      "loss": 0.3304,
+      "step": 5300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015652409558623086,
+      "loss": 0.4594,
+      "step": 5301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001565089701484153,
+      "loss": 0.7382,
+      "step": 5302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001564938428110577,
+      "loss": 0.5805,
+      "step": 5303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001564787135746665,
+      "loss": 0.2271,
+      "step": 5304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015646358243975026,
+      "loss": 0.3211,
+      "step": 5305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015644844940681764,
+      "loss": 0.3455,
+      "step": 5306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015643331447637735,
+      "loss": 0.6192,
+      "step": 5307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015641817764893814,
+      "loss": 0.3684,
+      "step": 5308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015640303892500883,
+      "loss": 0.4133,
+      "step": 5309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015638789830509832,
+      "loss": 0.2884,
+      "step": 5310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015637275578971553,
+      "loss": 0.2997,
+      "step": 5311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015635761137936945,
+      "loss": 0.2333,
+      "step": 5312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015634246507456923,
+      "loss": 0.1288,
+      "step": 5313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015632731687582397,
+      "loss": 0.5183,
+      "step": 5314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015631216678364286,
+      "loss": 0.415,
+      "step": 5315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001562970147985352,
+      "loss": 0.2699,
+      "step": 5316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015628186092101035,
+      "loss": 0.1665,
+      "step": 5317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015626670515157763,
+      "loss": 0.2884,
+      "step": 5318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015625154749074655,
+      "loss": 0.4074,
+      "step": 5319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001562363879390266,
+      "loss": 0.3377,
+      "step": 5320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015622122649692742,
+      "loss": 0.3801,
+      "step": 5321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015620606316495858,
+      "loss": 0.4286,
+      "step": 5322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001561908979436299,
+      "loss": 0.5192,
+      "step": 5323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015617573083345106,
+      "loss": 0.4477,
+      "step": 5324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001561605618349319,
+      "loss": 0.4899,
+      "step": 5325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015614539094858243,
+      "loss": 0.2523,
+      "step": 5326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001561302181749125,
+      "loss": 0.2141,
+      "step": 5327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001561150435144322,
+      "loss": 0.4465,
+      "step": 5328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015609986696765162,
+      "loss": 0.3026,
+      "step": 5329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015608468853508094,
+      "loss": 0.4984,
+      "step": 5330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015606950821723032,
+      "loss": 0.8155,
+      "step": 5331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015605432601461006,
+      "loss": 0.3161,
+      "step": 5332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015603914192773054,
+      "loss": 0.3681,
+      "step": 5333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015602395595710217,
+      "loss": 0.3858,
+      "step": 5334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015600876810323538,
+      "loss": 0.4439,
+      "step": 5335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015599357836664082,
+      "loss": 0.2291,
+      "step": 5336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015597838674782892,
+      "loss": 0.6037,
+      "step": 5337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015596319324731047,
+      "loss": 0.2561,
+      "step": 5338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015594799786559615,
+      "loss": 0.4613,
+      "step": 5339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015593280060319676,
+      "loss": 0.4688,
+      "step": 5340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001559176014606232,
+      "loss": 0.3757,
+      "step": 5341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00015590240043838628,
+      "loss": 0.2994,
+      "step": 5342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001558871975369971,
+      "loss": 0.4249,
+      "step": 5343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001558719927569666,
+      "loss": 0.2164,
+      "step": 5344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015585678609880596,
+      "loss": 0.1341,
+      "step": 5345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015584157756302634,
+      "loss": 0.3105,
+      "step": 5346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015582636715013895,
+      "loss": 0.3763,
+      "step": 5347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015581115486065506,
+      "loss": 0.463,
+      "step": 5348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015579594069508608,
+      "loss": 0.572,
+      "step": 5349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015578072465394343,
+      "loss": 0.4482,
+      "step": 5350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015576550673773859,
+      "loss": 0.3632,
+      "step": 5351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001557502869469831,
+      "loss": 0.333,
+      "step": 5352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015573506528218854,
+      "loss": 0.4105,
+      "step": 5353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001557198417438666,
+      "loss": 0.2526,
+      "step": 5354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001557046163325291,
+      "loss": 0.1573,
+      "step": 5355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015568938904868772,
+      "loss": 0.3712,
+      "step": 5356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001556741598928544,
+      "loss": 0.289,
+      "step": 5357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015565892886554102,
+      "loss": 0.5078,
+      "step": 5358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015564369596725964,
+      "loss": 0.5963,
+      "step": 5359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015562846119852218,
+      "loss": 0.4742,
+      "step": 5360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001556132245598409,
+      "loss": 0.3814,
+      "step": 5361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015559798605172792,
+      "loss": 0.5572,
+      "step": 5362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001555827456746955,
+      "loss": 0.3866,
+      "step": 5363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015556750342925587,
+      "loss": 0.3655,
+      "step": 5364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015555225931592143,
+      "loss": 0.5787,
+      "step": 5365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015553701333520466,
+      "loss": 0.4147,
+      "step": 5366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000155521765487618,
+      "loss": 0.3387,
+      "step": 5367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015550651577367407,
+      "loss": 0.2716,
+      "step": 5368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015549126419388536,
+      "loss": 0.4475,
+      "step": 5369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015547601074876469,
+      "loss": 0.3583,
+      "step": 5370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015546075543882473,
+      "loss": 0.2549,
+      "step": 5371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015544549826457826,
+      "loss": 0.3589,
+      "step": 5372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001554302392265382,
+      "loss": 0.5145,
+      "step": 5373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001554149783252175,
+      "loss": 0.5058,
+      "step": 5374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001553997155611291,
+      "loss": 0.4233,
+      "step": 5375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015538445093478607,
+      "loss": 0.5082,
+      "step": 5376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015536918444670152,
+      "loss": 0.33,
+      "step": 5377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015535391609738865,
+      "loss": 0.3241,
+      "step": 5378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015533864588736072,
+      "loss": 0.4132,
+      "step": 5379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000155323373817131,
+      "loss": 0.1756,
+      "step": 5380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001553080998872129,
+      "loss": 0.3749,
+      "step": 5381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015529282409811979,
+      "loss": 0.3,
+      "step": 5382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001552775464503652,
+      "loss": 0.4513,
+      "step": 5383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001552622669444627,
+      "loss": 0.413,
+      "step": 5384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015524698558092587,
+      "loss": 0.5505,
+      "step": 5385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015523170236026842,
+      "loss": 0.5174,
+      "step": 5386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001552164172830041,
+      "loss": 0.4528,
+      "step": 5387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001552011303496467,
+      "loss": 0.2552,
+      "step": 5388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001551858415607101,
+      "loss": 0.491,
+      "step": 5389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015517055091670822,
+      "loss": 0.3843,
+      "step": 5390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015515525841815507,
+      "loss": 0.2913,
+      "step": 5391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015513996406556465,
+      "loss": 0.5126,
+      "step": 5392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015512466785945115,
+      "loss": 0.6548,
+      "step": 5393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015510936980032874,
+      "loss": 0.6231,
+      "step": 5394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015509406988871161,
+      "loss": 0.2689,
+      "step": 5395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015507876812511411,
+      "loss": 0.7284,
+      "step": 5396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015506346451005056,
+      "loss": 0.3288,
+      "step": 5397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015504815904403545,
+      "loss": 0.3248,
+      "step": 5398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015503285172758323,
+      "loss": 0.2818,
+      "step": 5399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001550175425612085,
+      "loss": 0.3953,
+      "step": 5400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00015500223154542582,
+      "loss": 0.3777,
+      "step": 5401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001549869186807499,
+      "loss": 0.2814,
+      "step": 5402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015497160396769547,
+      "loss": 0.6174,
+      "step": 5403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015495628740677732,
+      "loss": 0.6025,
+      "step": 5404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015494096899851031,
+      "loss": 0.4117,
+      "step": 5405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015492564874340943,
+      "loss": 0.3008,
+      "step": 5406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001549103266419896,
+      "loss": 0.392,
+      "step": 5407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001548950026947659,
+      "loss": 0.3701,
+      "step": 5408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015487967690225345,
+      "loss": 0.37,
+      "step": 5409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015486434926496737,
+      "loss": 0.2549,
+      "step": 5410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015484901978342295,
+      "loss": 0.5157,
+      "step": 5411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001548336884581355,
+      "loss": 0.3357,
+      "step": 5412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015481835528962035,
+      "loss": 0.4081,
+      "step": 5413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001548030202783929,
+      "loss": 0.2226,
+      "step": 5414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001547876834249687,
+      "loss": 0.2673,
+      "step": 5415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015477234472986324,
+      "loss": 0.6716,
+      "step": 5416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015475700419359217,
+      "loss": 0.2039,
+      "step": 5417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001547416618166711,
+      "loss": 0.4624,
+      "step": 5418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015472631759961582,
+      "loss": 0.2792,
+      "step": 5419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001547109715429421,
+      "loss": 0.1634,
+      "step": 5420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001546956236471658,
+      "loss": 0.2305,
+      "step": 5421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015468027391280287,
+      "loss": 0.2974,
+      "step": 5422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015466492234036922,
+      "loss": 0.3853,
+      "step": 5423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015464956893038095,
+      "loss": 0.369,
+      "step": 5424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015463421368335412,
+      "loss": 0.5667,
+      "step": 5425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015461885659980493,
+      "loss": 0.5016,
+      "step": 5426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001546034976802496,
+      "loss": 0.2828,
+      "step": 5427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001545881369252044,
+      "loss": 0.3416,
+      "step": 5428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001545727743351857,
+      "loss": 0.3925,
+      "step": 5429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015455740991070989,
+      "loss": 0.2959,
+      "step": 5430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015454204365229342,
+      "loss": 0.398,
+      "step": 5431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015452667556045292,
+      "loss": 0.3619,
+      "step": 5432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001545113056357049,
+      "loss": 0.3582,
+      "step": 5433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015449593387856608,
+      "loss": 0.3519,
+      "step": 5434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015448056028955311,
+      "loss": 0.7429,
+      "step": 5435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001544651848691828,
+      "loss": 0.4105,
+      "step": 5436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015444980761797204,
+      "loss": 0.4948,
+      "step": 5437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015443442853643762,
+      "loss": 0.3506,
+      "step": 5438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015441904762509667,
+      "loss": 0.592,
+      "step": 5439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015440366488446607,
+      "loss": 0.2968,
+      "step": 5440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015438828031506298,
+      "loss": 0.4426,
+      "step": 5441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015437289391740454,
+      "loss": 0.4954,
+      "step": 5442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015435750569200795,
+      "loss": 0.3153,
+      "step": 5443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001543421156393905,
+      "loss": 0.5503,
+      "step": 5444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001543267237600695,
+      "loss": 0.4649,
+      "step": 5445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001543113300545624,
+      "loss": 0.5492,
+      "step": 5446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015429593452338659,
+      "loss": 0.2573,
+      "step": 5447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001542805371670596,
+      "loss": 0.3754,
+      "step": 5448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015426513798609906,
+      "loss": 0.2582,
+      "step": 5449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015424973698102256,
+      "loss": 0.5413,
+      "step": 5450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015423433415234783,
+      "loss": 0.2966,
+      "step": 5451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015421892950059264,
+      "loss": 0.1778,
+      "step": 5452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001542035230262748,
+      "loss": 0.4318,
+      "step": 5453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015418811472991218,
+      "loss": 0.3647,
+      "step": 5454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015417270461202278,
+      "loss": 0.4108,
+      "step": 5455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015415729267312454,
+      "loss": 0.6468,
+      "step": 5456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015414187891373563,
+      "loss": 0.3236,
+      "step": 5457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015412646333437405,
+      "loss": 0.3509,
+      "step": 5458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001541110459355581,
+      "loss": 0.4271,
+      "step": 5459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015409562671780597,
+      "loss": 0.2963,
+      "step": 5460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015408020568163602,
+      "loss": 0.4315,
+      "step": 5461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001540647828275666,
+      "loss": 0.3234,
+      "step": 5462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015404935815611616,
+      "loss": 0.221,
+      "step": 5463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001540339316678032,
+      "loss": 0.4882,
+      "step": 5464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015401850336314626,
+      "loss": 0.4908,
+      "step": 5465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015400307324266396,
+      "loss": 0.3262,
+      "step": 5466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015398764130687502,
+      "loss": 0.3246,
+      "step": 5467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015397220755629815,
+      "loss": 0.3311,
+      "step": 5468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015395677199145214,
+      "loss": 0.3695,
+      "step": 5469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001539413346128559,
+      "loss": 0.3127,
+      "step": 5470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001539258954210283,
+      "loss": 0.3626,
+      "step": 5471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001539104544164884,
+      "loss": 0.5847,
+      "step": 5472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001538950115997552,
+      "loss": 0.4289,
+      "step": 5473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001538795669713478,
+      "loss": 0.295,
+      "step": 5474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015386412053178535,
+      "loss": 0.6591,
+      "step": 5475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015384867228158714,
+      "loss": 0.3133,
+      "step": 5476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015383322222127247,
+      "loss": 0.1984,
+      "step": 5477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015381777035136063,
+      "loss": 0.2015,
+      "step": 5478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015380231667237105,
+      "loss": 0.3355,
+      "step": 5479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001537868611848232,
+      "loss": 0.5374,
+      "step": 5480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015377140388923667,
+      "loss": 0.4122,
+      "step": 5481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015375594478613097,
+      "loss": 0.4213,
+      "step": 5482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001537404838760258,
+      "loss": 0.3006,
+      "step": 5483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001537250211594409,
+      "loss": 0.2548,
+      "step": 5484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000153709556636896,
+      "loss": 0.3098,
+      "step": 5485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015369409030891097,
+      "loss": 0.3101,
+      "step": 5486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015367862217600568,
+      "loss": 0.2884,
+      "step": 5487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015366315223870012,
+      "loss": 0.3452,
+      "step": 5488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001536476804975143,
+      "loss": 0.3463,
+      "step": 5489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015363220695296828,
+      "loss": 0.3102,
+      "step": 5490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015361673160558227,
+      "loss": 0.2534,
+      "step": 5491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015360125445587638,
+      "loss": 0.338,
+      "step": 5492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015358577550437095,
+      "loss": 0.4745,
+      "step": 5493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015357029475158618,
+      "loss": 0.42,
+      "step": 5494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015355481219804263,
+      "loss": 0.2296,
+      "step": 5495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015353932784426062,
+      "loss": 0.2155,
+      "step": 5496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015352384169076068,
+      "loss": 0.4107,
+      "step": 5497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015350835373806343,
+      "loss": 0.4286,
+      "step": 5498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001534928639866894,
+      "loss": 0.4593,
+      "step": 5499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015347737243715933,
+      "loss": 0.2333,
+      "step": 5500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000153461879089994,
+      "loss": 0.1383,
+      "step": 5501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015344638394571412,
+      "loss": 0.3688,
+      "step": 5502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015343088700484064,
+      "loss": 0.3248,
+      "step": 5503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001534153882678945,
+      "loss": 0.3811,
+      "step": 5504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015339988773539664,
+      "loss": 0.2455,
+      "step": 5505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001533843854078681,
+      "loss": 0.3554,
+      "step": 5506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015336888128583,
+      "loss": 0.1994,
+      "step": 5507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015335337536980355,
+      "loss": 0.4806,
+      "step": 5508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015333786766030992,
+      "loss": 0.2416,
+      "step": 5509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015332235815787047,
+      "loss": 0.235,
+      "step": 5510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015330684686300648,
+      "loss": 0.3462,
+      "step": 5511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015329133377623937,
+      "loss": 0.2538,
+      "step": 5512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001532758188980907,
+      "loss": 0.249,
+      "step": 5513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015326030222908186,
+      "loss": 0.3907,
+      "step": 5514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015324478376973452,
+      "loss": 0.4273,
+      "step": 5515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015322926352057033,
+      "loss": 0.2844,
+      "step": 5516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015321374148211103,
+      "loss": 0.3016,
+      "step": 5517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015319821765487828,
+      "loss": 0.4413,
+      "step": 5518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015318269203939405,
+      "loss": 0.3763,
+      "step": 5519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015316716463618016,
+      "loss": 0.4158,
+      "step": 5520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015315163544575854,
+      "loss": 0.268,
+      "step": 5521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015313610446865126,
+      "loss": 0.44,
+      "step": 5522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015312057170538035,
+      "loss": 0.2865,
+      "step": 5523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015310503715646796,
+      "loss": 0.2494,
+      "step": 5524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015308950082243628,
+      "loss": 0.3624,
+      "step": 5525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015307396270380753,
+      "loss": 0.4139,
+      "step": 5526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001530584228011041,
+      "loss": 0.4598,
+      "step": 5527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015304288111484828,
+      "loss": 0.4408,
+      "step": 5528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015302733764556252,
+      "loss": 0.1732,
+      "step": 5529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015301179239376938,
+      "loss": 0.3812,
+      "step": 5530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001529962453599913,
+      "loss": 0.2737,
+      "step": 5531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015298069654475098,
+      "loss": 0.2994,
+      "step": 5532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015296514594857107,
+      "loss": 0.5436,
+      "step": 5533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015294959357197428,
+      "loss": 0.2987,
+      "step": 5534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001529340394154834,
+      "loss": 0.1853,
+      "step": 5535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015291848347962134,
+      "loss": 0.3597,
+      "step": 5536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015290292576491092,
+      "loss": 0.4135,
+      "step": 5537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015288736627187518,
+      "loss": 0.2704,
+      "step": 5538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001528718050010371,
+      "loss": 0.2522,
+      "step": 5539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015285624195291985,
+      "loss": 0.4723,
+      "step": 5540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015284067712804648,
+      "loss": 0.2575,
+      "step": 5541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015282511052694026,
+      "loss": 0.4893,
+      "step": 5542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015280954215012447,
+      "loss": 0.3403,
+      "step": 5543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015279397199812237,
+      "loss": 0.1535,
+      "step": 5544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015277840007145744,
+      "loss": 0.3888,
+      "step": 5545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015276282637065306,
+      "loss": 0.453,
+      "step": 5546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015274725089623275,
+      "loss": 0.2996,
+      "step": 5547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015273167364872008,
+      "loss": 0.4075,
+      "step": 5548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015271609462863875,
+      "loss": 0.3975,
+      "step": 5549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015270051383651235,
+      "loss": 0.26,
+      "step": 5550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015268493127286461,
+      "loss": 0.31,
+      "step": 5551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015266934693821944,
+      "loss": 0.4327,
+      "step": 5552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001526537608331006,
+      "loss": 0.2509,
+      "step": 5553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015263817295803214,
+      "loss": 0.2436,
+      "step": 5554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001526225833135379,
+      "loss": 0.4454,
+      "step": 5555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015260699190014203,
+      "loss": 0.3319,
+      "step": 5556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015259139871836861,
+      "loss": 0.3281,
+      "step": 5557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015257580376874173,
+      "loss": 0.5465,
+      "step": 5558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015256020705178573,
+      "loss": 0.2041,
+      "step": 5559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015254460856802479,
+      "loss": 0.6084,
+      "step": 5560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015252900831798334,
+      "loss": 0.4651,
+      "step": 5561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015251340630218572,
+      "loss": 0.3099,
+      "step": 5562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001524978025211564,
+      "loss": 0.5141,
+      "step": 5563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001524821969754199,
+      "loss": 0.338,
+      "step": 5564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015246658966550084,
+      "loss": 0.3151,
+      "step": 5565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001524509805919238,
+      "loss": 0.362,
+      "step": 5566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015243536975521345,
+      "loss": 0.4169,
+      "step": 5567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015241975715589464,
+      "loss": 0.4277,
+      "step": 5568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015240414279449213,
+      "loss": 0.2217,
+      "step": 5569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015238852667153081,
+      "loss": 0.2742,
+      "step": 5570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015237290878753563,
+      "loss": 0.2336,
+      "step": 5571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001523572891430315,
+      "loss": 0.3431,
+      "step": 5572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001523416677385436,
+      "loss": 0.3402,
+      "step": 5573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015232604457459696,
+      "loss": 0.2114,
+      "step": 5574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015231041965171676,
+      "loss": 0.4944,
+      "step": 5575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015229479297042823,
+      "loss": 0.6216,
+      "step": 5576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015227916453125668,
+      "loss": 0.3308,
+      "step": 5577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015226353433472745,
+      "loss": 0.2232,
+      "step": 5578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001522479023813659,
+      "loss": 0.1557,
+      "step": 5579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015223226867169758,
+      "loss": 0.3159,
+      "step": 5580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015221663320624795,
+      "loss": 0.4307,
+      "step": 5581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015220099598554262,
+      "loss": 0.4263,
+      "step": 5582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015218535701010723,
+      "loss": 0.3249,
+      "step": 5583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001521697162804675,
+      "loss": 0.3826,
+      "step": 5584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015215407379714913,
+      "loss": 0.3121,
+      "step": 5585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015213842956067803,
+      "loss": 0.8472,
+      "step": 5586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015212278357158,
+      "loss": 0.3752,
+      "step": 5587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015210713583038103,
+      "loss": 0.3957,
+      "step": 5588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001520914863376071,
+      "loss": 0.3022,
+      "step": 5589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015207583509378427,
+      "loss": 0.2749,
+      "step": 5590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001520601820994386,
+      "loss": 0.3848,
+      "step": 5591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015204452735509638,
+      "loss": 0.2973,
+      "step": 5592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015202887086128374,
+      "loss": 0.3075,
+      "step": 5593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000152013212618527,
+      "loss": 0.2765,
+      "step": 5594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001519975526273525,
+      "loss": 0.4228,
+      "step": 5595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015198189088828668,
+      "loss": 0.2833,
+      "step": 5596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015196622740185601,
+      "loss": 0.363,
+      "step": 5597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015195056216858697,
+      "loss": 0.576,
+      "step": 5598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001519348951890062,
+      "loss": 0.21,
+      "step": 5599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015191922646364027,
+      "loss": 0.2475,
+      "step": 5600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015190355599301598,
+      "loss": 0.3239,
+      "step": 5601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015188788377766001,
+      "loss": 0.4566,
+      "step": 5602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001518722098180992,
+      "loss": 0.2763,
+      "step": 5603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015185653411486048,
+      "loss": 0.2919,
+      "step": 5604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015184085666847068,
+      "loss": 0.2178,
+      "step": 5605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001518251774794569,
+      "loss": 0.3496,
+      "step": 5606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015180949654834615,
+      "loss": 0.3173,
+      "step": 5607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001517938138756655,
+      "loss": 0.3277,
+      "step": 5608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015177812946194226,
+      "loss": 0.4193,
+      "step": 5609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015176244330770348,
+      "loss": 0.3738,
+      "step": 5610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015174675541347658,
+      "loss": 0.2453,
+      "step": 5611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015173106577978884,
+      "loss": 1.0799,
+      "step": 5612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001517153744071677,
+      "loss": 0.5706,
+      "step": 5613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001516996812961406,
+      "loss": 0.5101,
+      "step": 5614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015168398644723507,
+      "loss": 0.4584,
+      "step": 5615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015166828986097868,
+      "loss": 0.396,
+      "step": 5616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001516525915378991,
+      "loss": 0.29,
+      "step": 5617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015163689147852405,
+      "loss": 0.3781,
+      "step": 5618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015162118968338118,
+      "loss": 0.3827,
+      "step": 5619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001516054861529984,
+      "loss": 0.2615,
+      "step": 5620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001515897808879035,
+      "loss": 0.2726,
+      "step": 5621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015157407388862452,
+      "loss": 0.4926,
+      "step": 5622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015155836515568937,
+      "loss": 0.399,
+      "step": 5623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015154265468962609,
+      "loss": 0.4297,
+      "step": 5624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015152694249096285,
+      "loss": 0.2194,
+      "step": 5625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015151122856022777,
+      "loss": 0.2116,
+      "step": 5626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015149551289794905,
+      "loss": 0.4634,
+      "step": 5627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015147979550465503,
+      "loss": 0.4547,
+      "step": 5628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015146407638087398,
+      "loss": 0.4119,
+      "step": 5629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015144835552713436,
+      "loss": 0.3933,
+      "step": 5630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015143263294396458,
+      "loss": 0.2156,
+      "step": 5631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001514169086318932,
+      "loss": 0.2183,
+      "step": 5632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015140118259144872,
+      "loss": 0.4831,
+      "step": 5633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015138545482315982,
+      "loss": 0.3273,
+      "step": 5634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001513697253275552,
+      "loss": 0.2768,
+      "step": 5635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015135399410516354,
+      "loss": 0.4413,
+      "step": 5636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015133826115651373,
+      "loss": 0.315,
+      "step": 5637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015132252648213454,
+      "loss": 0.2796,
+      "step": 5638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015130679008255497,
+      "loss": 0.4001,
+      "step": 5639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015129105195830396,
+      "loss": 0.3926,
+      "step": 5640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001512753121099105,
+      "loss": 0.4475,
+      "step": 5641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015125957053790378,
+      "loss": 0.529,
+      "step": 5642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015124382724281287,
+      "loss": 0.2058,
+      "step": 5643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015122808222516705,
+      "loss": 0.4101,
+      "step": 5644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001512123354854955,
+      "loss": 0.2684,
+      "step": 5645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015119658702432762,
+      "loss": 0.1959,
+      "step": 5646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015118083684219276,
+      "loss": 0.8218,
+      "step": 5647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015116508493962037,
+      "loss": 0.4579,
+      "step": 5648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015114933131713993,
+      "loss": 0.2613,
+      "step": 5649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000151133575975281,
+      "loss": 0.8248,
+      "step": 5650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001511178189145732,
+      "loss": 0.2842,
+      "step": 5651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015110206013554622,
+      "loss": 0.3168,
+      "step": 5652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015108629963872975,
+      "loss": 0.3073,
+      "step": 5653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015107053742465364,
+      "loss": 0.299,
+      "step": 5654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015105477349384766,
+      "loss": 0.3092,
+      "step": 5655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015103900784684178,
+      "loss": 0.3124,
+      "step": 5656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001510232404841659,
+      "loss": 0.2837,
+      "step": 5657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015100747140635008,
+      "loss": 0.3241,
+      "step": 5658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001509917006139244,
+      "loss": 0.187,
+      "step": 5659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015097592810741893,
+      "loss": 0.3541,
+      "step": 5660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015096015388736396,
+      "loss": 0.6193,
+      "step": 5661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015094437795428963,
+      "loss": 0.3845,
+      "step": 5662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015092860030872633,
+      "loss": 0.4943,
+      "step": 5663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015091282095120438,
+      "loss": 0.5994,
+      "step": 5664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015089703988225428,
+      "loss": 0.3841,
+      "step": 5665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015088125710240635,
+      "loss": 0.3662,
+      "step": 5666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015086547261219127,
+      "loss": 0.5372,
+      "step": 5667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015084968641213958,
+      "loss": 0.34,
+      "step": 5668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001508338985027819,
+      "loss": 0.4402,
+      "step": 5669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000150818108884649,
+      "loss": 0.3416,
+      "step": 5670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015080231755827163,
+      "loss": 0.4326,
+      "step": 5671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015078652452418063,
+      "loss": 0.2407,
+      "step": 5672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015077072978290682,
+      "loss": 0.496,
+      "step": 5673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015075493333498115,
+      "loss": 0.5143,
+      "step": 5674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015073913518093466,
+      "loss": 0.4357,
+      "step": 5675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015072333532129836,
+      "loss": 0.3915,
+      "step": 5676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015070753375660343,
+      "loss": 0.2663,
+      "step": 5677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015069173048738098,
+      "loss": 0.3058,
+      "step": 5678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001506759255141622,
+      "loss": 0.2538,
+      "step": 5679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015066011883747846,
+      "loss": 0.2759,
+      "step": 5680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015064431045786101,
+      "loss": 0.3654,
+      "step": 5681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015062850037584133,
+      "loss": 0.4035,
+      "step": 5682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015061268859195082,
+      "loss": 0.5856,
+      "step": 5683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015059687510672105,
+      "loss": 0.438,
+      "step": 5684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001505810599206835,
+      "loss": 0.3031,
+      "step": 5685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015056524303436986,
+      "loss": 0.4146,
+      "step": 5686
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001505494244483118,
+      "loss": 0.3443,
+      "step": 5687
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000150533604163041,
+      "loss": 0.504,
+      "step": 5688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001505177821790894,
+      "loss": 0.5538,
+      "step": 5689
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015050195849698866,
+      "loss": 0.2828,
+      "step": 5690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001504861331172709,
+      "loss": 0.4611,
+      "step": 5691
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001504703060404679,
+      "loss": 0.5255,
+      "step": 5692
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015045447726711182,
+      "loss": 0.3207,
+      "step": 5693
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001504386467977347,
+      "loss": 0.4796,
+      "step": 5694
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001504228146328686,
+      "loss": 0.4783,
+      "step": 5695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015040698077304588,
+      "loss": 0.3358,
+      "step": 5696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015039114521879867,
+      "loss": 0.4655,
+      "step": 5697
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015037530797065932,
+      "loss": 0.3718,
+      "step": 5698
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015035946902916013,
+      "loss": 0.2414,
+      "step": 5699
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001503436283948336,
+      "loss": 0.322,
+      "step": 5700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015032778606821222,
+      "loss": 0.3842,
+      "step": 5701
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001503119420498285,
+      "loss": 0.3081,
+      "step": 5702
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015029609634021503,
+      "loss": 0.3463,
+      "step": 5703
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001502802489399045,
+      "loss": 0.3989,
+      "step": 5704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015026439984942952,
+      "loss": 0.3247,
+      "step": 5705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015024854906932298,
+      "loss": 0.3899,
+      "step": 5706
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001502326966001176,
+      "loss": 0.1889,
+      "step": 5707
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015021684244234632,
+      "loss": 0.3733,
+      "step": 5708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015020098659654206,
+      "loss": 0.173,
+      "step": 5709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001501851290632378,
+      "loss": 0.3331,
+      "step": 5710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001501692698429666,
+      "loss": 0.5037,
+      "step": 5711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001501534089362616,
+      "loss": 0.3636,
+      "step": 5712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015013754634365587,
+      "loss": 0.544,
+      "step": 5713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015012168206568268,
+      "loss": 0.247,
+      "step": 5714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015010581610287534,
+      "loss": 0.3781,
+      "step": 5715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001500899484557672,
+      "loss": 0.3132,
+      "step": 5716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015007407912489158,
+      "loss": 0.262,
+      "step": 5717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001500582081107819,
+      "loss": 0.3128,
+      "step": 5718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015004233541397174,
+      "loss": 0.3426,
+      "step": 5719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015002646103499462,
+      "loss": 0.544,
+      "step": 5720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015001058497438415,
+      "loss": 0.3115,
+      "step": 5721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014999470723267406,
+      "loss": 0.1761,
+      "step": 5722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014997882781039802,
+      "loss": 0.5648,
+      "step": 5723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014996294670808984,
+      "loss": 0.3598,
+      "step": 5724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001499470639262833,
+      "loss": 0.3937,
+      "step": 5725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014993117946551242,
+      "loss": 0.2553,
+      "step": 5726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014991529332631105,
+      "loss": 0.4713,
+      "step": 5727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001498994055092132,
+      "loss": 0.2614,
+      "step": 5728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014988351601475302,
+      "loss": 0.2969,
+      "step": 5729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014986762484346457,
+      "loss": 0.3652,
+      "step": 5730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014985173199588204,
+      "loss": 0.2872,
+      "step": 5731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014983583747253966,
+      "loss": 0.4096,
+      "step": 5732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014981994127397175,
+      "loss": 0.4997,
+      "step": 5733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014980404340071265,
+      "loss": 0.4963,
+      "step": 5734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014978814385329674,
+      "loss": 0.2212,
+      "step": 5735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001497722426322585,
+      "loss": 0.3146,
+      "step": 5736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014975633973813242,
+      "loss": 0.3517,
+      "step": 5737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014974043517145317,
+      "loss": 0.4945,
+      "step": 5738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014972452893275524,
+      "loss": 0.2966,
+      "step": 5739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001497086210225734,
+      "loss": 0.3627,
+      "step": 5740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001496927114414424,
+      "loss": 0.2054,
+      "step": 5741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000149676800189897,
+      "loss": 0.3741,
+      "step": 5742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014966088726847207,
+      "loss": 0.2706,
+      "step": 5743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014964497267770253,
+      "loss": 0.1911,
+      "step": 5744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014962905641812335,
+      "loss": 0.4899,
+      "step": 5745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001496131384902695,
+      "loss": 0.3665,
+      "step": 5746
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014959721889467613,
+      "loss": 0.5252,
+      "step": 5747
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014958129763187835,
+      "loss": 0.3239,
+      "step": 5748
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014956537470241135,
+      "loss": 0.4332,
+      "step": 5749
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014954945010681037,
+      "loss": 0.3133,
+      "step": 5750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001495335238456107,
+      "loss": 0.3175,
+      "step": 5751
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014951759591934775,
+      "loss": 0.5612,
+      "step": 5752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014950166632855688,
+      "loss": 0.2336,
+      "step": 5753
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001494857350737736,
+      "loss": 0.4036,
+      "step": 5754
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001494698021555334,
+      "loss": 0.1913,
+      "step": 5755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014945386757437193,
+      "loss": 0.3569,
+      "step": 5756
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014943793133082473,
+      "loss": 0.3295,
+      "step": 5757
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014942199342542756,
+      "loss": 0.2296,
+      "step": 5758
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014940605385871616,
+      "loss": 0.1391,
+      "step": 5759
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014939011263122634,
+      "loss": 0.325,
+      "step": 5760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014937416974349397,
+      "loss": 0.4382,
+      "step": 5761
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000149358225196055,
+      "loss": 0.3071,
+      "step": 5762
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001493422789894453,
+      "loss": 0.2923,
+      "step": 5763
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014932633112420095,
+      "loss": 0.2968,
+      "step": 5764
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014931038160085805,
+      "loss": 0.2423,
+      "step": 5765
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014929443041995278,
+      "loss": 0.3216,
+      "step": 5766
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014927847758202124,
+      "loss": 0.1285,
+      "step": 5767
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001492625230875998,
+      "loss": 0.3961,
+      "step": 5768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014924656693722467,
+      "loss": 0.2972,
+      "step": 5769
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014923060913143223,
+      "loss": 0.3179,
+      "step": 5770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014921464967075895,
+      "loss": 0.2729,
+      "step": 5771
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014919868855574125,
+      "loss": 0.3846,
+      "step": 5772
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001491827257869157,
+      "loss": 0.2405,
+      "step": 5773
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014916676136481888,
+      "loss": 0.5234,
+      "step": 5774
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014915079528998742,
+      "loss": 0.2966,
+      "step": 5775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000149134827562958,
+      "loss": 0.3219,
+      "step": 5776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014911885818426741,
+      "loss": 0.2587,
+      "step": 5777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014910288715445244,
+      "loss": 0.3717,
+      "step": 5778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014908691447404994,
+      "loss": 0.1779,
+      "step": 5779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014907094014359688,
+      "loss": 0.2485,
+      "step": 5780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001490549641636302,
+      "loss": 0.2307,
+      "step": 5781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014903898653468691,
+      "loss": 0.2424,
+      "step": 5782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014902300725730413,
+      "loss": 0.3664,
+      "step": 5783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014900702633201897,
+      "loss": 0.3216,
+      "step": 5784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014899104375936867,
+      "loss": 0.3824,
+      "step": 5785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014897505953989046,
+      "loss": 0.4631,
+      "step": 5786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014895907367412165,
+      "loss": 0.2475,
+      "step": 5787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001489430861625996,
+      "loss": 0.3008,
+      "step": 5788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014892709700586168,
+      "loss": 0.3063,
+      "step": 5789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001489111062044455,
+      "loss": 0.1956,
+      "step": 5790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014889511375888842,
+      "loss": 0.2844,
+      "step": 5791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014887911966972813,
+      "loss": 0.1375,
+      "step": 5792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014886312393750227,
+      "loss": 0.2463,
+      "step": 5793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014884712656274847,
+      "loss": 0.2576,
+      "step": 5794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014883112754600453,
+      "loss": 0.3972,
+      "step": 5795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001488151268878083,
+      "loss": 0.3128,
+      "step": 5796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014879912458869748,
+      "loss": 0.3785,
+      "step": 5797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014878312064921017,
+      "loss": 0.2284,
+      "step": 5798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001487671150698842,
+      "loss": 0.2128,
+      "step": 5799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001487511078512577,
+      "loss": 0.4348,
+      "step": 5800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014873509899386868,
+      "loss": 0.236,
+      "step": 5801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014871908849825534,
+      "loss": 0.3222,
+      "step": 5802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014870307636495578,
+      "loss": 0.2888,
+      "step": 5803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014868706259450832,
+      "loss": 0.3462,
+      "step": 5804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014867104718745125,
+      "loss": 0.2998,
+      "step": 5805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014865503014432292,
+      "loss": 0.3337,
+      "step": 5806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014863901146566172,
+      "loss": 0.2411,
+      "step": 5807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014862299115200612,
+      "loss": 0.476,
+      "step": 5808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014860696920389468,
+      "loss": 0.6119,
+      "step": 5809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001485909456218659,
+      "loss": 0.2369,
+      "step": 5810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014857492040645852,
+      "loss": 0.4807,
+      "step": 5811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014855889355821112,
+      "loss": 0.3317,
+      "step": 5812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014854286507766251,
+      "loss": 0.2689,
+      "step": 5813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014852683496535145,
+      "loss": 0.1833,
+      "step": 5814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001485108032218168,
+      "loss": 0.223,
+      "step": 5815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014849476984759747,
+      "loss": 0.2532,
+      "step": 5816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001484787348432324,
+      "loss": 0.2838,
+      "step": 5817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014846269820926065,
+      "loss": 0.1792,
+      "step": 5818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014844665994622123,
+      "loss": 0.1903,
+      "step": 5819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001484306200546533,
+      "loss": 0.3362,
+      "step": 5820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014841457853509606,
+      "loss": 0.3904,
+      "step": 5821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001483985353880887,
+      "loss": 0.3099,
+      "step": 5822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014838249061417054,
+      "loss": 0.2633,
+      "step": 5823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014836644421388088,
+      "loss": 0.1684,
+      "step": 5824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014835039618775917,
+      "loss": 0.3104,
+      "step": 5825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001483343465363448,
+      "loss": 0.4749,
+      "step": 5826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014831829526017735,
+      "loss": 0.3353,
+      "step": 5827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014830224235979635,
+      "loss": 0.2957,
+      "step": 5828
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001482861878357414,
+      "loss": 0.2574,
+      "step": 5829
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001482701316885522,
+      "loss": 0.2881,
+      "step": 5830
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001482540739187684,
+      "loss": 0.2188,
+      "step": 5831
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014823801452692992,
+      "loss": 0.2034,
+      "step": 5832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014822195351357648,
+      "loss": 0.3192,
+      "step": 5833
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000148205890879248,
+      "loss": 0.2923,
+      "step": 5834
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014818982662448445,
+      "loss": 0.1803,
+      "step": 5835
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001481737607498258,
+      "loss": 0.4081,
+      "step": 5836
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014815769325581208,
+      "loss": 0.3967,
+      "step": 5837
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014814162414298343,
+      "loss": 0.2463,
+      "step": 5838
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014812555341188002,
+      "loss": 0.4848,
+      "step": 5839
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014810948106304205,
+      "loss": 0.3998,
+      "step": 5840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001480934070970098,
+      "loss": 0.5706,
+      "step": 5841
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014807733151432357,
+      "loss": 0.2028,
+      "step": 5842
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014806125431552372,
+      "loss": 0.2023,
+      "step": 5843
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014804517550115074,
+      "loss": 0.2898,
+      "step": 5844
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014802909507174512,
+      "loss": 0.3115,
+      "step": 5845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001480130130278474,
+      "loss": 0.2663,
+      "step": 5846
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001479969293699981,
+      "loss": 0.1546,
+      "step": 5847
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014798084409873793,
+      "loss": 0.2011,
+      "step": 5848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014796475721460758,
+      "loss": 0.2037,
+      "step": 5849
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014794866871814782,
+      "loss": 0.1176,
+      "step": 5850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014793257860989947,
+      "loss": 0.315,
+      "step": 5851
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001479164868904034,
+      "loss": 0.4211,
+      "step": 5852
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014790039356020048,
+      "loss": 0.2828,
+      "step": 5853
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014788429861983175,
+      "loss": 0.2144,
+      "step": 5854
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001478682020698382,
+      "loss": 0.3172,
+      "step": 5855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014785210391076093,
+      "loss": 0.1317,
+      "step": 5856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014783600414314108,
+      "loss": 0.0739,
+      "step": 5857
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001478199027675198,
+      "loss": 0.2193,
+      "step": 5858
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014780379978443842,
+      "loss": 0.1174,
+      "step": 5859
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014778769519443815,
+      "loss": 0.2677,
+      "step": 5860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014777158899806039,
+      "loss": 0.1483,
+      "step": 5861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014775548119584655,
+      "loss": 0.3587,
+      "step": 5862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014773937178833807,
+      "loss": 0.1842,
+      "step": 5863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014772326077607648,
+      "loss": 0.1676,
+      "step": 5864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014770714815960336,
+      "loss": 0.1415,
+      "step": 5865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014769103393946032,
+      "loss": 0.111,
+      "step": 5866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014767491811618904,
+      "loss": 0.2265,
+      "step": 5867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014765880069033123,
+      "loss": 0.1394,
+      "step": 5868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014764268166242873,
+      "loss": 0.5216,
+      "step": 5869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014762656103302333,
+      "loss": 0.2988,
+      "step": 5870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014761043880265692,
+      "loss": 0.1622,
+      "step": 5871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00014759431497187147,
+      "loss": 0.2855,
+      "step": 5872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014757818954120902,
+      "loss": 0.1995,
+      "step": 5873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014756206251121152,
+      "loss": 0.1384,
+      "step": 5874
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014754593388242117,
+      "loss": 0.3499,
+      "step": 5875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001475298036553801,
+      "loss": 0.1567,
+      "step": 5876
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001475136718306305,
+      "loss": 0.2129,
+      "step": 5877
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001474975384087147,
+      "loss": 0.1299,
+      "step": 5878
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014748140339017498,
+      "loss": 0.2237,
+      "step": 5879
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014746526677555374,
+      "loss": 0.1636,
+      "step": 5880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014744912856539335,
+      "loss": 0.1612,
+      "step": 5881
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014743298876023636,
+      "loss": 0.2505,
+      "step": 5882
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001474168473606253,
+      "loss": 0.2564,
+      "step": 5883
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014740070436710273,
+      "loss": 0.3087,
+      "step": 5884
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014738455978021136,
+      "loss": 0.1096,
+      "step": 5885
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014736841360049377,
+      "loss": 0.191,
+      "step": 5886
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014735226582849283,
+      "loss": 0.1871,
+      "step": 5887
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014733611646475128,
+      "loss": 0.4793,
+      "step": 5888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000147319965509812,
+      "loss": 0.1564,
+      "step": 5889
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014730381296421786,
+      "loss": 0.2984,
+      "step": 5890
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014728765882851192,
+      "loss": 0.1832,
+      "step": 5891
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014727150310323714,
+      "loss": 0.1116,
+      "step": 5892
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014725534578893657,
+      "loss": 0.114,
+      "step": 5893
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014723918688615338,
+      "loss": 0.1626,
+      "step": 5894
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014722302639543072,
+      "loss": 0.2845,
+      "step": 5895
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014720686431731185,
+      "loss": 0.148,
+      "step": 5896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014719070065234005,
+      "loss": 0.1458,
+      "step": 5897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001471745354010586,
+      "loss": 0.1885,
+      "step": 5898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000147158368564011,
+      "loss": 0.2814,
+      "step": 5899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014714220014174062,
+      "loss": 0.4326,
+      "step": 5900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014712603013479098,
+      "loss": 0.2493,
+      "step": 5901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014710985854370561,
+      "loss": 0.294,
+      "step": 5902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014709368536902818,
+      "loss": 0.1839,
+      "step": 5903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001470775106113023,
+      "loss": 0.1875,
+      "step": 5904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014706133427107169,
+      "loss": 0.2471,
+      "step": 5905
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014704515634888013,
+      "loss": 0.1728,
+      "step": 5906
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014702897684527137,
+      "loss": 0.1265,
+      "step": 5907
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001470127957607894,
+      "loss": 0.1549,
+      "step": 5908
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014699661309597803,
+      "loss": 0.1381,
+      "step": 5909
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014698042885138136,
+      "loss": 0.1218,
+      "step": 5910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001469642430275433,
+      "loss": 0.2206,
+      "step": 5911
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014694805562500798,
+      "loss": 0.266,
+      "step": 5912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014693186664431956,
+      "loss": 0.1922,
+      "step": 5913
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001469156760860222,
+      "loss": 0.1813,
+      "step": 5914
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014689948395066018,
+      "loss": 0.2151,
+      "step": 5915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014688329023877775,
+      "loss": 0.1908,
+      "step": 5916
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014686709495091932,
+      "loss": 0.0744,
+      "step": 5917
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001468508980876292,
+      "loss": 0.2085,
+      "step": 5918
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014683469964945196,
+      "loss": 0.1937,
+      "step": 5919
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014681849963693204,
+      "loss": 0.1503,
+      "step": 5920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000146802298050614,
+      "loss": 0.1541,
+      "step": 5921
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014678609489104247,
+      "loss": 0.1791,
+      "step": 5922
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014676989015876212,
+      "loss": 0.1618,
+      "step": 5923
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014675368385431767,
+      "loss": 0.185,
+      "step": 5924
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014673747597825386,
+      "loss": 0.1912,
+      "step": 5925
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014672126653111556,
+      "loss": 0.1848,
+      "step": 5926
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014670505551344764,
+      "loss": 0.1078,
+      "step": 5927
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014668884292579502,
+      "loss": 0.3315,
+      "step": 5928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00014667262876870267,
+      "loss": 0.2138,
+      "step": 5929
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014665641304271567,
+      "loss": 0.1938,
+      "step": 5930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014664019574837906,
+      "loss": 0.179,
+      "step": 5931
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014662397688623802,
+      "loss": 0.1423,
+      "step": 5932
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014660775645683772,
+      "loss": 0.0989,
+      "step": 5933
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014659153446072343,
+      "loss": 0.165,
+      "step": 5934
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014657531089844043,
+      "loss": 0.3503,
+      "step": 5935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001465590857705341,
+      "loss": 0.2128,
+      "step": 5936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014654285907754984,
+      "loss": 0.2313,
+      "step": 5937
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014652663082003305,
+      "loss": 0.1642,
+      "step": 5938
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014651040099852934,
+      "loss": 0.0711,
+      "step": 5939
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014649416961358416,
+      "loss": 0.1996,
+      "step": 5940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001464779366657432,
+      "loss": 0.2162,
+      "step": 5941
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014646170215555215,
+      "loss": 0.161,
+      "step": 5942
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001464454660835567,
+      "loss": 0.262,
+      "step": 5943
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014642922845030257,
+      "loss": 0.1115,
+      "step": 5944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014641298925633566,
+      "loss": 0.3265,
+      "step": 5945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014639674850220184,
+      "loss": 0.1425,
+      "step": 5946
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000146380506188447,
+      "loss": 0.1798,
+      "step": 5947
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014636426231561718,
+      "loss": 0.0913,
+      "step": 5948
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014634801688425837,
+      "loss": 0.1329,
+      "step": 5949
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014633176989491668,
+      "loss": 0.3306,
+      "step": 5950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014631552134813821,
+      "loss": 0.2053,
+      "step": 5951
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001462992712444692,
+      "loss": 0.1357,
+      "step": 5952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014628301958445592,
+      "loss": 0.1684,
+      "step": 5953
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014626676636864456,
+      "loss": 0.1631,
+      "step": 5954
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001462505115975816,
+      "loss": 0.1549,
+      "step": 5955
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014623425527181337,
+      "loss": 0.2897,
+      "step": 5956
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014621799739188634,
+      "loss": 0.1328,
+      "step": 5957
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014620173795834698,
+      "loss": 0.3354,
+      "step": 5958
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014618547697174187,
+      "loss": 0.2606,
+      "step": 5959
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014616921443261766,
+      "loss": 0.2569,
+      "step": 5960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014615295034152098,
+      "loss": 0.3164,
+      "step": 5961
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014613668469899856,
+      "loss": 0.1297,
+      "step": 5962
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014612041750559714,
+      "loss": 0.1236,
+      "step": 5963
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001461041487618636,
+      "loss": 0.1467,
+      "step": 5964
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014608787846834474,
+      "loss": 0.1723,
+      "step": 5965
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014607160662558747,
+      "loss": 0.3853,
+      "step": 5966
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014605533323413887,
+      "loss": 0.1984,
+      "step": 5967
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014603905829454588,
+      "loss": 0.1069,
+      "step": 5968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014602278180735564,
+      "loss": 0.1106,
+      "step": 5969
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014600650377311522,
+      "loss": 0.1096,
+      "step": 5970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014599022419237184,
+      "loss": 0.1304,
+      "step": 5971
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014597394306567273,
+      "loss": 0.2327,
+      "step": 5972
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001459576603935652,
+      "loss": 0.1165,
+      "step": 5973
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001459413761765965,
+      "loss": 0.2105,
+      "step": 5974
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014592509041531416,
+      "loss": 0.1978,
+      "step": 5975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014590880311026556,
+      "loss": 0.2197,
+      "step": 5976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014589251426199814,
+      "loss": 0.1349,
+      "step": 5977
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001458762238710595,
+      "loss": 0.1203,
+      "step": 5978
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001458599319379973,
+      "loss": 0.2135,
+      "step": 5979
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014584363846335902,
+      "loss": 0.1776,
+      "step": 5980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014582734344769258,
+      "loss": 0.2638,
+      "step": 5981
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014581104689154555,
+      "loss": 0.4038,
+      "step": 5982
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014579474879546585,
+      "loss": 0.2109,
+      "step": 5983
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001457784491600013,
+      "loss": 0.2405,
+      "step": 5984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014576214798569978,
+      "loss": 0.1609,
+      "step": 5985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00014574584527310931,
+      "loss": 0.1223,
+      "step": 5986
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014572954102277786,
+      "loss": 0.1698,
+      "step": 5987
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014571323523525356,
+      "loss": 0.2596,
+      "step": 5988
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014569692791108446,
+      "loss": 0.3458,
+      "step": 5989
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014568061905081875,
+      "loss": 0.13,
+      "step": 5990
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014566430865500462,
+      "loss": 0.2134,
+      "step": 5991
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014564799672419038,
+      "loss": 0.1763,
+      "step": 5992
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014563168325892436,
+      "loss": 0.1186,
+      "step": 5993
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001456153682597549,
+      "loss": 0.1367,
+      "step": 5994
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014559905172723046,
+      "loss": 0.2157,
+      "step": 5995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001455827336618995,
+      "loss": 0.312,
+      "step": 5996
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014556641406431058,
+      "loss": 0.336,
+      "step": 5997
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001455500929350122,
+      "loss": 0.1156,
+      "step": 5998
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001455337702745531,
+      "loss": 0.3107,
+      "step": 5999
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014551744608348188,
+      "loss": 0.1931,
+      "step": 6000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014550112036234731,
+      "loss": 0.1629,
+      "step": 6001
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014548479311169817,
+      "loss": 0.1542,
+      "step": 6002
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014546846433208333,
+      "loss": 0.166,
+      "step": 6003
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014545213402405164,
+      "loss": 0.1471,
+      "step": 6004
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014543580218815205,
+      "loss": 0.1103,
+      "step": 6005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014541946882493356,
+      "loss": 0.1521,
+      "step": 6006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014540313393494522,
+      "loss": 0.1382,
+      "step": 6007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014538679751873612,
+      "loss": 0.4787,
+      "step": 6008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001453704595768554,
+      "loss": 0.253,
+      "step": 6009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014535412010985223,
+      "loss": 0.1668,
+      "step": 6010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014533777911827594,
+      "loss": 0.1271,
+      "step": 6011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014532143660267574,
+      "loss": 0.1976,
+      "step": 6012
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014530509256360102,
+      "loss": 0.1598,
+      "step": 6013
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014528874700160124,
+      "loss": 0.1236,
+      "step": 6014
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014527239991722575,
+      "loss": 0.1339,
+      "step": 6015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014525605131102413,
+      "loss": 0.255,
+      "step": 6016
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014523970118354587,
+      "loss": 0.2384,
+      "step": 6017
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014522334953534068,
+      "loss": 0.178,
+      "step": 6018
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001452069963669581,
+      "loss": 0.1153,
+      "step": 6019
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014519064167894794,
+      "loss": 0.1762,
+      "step": 6020
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014517428547185989,
+      "loss": 0.1893,
+      "step": 6021
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014515792774624382,
+      "loss": 0.1554,
+      "step": 6022
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014514156850264952,
+      "loss": 0.1402,
+      "step": 6023
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014512520774162698,
+      "loss": 0.2387,
+      "step": 6024
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014510884546372612,
+      "loss": 0.1508,
+      "step": 6025
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014509248166949697,
+      "loss": 0.2439,
+      "step": 6026
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001450761163594896,
+      "loss": 0.2615,
+      "step": 6027
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014505974953425412,
+      "loss": 0.101,
+      "step": 6028
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014504338119434066,
+      "loss": 0.134,
+      "step": 6029
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014502701134029954,
+      "loss": 0.175,
+      "step": 6030
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014501063997268094,
+      "loss": 0.1827,
+      "step": 6031
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014499426709203522,
+      "loss": 0.1822,
+      "step": 6032
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014497789269891275,
+      "loss": 0.175,
+      "step": 6033
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014496151679386395,
+      "loss": 0.2122,
+      "step": 6034
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001449451393774393,
+      "loss": 0.1909,
+      "step": 6035
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001449287604501893,
+      "loss": 0.0916,
+      "step": 6036
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014491238001266454,
+      "loss": 0.1296,
+      "step": 6037
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014489599806541566,
+      "loss": 0.1139,
+      "step": 6038
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014487961460899333,
+      "loss": 0.2522,
+      "step": 6039
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014486322964394827,
+      "loss": 0.1665,
+      "step": 6040
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014484684317083127,
+      "loss": 0.1402,
+      "step": 6041
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014483045519019314,
+      "loss": 0.1743,
+      "step": 6042
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00014481406570258476,
+      "loss": 0.2429,
+      "step": 6043
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001447976747085571,
+      "loss": 0.2341,
+      "step": 6044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014478128220866108,
+      "loss": 0.1355,
+      "step": 6045
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014476488820344783,
+      "loss": 0.152,
+      "step": 6046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014474849269346832,
+      "loss": 0.0878,
+      "step": 6047
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014473209567927373,
+      "loss": 0.1175,
+      "step": 6048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014471569716141525,
+      "loss": 0.1949,
+      "step": 6049
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014469929714044412,
+      "loss": 0.2282,
+      "step": 6050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014468289561691162,
+      "loss": 0.1851,
+      "step": 6051
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001446664925913691,
+      "loss": 0.3061,
+      "step": 6052
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014465008806436786,
+      "loss": 0.3132,
+      "step": 6053
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014463368203645945,
+      "loss": 0.106,
+      "step": 6054
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014461727450819527,
+      "loss": 0.2694,
+      "step": 6055
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014460086548012693,
+      "loss": 0.2528,
+      "step": 6056
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014458445495280593,
+      "loss": 0.2096,
+      "step": 6057
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014456804292678404,
+      "loss": 0.219,
+      "step": 6058
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014455162940261285,
+      "loss": 0.2183,
+      "step": 6059
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014453521438084405,
+      "loss": 0.2113,
+      "step": 6060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014451879786202957,
+      "loss": 0.1448,
+      "step": 6061
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014450237984672113,
+      "loss": 0.1896,
+      "step": 6062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014448596033547067,
+      "loss": 0.2963,
+      "step": 6063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014446953932883015,
+      "loss": 0.1415,
+      "step": 6064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001444531168273515,
+      "loss": 0.2599,
+      "step": 6065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014443669283158682,
+      "loss": 0.158,
+      "step": 6066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014442026734208817,
+      "loss": 0.1234,
+      "step": 6067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001444038403594077,
+      "loss": 0.0889,
+      "step": 6068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001443874118840976,
+      "loss": 0.3436,
+      "step": 6069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001443709819167101,
+      "loss": 0.2221,
+      "step": 6070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001443545504577975,
+      "loss": 0.136,
+      "step": 6071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014433811750791216,
+      "loss": 0.1284,
+      "step": 6072
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014432168306760643,
+      "loss": 0.2856,
+      "step": 6073
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014430524713743276,
+      "loss": 0.1368,
+      "step": 6074
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001442888097179437,
+      "loss": 0.0842,
+      "step": 6075
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001442723708096917,
+      "loss": 0.1697,
+      "step": 6076
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014425593041322943,
+      "loss": 0.0865,
+      "step": 6077
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001442394885291095,
+      "loss": 0.1686,
+      "step": 6078
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001442230451578846,
+      "loss": 0.1276,
+      "step": 6079
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001442066003001075,
+      "loss": 0.1381,
+      "step": 6080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014419015395633093,
+      "loss": 0.1658,
+      "step": 6081
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014417370612710778,
+      "loss": 0.2773,
+      "step": 6082
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014415725681299094,
+      "loss": 0.2283,
+      "step": 6083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014414080601453334,
+      "loss": 0.2483,
+      "step": 6084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014412435373228799,
+      "loss": 0.313,
+      "step": 6085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001441078999668079,
+      "loss": 0.1688,
+      "step": 6086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014409144471864619,
+      "loss": 0.3464,
+      "step": 6087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014407498798835595,
+      "loss": 0.1317,
+      "step": 6088
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014405852977649044,
+      "loss": 0.1784,
+      "step": 6089
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014404207008360284,
+      "loss": 0.1548,
+      "step": 6090
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014402560891024652,
+      "loss": 0.2433,
+      "step": 6091
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014400914625697478,
+      "loss": 0.1297,
+      "step": 6092
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014399268212434094,
+      "loss": 0.1735,
+      "step": 6093
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014397621651289852,
+      "loss": 0.1406,
+      "step": 6094
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000143959749423201,
+      "loss": 0.2254,
+      "step": 6095
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001439432808558019,
+      "loss": 0.1651,
+      "step": 6096
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014392681081125486,
+      "loss": 0.1128,
+      "step": 6097
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014391033929011342,
+      "loss": 0.175,
+      "step": 6098
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014389386629293135,
+      "loss": 0.1216,
+      "step": 6099
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014387739182026237,
+      "loss": 0.2081,
+      "step": 6100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00014386091587266026,
+      "loss": 0.2274,
+      "step": 6101
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014384443845067883,
+      "loss": 0.1954,
+      "step": 6102
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014382795955487202,
+      "loss": 0.1363,
+      "step": 6103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014381147918579376,
+      "loss": 0.0979,
+      "step": 6104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014379499734399798,
+      "loss": 0.101,
+      "step": 6105
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014377851403003878,
+      "loss": 0.2276,
+      "step": 6106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014376202924447015,
+      "loss": 0.1473,
+      "step": 6107
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014374554298784634,
+      "loss": 0.146,
+      "step": 6108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014372905526072147,
+      "loss": 0.2298,
+      "step": 6109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001437125660636498,
+      "loss": 0.1512,
+      "step": 6110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001436960753971856,
+      "loss": 0.1879,
+      "step": 6111
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014367958326188317,
+      "loss": 0.1779,
+      "step": 6112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014366308965829696,
+      "loss": 0.1386,
+      "step": 6113
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014364659458698134,
+      "loss": 0.1423,
+      "step": 6114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001436300980484908,
+      "loss": 0.1269,
+      "step": 6115
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001436136000433799,
+      "loss": 0.2327,
+      "step": 6116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014359710057220318,
+      "loss": 0.2816,
+      "step": 6117
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014358059963551532,
+      "loss": 0.0976,
+      "step": 6118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001435640972338709,
+      "loss": 0.1771,
+      "step": 6119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001435475933678248,
+      "loss": 0.2497,
+      "step": 6120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014353108803793162,
+      "loss": 0.1369,
+      "step": 6121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001435145812447463,
+      "loss": 0.1391,
+      "step": 6122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001434980729888237,
+      "loss": 0.1998,
+      "step": 6123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014348156327071871,
+      "loss": 0.2011,
+      "step": 6124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014346505209098632,
+      "loss": 0.153,
+      "step": 6125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014344853945018156,
+      "loss": 0.1237,
+      "step": 6126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014343202534885946,
+      "loss": 0.261,
+      "step": 6127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001434155097875752,
+      "loss": 0.0984,
+      "step": 6128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001433989927668839,
+      "loss": 0.2293,
+      "step": 6129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014338247428734082,
+      "loss": 0.2082,
+      "step": 6130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014336595434950118,
+      "loss": 0.1972,
+      "step": 6131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014334943295392033,
+      "loss": 0.1097,
+      "step": 6132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014333291010115362,
+      "loss": 0.1807,
+      "step": 6133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014331638579175645,
+      "loss": 0.229,
+      "step": 6134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014329986002628432,
+      "loss": 0.2146,
+      "step": 6135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014328333280529273,
+      "loss": 0.1698,
+      "step": 6136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001432668041293372,
+      "loss": 0.266,
+      "step": 6137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014325027399897337,
+      "loss": 0.1919,
+      "step": 6138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001432337424147569,
+      "loss": 0.2402,
+      "step": 6139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014321720937724352,
+      "loss": 0.4199,
+      "step": 6140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014320067488698897,
+      "loss": 0.1518,
+      "step": 6141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000143184138944549,
+      "loss": 0.1712,
+      "step": 6142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014316760155047955,
+      "loss": 0.1288,
+      "step": 6143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014315106270533647,
+      "loss": 0.1463,
+      "step": 6144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014313452240967574,
+      "loss": 0.083,
+      "step": 6145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014311798066405333,
+      "loss": 0.3014,
+      "step": 6146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014310143746902528,
+      "loss": 0.2068,
+      "step": 6147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014308489282514774,
+      "loss": 0.0999,
+      "step": 6148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014306834673297682,
+      "loss": 0.1794,
+      "step": 6149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014305179919306873,
+      "loss": 0.2056,
+      "step": 6150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001430352502059797,
+      "loss": 0.1706,
+      "step": 6151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000143018699772266,
+      "loss": 0.2093,
+      "step": 6152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014300214789248402,
+      "loss": 0.4479,
+      "step": 6153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014298559456719012,
+      "loss": 0.1813,
+      "step": 6154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001429690397969408,
+      "loss": 0.1937,
+      "step": 6155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001429524835822924,
+      "loss": 0.1036,
+      "step": 6156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001429359259238016,
+      "loss": 0.1262,
+      "step": 6157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00014291936682202494,
+      "loss": 0.2672,
+      "step": 6158
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000142902806277519,
+      "loss": 0.1173,
+      "step": 6159
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014288624429084054,
+      "loss": 0.2278,
+      "step": 6160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014286968086254622,
+      "loss": 0.2021,
+      "step": 6161
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001428531159931929,
+      "loss": 0.1578,
+      "step": 6162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014283654968333734,
+      "loss": 0.1683,
+      "step": 6163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014281998193353638,
+      "loss": 0.1916,
+      "step": 6164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014280341274434705,
+      "loss": 0.1996,
+      "step": 6165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014278684211632621,
+      "loss": 0.0768,
+      "step": 6166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014277027005003102,
+      "loss": 0.4681,
+      "step": 6167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001427536965460184,
+      "loss": 0.3539,
+      "step": 6168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014273712160484554,
+      "loss": 0.2549,
+      "step": 6169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001427205452270696,
+      "loss": 0.153,
+      "step": 6170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014270396741324776,
+      "loss": 0.1518,
+      "step": 6171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014268738816393736,
+      "loss": 0.2656,
+      "step": 6172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014267080747969562,
+      "loss": 0.1111,
+      "step": 6173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014265422536107993,
+      "loss": 0.1677,
+      "step": 6174
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014263764180864774,
+      "loss": 0.257,
+      "step": 6175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014262105682295642,
+      "loss": 0.2127,
+      "step": 6176
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014260447040456353,
+      "loss": 0.1356,
+      "step": 6177
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001425878825540266,
+      "loss": 0.0925,
+      "step": 6178
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014257129327190326,
+      "loss": 0.4354,
+      "step": 6179
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001425547025587511,
+      "loss": 0.1255,
+      "step": 6180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001425381104151279,
+      "loss": 0.1851,
+      "step": 6181
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001425215168415913,
+      "loss": 0.1228,
+      "step": 6182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014250492183869913,
+      "loss": 0.1366,
+      "step": 6183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014248832540700928,
+      "loss": 0.1139,
+      "step": 6184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014247172754707957,
+      "loss": 0.2247,
+      "step": 6185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014245512825946798,
+      "loss": 0.1314,
+      "step": 6186
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014243852754473245,
+      "loss": 0.206,
+      "step": 6187
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001424219254034311,
+      "loss": 0.1238,
+      "step": 6188
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014240532183612187,
+      "loss": 0.1899,
+      "step": 6189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014238871684336298,
+      "loss": 0.1676,
+      "step": 6190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001423721104257126,
+      "loss": 0.2477,
+      "step": 6191
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014235550258372894,
+      "loss": 0.1523,
+      "step": 6192
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001423388933179703,
+      "loss": 0.21,
+      "step": 6193
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014232228262899495,
+      "loss": 0.225,
+      "step": 6194
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014230567051736124,
+      "loss": 0.4526,
+      "step": 6195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014228905698362762,
+      "loss": 0.1284,
+      "step": 6196
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014227244202835257,
+      "loss": 0.1858,
+      "step": 6197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014225582565209459,
+      "loss": 0.1393,
+      "step": 6198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001422392078554122,
+      "loss": 0.2687,
+      "step": 6199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014222258863886406,
+      "loss": 0.2643,
+      "step": 6200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014220596800300882,
+      "loss": 0.2078,
+      "step": 6201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014218934594840513,
+      "loss": 0.1273,
+      "step": 6202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014217272247561173,
+      "loss": 0.1793,
+      "step": 6203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014215609758518746,
+      "loss": 0.2849,
+      "step": 6204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001421394712776912,
+      "loss": 0.1327,
+      "step": 6205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014212284355368172,
+      "loss": 0.2272,
+      "step": 6206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014210621441371811,
+      "loss": 0.1134,
+      "step": 6207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014208958385835926,
+      "loss": 0.2518,
+      "step": 6208
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014207295188816424,
+      "loss": 0.141,
+      "step": 6209
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014205631850369206,
+      "loss": 0.1691,
+      "step": 6210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014203968370550194,
+      "loss": 0.1759,
+      "step": 6211
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014202304749415303,
+      "loss": 0.2015,
+      "step": 6212
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014200640987020453,
+      "loss": 0.1536,
+      "step": 6213
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014198977083421577,
+      "loss": 0.2531,
+      "step": 6214
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000141973130386746,
+      "loss": 0.1691,
+      "step": 6215
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001419564885283546,
+      "loss": 0.1472,
+      "step": 6216
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014193984525960097,
+      "loss": 0.0899,
+      "step": 6217
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014192320058104466,
+      "loss": 0.145,
+      "step": 6218
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001419065544932451,
+      "loss": 0.1459,
+      "step": 6219
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014188990699676184,
+      "loss": 0.1835,
+      "step": 6220
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014187325809215454,
+      "loss": 0.2108,
+      "step": 6221
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014185660777998283,
+      "loss": 0.1441,
+      "step": 6222
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014183995606080636,
+      "loss": 0.187,
+      "step": 6223
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014182330293518493,
+      "loss": 0.1722,
+      "step": 6224
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001418066484036783,
+      "loss": 0.3168,
+      "step": 6225
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014178999246684636,
+      "loss": 0.2687,
+      "step": 6226
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014177333512524896,
+      "loss": 0.1853,
+      "step": 6227
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000141756676379446,
+      "loss": 0.1141,
+      "step": 6228
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001417400162299975,
+      "loss": 0.3532,
+      "step": 6229
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014172335467746352,
+      "loss": 0.2227,
+      "step": 6230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001417066917224041,
+      "loss": 0.0799,
+      "step": 6231
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014169002736537934,
+      "loss": 0.1962,
+      "step": 6232
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014167336160694947,
+      "loss": 0.1885,
+      "step": 6233
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014165669444767467,
+      "loss": 0.2803,
+      "step": 6234
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001416400258881152,
+      "loss": 0.2156,
+      "step": 6235
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014162335592883137,
+      "loss": 0.1526,
+      "step": 6236
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014160668457038352,
+      "loss": 0.1451,
+      "step": 6237
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014159001181333216,
+      "loss": 0.1218,
+      "step": 6238
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014157333765823764,
+      "loss": 0.1895,
+      "step": 6239
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014155666210566048,
+      "loss": 0.1349,
+      "step": 6240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014153998515616123,
+      "loss": 0.5869,
+      "step": 6241
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001415233068103005,
+      "loss": 0.1733,
+      "step": 6242
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001415066270686389,
+      "loss": 0.1101,
+      "step": 6243
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014148994593173714,
+      "loss": 0.1238,
+      "step": 6244
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014147326340015597,
+      "loss": 0.1387,
+      "step": 6245
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014145657947445614,
+      "loss": 0.3023,
+      "step": 6246
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001414398941551985,
+      "loss": 0.2157,
+      "step": 6247
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014142320744294388,
+      "loss": 0.0898,
+      "step": 6248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014140651933825326,
+      "loss": 0.1243,
+      "step": 6249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001413898298416876,
+      "loss": 0.3579,
+      "step": 6250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014137313895380785,
+      "loss": 0.1528,
+      "step": 6251
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001413564466751752,
+      "loss": 0.2194,
+      "step": 6252
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014133975300635062,
+      "loss": 0.2244,
+      "step": 6253
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014132305794789534,
+      "loss": 0.1208,
+      "step": 6254
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014130636150037055,
+      "loss": 0.2004,
+      "step": 6255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014128966366433752,
+      "loss": 0.1908,
+      "step": 6256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001412729644403575,
+      "loss": 0.1909,
+      "step": 6257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014125626382899183,
+      "loss": 0.0851,
+      "step": 6258
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000141239561830802,
+      "loss": 0.2156,
+      "step": 6259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001412228584463493,
+      "loss": 0.2633,
+      "step": 6260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014120615367619533,
+      "loss": 0.1373,
+      "step": 6261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014118944752090155,
+      "loss": 0.3196,
+      "step": 6262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014117273998102959,
+      "loss": 0.1998,
+      "step": 6263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014115603105714096,
+      "loss": 0.1544,
+      "step": 6264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014113932074979747,
+      "loss": 0.4805,
+      "step": 6265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001411226090595608,
+      "loss": 0.2195,
+      "step": 6266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014110589598699262,
+      "loss": 0.1599,
+      "step": 6267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014108918153265485,
+      "loss": 0.1463,
+      "step": 6268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014107246569710925,
+      "loss": 0.1164,
+      "step": 6269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001410557484809178,
+      "loss": 0.1362,
+      "step": 6270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001410390298846424,
+      "loss": 0.1939,
+      "step": 6271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014102230990884507,
+      "loss": 0.1278,
+      "step": 6272
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014100558855408783,
+      "loss": 0.2479,
+      "step": 6273
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014098886582093276,
+      "loss": 0.1942,
+      "step": 6274
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014097214170994202,
+      "loss": 0.2125,
+      "step": 6275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001409554162216778,
+      "loss": 0.1704,
+      "step": 6276
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014093868935670227,
+      "loss": 0.1915,
+      "step": 6277
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014092196111557774,
+      "loss": 0.1437,
+      "step": 6278
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001409052314988665,
+      "loss": 0.1184,
+      "step": 6279
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014088850050713095,
+      "loss": 0.1342,
+      "step": 6280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014087176814093347,
+      "loss": 0.1556,
+      "step": 6281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001408550344008366,
+      "loss": 0.2451,
+      "step": 6282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014083829928740268,
+      "loss": 0.1314,
+      "step": 6283
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001408215628011944,
+      "loss": 0.2094,
+      "step": 6284
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014080482494277432,
+      "loss": 0.1415,
+      "step": 6285
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014078808571270504,
+      "loss": 0.111,
+      "step": 6286
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014077134511154927,
+      "loss": 0.252,
+      "step": 6287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014075460313986974,
+      "loss": 0.1264,
+      "step": 6288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001407378597982293,
+      "loss": 0.2114,
+      "step": 6289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014072111508719063,
+      "loss": 0.329,
+      "step": 6290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014070436900731674,
+      "loss": 0.2385,
+      "step": 6291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001406876215591705,
+      "loss": 0.2107,
+      "step": 6292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014067087274331484,
+      "loss": 0.1773,
+      "step": 6293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001406541225603128,
+      "loss": 0.0859,
+      "step": 6294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001406373710107274,
+      "loss": 0.1779,
+      "step": 6295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014062061809512182,
+      "loss": 0.1565,
+      "step": 6296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014060386381405915,
+      "loss": 0.1178,
+      "step": 6297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014058710816810255,
+      "loss": 0.1291,
+      "step": 6298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014057035115781535,
+      "loss": 0.1236,
+      "step": 6299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014055359278376077,
+      "loss": 0.161,
+      "step": 6300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014053683304650214,
+      "loss": 0.1529,
+      "step": 6301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014052007194660285,
+      "loss": 0.21,
+      "step": 6302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014050330948462637,
+      "loss": 0.2344,
+      "step": 6303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001404865456611361,
+      "loss": 0.1388,
+      "step": 6304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014046978047669553,
+      "loss": 0.1139,
+      "step": 6305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014045301393186835,
+      "loss": 0.2202,
+      "step": 6306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014043624602721802,
+      "loss": 0.1457,
+      "step": 6307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014041947676330827,
+      "loss": 0.2216,
+      "step": 6308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014040270614070277,
+      "loss": 0.1525,
+      "step": 6309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001403859341599653,
+      "loss": 0.1269,
+      "step": 6310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014036916082165962,
+      "loss": 0.1094,
+      "step": 6311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001403523861263495,
+      "loss": 0.2816,
+      "step": 6312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014033561007459894,
+      "loss": 0.2269,
+      "step": 6313
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001403188326669718,
+      "loss": 0.1513,
+      "step": 6314
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014030205390403206,
+      "loss": 0.138,
+      "step": 6315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014028527378634372,
+      "loss": 0.1258,
+      "step": 6316
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014026849231447088,
+      "loss": 0.1476,
+      "step": 6317
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014025170948897761,
+      "loss": 0.2633,
+      "step": 6318
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001402349253104281,
+      "loss": 0.1606,
+      "step": 6319
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014021813977938648,
+      "loss": 0.1792,
+      "step": 6320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014020135289641704,
+      "loss": 0.2218,
+      "step": 6321
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014018456466208408,
+      "loss": 0.2987,
+      "step": 6322
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014016777507695196,
+      "loss": 0.2494,
+      "step": 6323
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014015098414158498,
+      "loss": 0.2237,
+      "step": 6324
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014013419185654761,
+      "loss": 0.3442,
+      "step": 6325
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001401173982224043,
+      "loss": 0.2982,
+      "step": 6326
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014010060323971964,
+      "loss": 0.2189,
+      "step": 6327
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014008380690905806,
+      "loss": 0.2079,
+      "step": 6328
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001400670092309843,
+      "loss": 0.2126,
+      "step": 6329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001400502102060629,
+      "loss": 0.2225,
+      "step": 6330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014003340983485867,
+      "loss": 0.116,
+      "step": 6331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014001660811793624,
+      "loss": 0.1309,
+      "step": 6332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001399998050558605,
+      "loss": 0.1792,
+      "step": 6333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013998300064919618,
+      "loss": 0.2135,
+      "step": 6334
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013996619489850822,
+      "loss": 0.156,
+      "step": 6335
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013994938780436157,
+      "loss": 0.2225,
+      "step": 6336
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013993257936732113,
+      "loss": 0.1906,
+      "step": 6337
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013991576958795196,
+      "loss": 0.1932,
+      "step": 6338
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001398989584668191,
+      "loss": 0.1832,
+      "step": 6339
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013988214600448766,
+      "loss": 0.2077,
+      "step": 6340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013986533220152278,
+      "loss": 0.2371,
+      "step": 6341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013984851705848965,
+      "loss": 0.2258,
+      "step": 6342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013983170057595357,
+      "loss": 0.131,
+      "step": 6343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013981488275447974,
+      "loss": 0.1925,
+      "step": 6344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001397980635946335,
+      "loss": 0.1824,
+      "step": 6345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001397812430969803,
+      "loss": 0.1146,
+      "step": 6346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013976442126208545,
+      "loss": 0.2275,
+      "step": 6347
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001397475980905145,
+      "loss": 0.1554,
+      "step": 6348
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013973077358283293,
+      "loss": 0.0844,
+      "step": 6349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013971394773960627,
+      "loss": 0.1783,
+      "step": 6350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013969712056140015,
+      "loss": 0.1151,
+      "step": 6351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013968029204878022,
+      "loss": 0.3345,
+      "step": 6352
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013966346220231214,
+      "loss": 0.1829,
+      "step": 6353
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013964663102256165,
+      "loss": 0.2255,
+      "step": 6354
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013962979851009455,
+      "loss": 0.1843,
+      "step": 6355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013961296466547665,
+      "loss": 0.1426,
+      "step": 6356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013959612948927382,
+      "loss": 0.2905,
+      "step": 6357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013957929298205195,
+      "loss": 0.1769,
+      "step": 6358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000139562455144377,
+      "loss": 0.3054,
+      "step": 6359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013954561597681503,
+      "loss": 0.3317,
+      "step": 6360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000139528775479932,
+      "loss": 0.089,
+      "step": 6361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013951193365429408,
+      "loss": 0.1933,
+      "step": 6362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013949509050046734,
+      "loss": 0.125,
+      "step": 6363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000139478246019018,
+      "loss": 0.1409,
+      "step": 6364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013946140021051226,
+      "loss": 0.2733,
+      "step": 6365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013944455307551642,
+      "loss": 0.1111,
+      "step": 6366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001394277046145968,
+      "loss": 0.1702,
+      "step": 6367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013941085482831972,
+      "loss": 0.1842,
+      "step": 6368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001393940037172516,
+      "loss": 0.1403,
+      "step": 6369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013937715128195887,
+      "loss": 0.247,
+      "step": 6370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013936029752300807,
+      "loss": 0.2319,
+      "step": 6371
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001393434424409657,
+      "loss": 0.2292,
+      "step": 6372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013932658603639834,
+      "loss": 0.165,
+      "step": 6373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013930972830987265,
+      "loss": 0.1482,
+      "step": 6374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013929286926195528,
+      "loss": 0.1355,
+      "step": 6375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013927600889321293,
+      "loss": 0.1569,
+      "step": 6376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001392591472042124,
+      "loss": 0.1577,
+      "step": 6377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013924228419552042,
+      "loss": 0.349,
+      "step": 6378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013922541986770387,
+      "loss": 0.2648,
+      "step": 6379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013920855422132968,
+      "loss": 0.186,
+      "step": 6380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001391916872569648,
+      "loss": 0.2022,
+      "step": 6381
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013917481897517613,
+      "loss": 0.0978,
+      "step": 6382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013915794937653076,
+      "loss": 0.249,
+      "step": 6383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013914107846159572,
+      "loss": 0.2532,
+      "step": 6384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013912420623093813,
+      "loss": 0.1561,
+      "step": 6385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00013910733268512518,
+      "loss": 0.1104,
+      "step": 6386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013909045782472404,
+      "loss": 0.1738,
+      "step": 6387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013907358165030198,
+      "loss": 0.1241,
+      "step": 6388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001390567041624263,
+      "loss": 0.169,
+      "step": 6389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013903982536166428,
+      "loss": 0.109,
+      "step": 6390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013902294524858336,
+      "loss": 0.1328,
+      "step": 6391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013900606382375087,
+      "loss": 0.2456,
+      "step": 6392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001389891810877344,
+      "loss": 0.1512,
+      "step": 6393
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001389722970411014,
+      "loss": 0.1452,
+      "step": 6394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001389554116844194,
+      "loss": 0.2048,
+      "step": 6395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013893852501825607,
+      "loss": 0.2891,
+      "step": 6396
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013892163704317894,
+      "loss": 0.0854,
+      "step": 6397
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013890474775975583,
+      "loss": 0.1137,
+      "step": 6398
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001388878571685544,
+      "loss": 0.1367,
+      "step": 6399
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013887096527014244,
+      "loss": 0.2865,
+      "step": 6400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013885407206508773,
+      "loss": 0.1654,
+      "step": 6401
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001388371775539582,
+      "loss": 0.1396,
+      "step": 6402
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013882028173732168,
+      "loss": 0.2113,
+      "step": 6403
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001388033846157462,
+      "loss": 0.1538,
+      "step": 6404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001387864861897997,
+      "loss": 0.1641,
+      "step": 6405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013876958646005024,
+      "loss": 0.1137,
+      "step": 6406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013875268542706595,
+      "loss": 0.1664,
+      "step": 6407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013873578309141484,
+      "loss": 0.1182,
+      "step": 6408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013871887945366516,
+      "loss": 0.1526,
+      "step": 6409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013870197451438512,
+      "loss": 0.2359,
+      "step": 6410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013868506827414295,
+      "loss": 0.098,
+      "step": 6411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013866816073350698,
+      "loss": 0.306,
+      "step": 6412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013865125189304554,
+      "loss": 0.1814,
+      "step": 6413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013863434175332703,
+      "loss": 0.0881,
+      "step": 6414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013861743031491988,
+      "loss": 0.2243,
+      "step": 6415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001386005175783925,
+      "loss": 0.1634,
+      "step": 6416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013858360354431355,
+      "loss": 0.1111,
+      "step": 6417
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013856668821325148,
+      "loss": 0.1284,
+      "step": 6418
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013854977158577492,
+      "loss": 0.1466,
+      "step": 6419
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013853285366245254,
+      "loss": 0.2167,
+      "step": 6420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000138515934443853,
+      "loss": 0.243,
+      "step": 6421
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001384990139305451,
+      "loss": 0.2811,
+      "step": 6422
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013848209212309755,
+      "loss": 0.177,
+      "step": 6423
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013846516902207923,
+      "loss": 0.2377,
+      "step": 6424
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013844824462805897,
+      "loss": 0.1455,
+      "step": 6425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013843131894160571,
+      "loss": 0.1377,
+      "step": 6426
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013841439196328836,
+      "loss": 0.1425,
+      "step": 6427
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013839746369367597,
+      "loss": 0.1159,
+      "step": 6428
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013838053413333756,
+      "loss": 0.2114,
+      "step": 6429
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013836360328284218,
+      "loss": 0.0982,
+      "step": 6430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013834667114275906,
+      "loss": 0.2372,
+      "step": 6431
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013832973771365725,
+      "loss": 0.1697,
+      "step": 6432
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013831280299610605,
+      "loss": 0.2432,
+      "step": 6433
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013829586699067468,
+      "loss": 0.2549,
+      "step": 6434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013827892969793244,
+      "loss": 0.1515,
+      "step": 6435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013826199111844872,
+      "loss": 0.1273,
+      "step": 6436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001382450512527928,
+      "loss": 0.115,
+      "step": 6437
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013822811010153426,
+      "loss": 0.1479,
+      "step": 6438
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013821116766524248,
+      "loss": 0.1249,
+      "step": 6439
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013819422394448702,
+      "loss": 0.1901,
+      "step": 6440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001381772789398374,
+      "loss": 0.2999,
+      "step": 6441
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013816033265186321,
+      "loss": 0.1395,
+      "step": 6442
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00013814338508113418,
+      "loss": 0.1146,
+      "step": 6443
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001381264362282199,
+      "loss": 0.0883,
+      "step": 6444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001381094860936902,
+      "loss": 0.1844,
+      "step": 6445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013809253467811483,
+      "loss": 0.2221,
+      "step": 6446
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013807558198206357,
+      "loss": 0.1535,
+      "step": 6447
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013805862800610628,
+      "loss": 0.3022,
+      "step": 6448
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001380416727508129,
+      "loss": 0.2831,
+      "step": 6449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013802471621675338,
+      "loss": 0.2515,
+      "step": 6450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001380077584044977,
+      "loss": 0.3944,
+      "step": 6451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001379907993146159,
+      "loss": 0.2448,
+      "step": 6452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013797383894767804,
+      "loss": 0.2575,
+      "step": 6453
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013795687730425426,
+      "loss": 0.2148,
+      "step": 6454
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013793991438491472,
+      "loss": 0.1814,
+      "step": 6455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013792295019022957,
+      "loss": 0.0877,
+      "step": 6456
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001379059847207692,
+      "loss": 0.1486,
+      "step": 6457
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013788901797710375,
+      "loss": 0.108,
+      "step": 6458
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001378720499598037,
+      "loss": 0.1452,
+      "step": 6459
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001378550806694393,
+      "loss": 0.1615,
+      "step": 6460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013783811010658102,
+      "loss": 0.083,
+      "step": 6461
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001378211382717993,
+      "loss": 0.1872,
+      "step": 6462
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013780416516566467,
+      "loss": 0.1777,
+      "step": 6463
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001377871907887477,
+      "loss": 0.159,
+      "step": 6464
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013777021514161896,
+      "loss": 0.1882,
+      "step": 6465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001377532382248491,
+      "loss": 0.2421,
+      "step": 6466
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013773626003900876,
+      "loss": 0.0908,
+      "step": 6467
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013771928058466866,
+      "loss": 0.3021,
+      "step": 6468
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001377022998623996,
+      "loss": 0.1568,
+      "step": 6469
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013768531787277234,
+      "loss": 0.3691,
+      "step": 6470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013766833461635779,
+      "loss": 0.169,
+      "step": 6471
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013765135009372682,
+      "loss": 0.1094,
+      "step": 6472
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013763436430545034,
+      "loss": 0.1348,
+      "step": 6473
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013761737725209928,
+      "loss": 0.1279,
+      "step": 6474
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013760038893424477,
+      "loss": 0.1424,
+      "step": 6475
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013758339935245777,
+      "loss": 0.0812,
+      "step": 6476
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013756640850730943,
+      "loss": 0.1349,
+      "step": 6477
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013754941639937092,
+      "loss": 0.1599,
+      "step": 6478
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013753242302921335,
+      "loss": 0.3021,
+      "step": 6479
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013751542839740804,
+      "loss": 0.1738,
+      "step": 6480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013749843250452618,
+      "loss": 0.1451,
+      "step": 6481
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013748143535113915,
+      "loss": 0.1659,
+      "step": 6482
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013746443693781827,
+      "loss": 0.1402,
+      "step": 6483
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013744743726513495,
+      "loss": 0.1753,
+      "step": 6484
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013743043633366065,
+      "loss": 0.14,
+      "step": 6485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001374134341439668,
+      "loss": 0.1088,
+      "step": 6486
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000137396430696625,
+      "loss": 0.0941,
+      "step": 6487
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013737942599220676,
+      "loss": 0.1948,
+      "step": 6488
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001373624200312837,
+      "loss": 0.1622,
+      "step": 6489
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001373454128144275,
+      "loss": 0.2111,
+      "step": 6490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013732840434220987,
+      "loss": 0.1836,
+      "step": 6491
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013731139461520248,
+      "loss": 0.1635,
+      "step": 6492
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013729438363397714,
+      "loss": 0.1226,
+      "step": 6493
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013727737139910572,
+      "loss": 0.1501,
+      "step": 6494
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013726035791116004,
+      "loss": 0.3113,
+      "step": 6495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013724334317071198,
+      "loss": 0.2683,
+      "step": 6496
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013722632717833356,
+      "loss": 0.2837,
+      "step": 6497
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013720930993459668,
+      "loss": 0.1078,
+      "step": 6498
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013719229144007347,
+      "loss": 0.146,
+      "step": 6499
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013717527169533592,
+      "loss": 0.1802,
+      "step": 6500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00013715825070095621,
+      "loss": 0.2017,
+      "step": 6501
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013714122845750646,
+      "loss": 0.1379,
+      "step": 6502
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013712420496555887,
+      "loss": 0.1576,
+      "step": 6503
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013710718022568575,
+      "loss": 0.2455,
+      "step": 6504
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013709015423845927,
+      "loss": 0.4832,
+      "step": 6505
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013707312700445186,
+      "loss": 0.2752,
+      "step": 6506
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013705609852423582,
+      "loss": 0.082,
+      "step": 6507
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001370390687983836,
+      "loss": 0.0979,
+      "step": 6508
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001370220378274676,
+      "loss": 0.1947,
+      "step": 6509
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001370050056120604,
+      "loss": 0.2362,
+      "step": 6510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013698797215273446,
+      "loss": 0.1661,
+      "step": 6511
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013697093745006239,
+      "loss": 0.2229,
+      "step": 6512
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013695390150461682,
+      "loss": 0.1553,
+      "step": 6513
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001369368643169704,
+      "loss": 0.1286,
+      "step": 6514
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013691982588769582,
+      "loss": 0.2864,
+      "step": 6515
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013690278621736579,
+      "loss": 0.1967,
+      "step": 6516
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001368857453065532,
+      "loss": 0.2763,
+      "step": 6517
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001368687031558308,
+      "loss": 0.1672,
+      "step": 6518
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013685165976577146,
+      "loss": 0.1164,
+      "step": 6519
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013683461513694812,
+      "loss": 0.1455,
+      "step": 6520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013681756926993372,
+      "loss": 0.2053,
+      "step": 6521
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013680052216530125,
+      "loss": 0.2086,
+      "step": 6522
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013678347382362378,
+      "loss": 0.252,
+      "step": 6523
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013676642424547433,
+      "loss": 0.1325,
+      "step": 6524
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013674937343142608,
+      "loss": 0.164,
+      "step": 6525
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013673232138205213,
+      "loss": 0.3782,
+      "step": 6526
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013671526809792573,
+      "loss": 0.161,
+      "step": 6527
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013669821357962012,
+      "loss": 0.2494,
+      "step": 6528
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013668115782770857,
+      "loss": 0.1567,
+      "step": 6529
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013666410084276441,
+      "loss": 0.1042,
+      "step": 6530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000136647042625361,
+      "loss": 0.1573,
+      "step": 6531
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001366299831760718,
+      "loss": 0.1626,
+      "step": 6532
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013661292249547016,
+      "loss": 0.1388,
+      "step": 6533
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013659586058412966,
+      "loss": 0.1638,
+      "step": 6534
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013657879744262378,
+      "loss": 0.1458,
+      "step": 6535
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013656173307152617,
+      "loss": 0.1257,
+      "step": 6536
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013654466747141036,
+      "loss": 0.2188,
+      "step": 6537
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013652760064285008,
+      "loss": 0.1405,
+      "step": 6538
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013651053258641898,
+      "loss": 0.2731,
+      "step": 6539
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013649346330269082,
+      "loss": 0.1859,
+      "step": 6540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013647639279223937,
+      "loss": 0.1353,
+      "step": 6541
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013645932105563844,
+      "loss": 0.1157,
+      "step": 6542
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013644224809346197,
+      "loss": 0.1585,
+      "step": 6543
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001364251739062838,
+      "loss": 0.105,
+      "step": 6544
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013640809849467788,
+      "loss": 0.2374,
+      "step": 6545
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001363910218592182,
+      "loss": 0.1615,
+      "step": 6546
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001363739440004788,
+      "loss": 0.1368,
+      "step": 6547
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013635686491903374,
+      "loss": 0.2324,
+      "step": 6548
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013633978461545716,
+      "loss": 0.1056,
+      "step": 6549
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013632270309032316,
+      "loss": 0.3642,
+      "step": 6550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000136305620344206,
+      "loss": 0.0724,
+      "step": 6551
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013628853637767985,
+      "loss": 0.1259,
+      "step": 6552
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013627145119131902,
+      "loss": 0.1183,
+      "step": 6553
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001362543647856978,
+      "loss": 0.2194,
+      "step": 6554
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001362372771613906,
+      "loss": 0.1328,
+      "step": 6555
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013622018831897177,
+      "loss": 0.129,
+      "step": 6556
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013620309825901575,
+      "loss": 0.1756,
+      "step": 6557
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00013618600698209706,
+      "loss": 0.1032,
+      "step": 6558
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013616891448879016,
+      "loss": 0.1453,
+      "step": 6559
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013615182077966967,
+      "loss": 0.1086,
+      "step": 6560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013613472585531015,
+      "loss": 0.2524,
+      "step": 6561
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001361176297162863,
+      "loss": 0.1204,
+      "step": 6562
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013610053236317275,
+      "loss": 0.1404,
+      "step": 6563
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013608343379654424,
+      "loss": 0.1569,
+      "step": 6564
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013606633401697557,
+      "loss": 0.0656,
+      "step": 6565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013604923302504147,
+      "loss": 0.1328,
+      "step": 6566
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013603213082131687,
+      "loss": 0.1512,
+      "step": 6567
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013601502740637664,
+      "loss": 0.205,
+      "step": 6568
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013599792278079563,
+      "loss": 0.0883,
+      "step": 6569
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013598081694514894,
+      "loss": 0.1057,
+      "step": 6570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013596370990001145,
+      "loss": 0.1843,
+      "step": 6571
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013594660164595833,
+      "loss": 0.3454,
+      "step": 6572
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001359294921835646,
+      "loss": 0.1906,
+      "step": 6573
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001359123815134054,
+      "loss": 0.1161,
+      "step": 6574
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013589526963605594,
+      "loss": 0.1451,
+      "step": 6575
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013587815655209136,
+      "loss": 0.0949,
+      "step": 6576
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000135861042262087,
+      "loss": 0.0959,
+      "step": 6577
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013584392676661809,
+      "loss": 0.2659,
+      "step": 6578
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013582681006626,
+      "loss": 0.1839,
+      "step": 6579
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013580969216158808,
+      "loss": 0.1677,
+      "step": 6580
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013579257305317778,
+      "loss": 0.1204,
+      "step": 6581
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013577545274160455,
+      "loss": 0.2162,
+      "step": 6582
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013575833122744387,
+      "loss": 0.2352,
+      "step": 6583
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013574120851127125,
+      "loss": 0.162,
+      "step": 6584
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013572408459366233,
+      "loss": 0.1501,
+      "step": 6585
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001357069594751927,
+      "loss": 0.1769,
+      "step": 6586
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013568983315643803,
+      "loss": 0.1142,
+      "step": 6587
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013567270563797398,
+      "loss": 0.1858,
+      "step": 6588
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013565557692037634,
+      "loss": 0.2158,
+      "step": 6589
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013563844700422085,
+      "loss": 0.0922,
+      "step": 6590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013562131589008335,
+      "loss": 0.2385,
+      "step": 6591
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001356041835785397,
+      "loss": 0.1449,
+      "step": 6592
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013558705007016582,
+      "loss": 0.21,
+      "step": 6593
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013556991536553762,
+      "loss": 0.1198,
+      "step": 6594
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013555277946523107,
+      "loss": 0.1186,
+      "step": 6595
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001355356423698222,
+      "loss": 0.1028,
+      "step": 6596
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001355185040798871,
+      "loss": 0.1352,
+      "step": 6597
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013550136459600186,
+      "loss": 0.1786,
+      "step": 6598
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001354842239187426,
+      "loss": 0.0923,
+      "step": 6599
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013546708204868554,
+      "loss": 0.1079,
+      "step": 6600
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013544993898640688,
+      "loss": 0.0866,
+      "step": 6601
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013543279473248284,
+      "loss": 0.2283,
+      "step": 6602
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001354156492874898,
+      "loss": 0.2754,
+      "step": 6603
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013539850265200404,
+      "loss": 0.1176,
+      "step": 6604
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000135381354826602,
+      "loss": 0.1261,
+      "step": 6605
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013536420581186003,
+      "loss": 0.1293,
+      "step": 6606
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013534705560835466,
+      "loss": 0.1368,
+      "step": 6607
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013532990421666235,
+      "loss": 0.1489,
+      "step": 6608
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013531275163735965,
+      "loss": 0.2541,
+      "step": 6609
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013529559787102318,
+      "loss": 0.1233,
+      "step": 6610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013527844291822948,
+      "loss": 0.2073,
+      "step": 6611
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001352612867795553,
+      "loss": 0.136,
+      "step": 6612
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013524412945557727,
+      "loss": 0.1511,
+      "step": 6613
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00013522697094687217,
+      "loss": 0.2324,
+      "step": 6614
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001352098112540168,
+      "loss": 0.1429,
+      "step": 6615
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013519265037758796,
+      "loss": 0.1433,
+      "step": 6616
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013517548831816246,
+      "loss": 0.1981,
+      "step": 6617
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013515832507631726,
+      "loss": 0.0866,
+      "step": 6618
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013514116065262933,
+      "loss": 0.2386,
+      "step": 6619
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013512399504767556,
+      "loss": 0.1221,
+      "step": 6620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013510682826203303,
+      "loss": 0.1621,
+      "step": 6621
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013508966029627877,
+      "loss": 0.165,
+      "step": 6622
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001350724911509899,
+      "loss": 0.1454,
+      "step": 6623
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001350553208267436,
+      "loss": 0.2629,
+      "step": 6624
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013503814932411693,
+      "loss": 0.1261,
+      "step": 6625
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013502097664368722,
+      "loss": 0.0819,
+      "step": 6626
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013500380278603168,
+      "loss": 0.1198,
+      "step": 6627
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013498662775172763,
+      "loss": 0.1545,
+      "step": 6628
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013496945154135238,
+      "loss": 0.1096,
+      "step": 6629
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013495227415548331,
+      "loss": 0.4158,
+      "step": 6630
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013493509559469783,
+      "loss": 0.1182,
+      "step": 6631
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013491791585957342,
+      "loss": 0.1491,
+      "step": 6632
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013490073495068757,
+      "loss": 0.2682,
+      "step": 6633
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013488355286861783,
+      "loss": 0.0804,
+      "step": 6634
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001348663696139417,
+      "loss": 0.3955,
+      "step": 6635
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001348491851872369,
+      "loss": 0.1502,
+      "step": 6636
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013483199958908097,
+      "loss": 0.1061,
+      "step": 6637
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013481481282005172,
+      "loss": 0.1124,
+      "step": 6638
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013479762488072676,
+      "loss": 0.2267,
+      "step": 6639
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013478043577168394,
+      "loss": 0.2314,
+      "step": 6640
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013476324549350105,
+      "loss": 0.1068,
+      "step": 6641
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013474605404675594,
+      "loss": 0.2047,
+      "step": 6642
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001347288614320265,
+      "loss": 0.1311,
+      "step": 6643
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013471166764989062,
+      "loss": 0.1921,
+      "step": 6644
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013469447270092634,
+      "loss": 0.1884,
+      "step": 6645
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013467727658571162,
+      "loss": 0.2551,
+      "step": 6646
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013466007930482452,
+      "loss": 0.1657,
+      "step": 6647
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013464288085884312,
+      "loss": 0.3007,
+      "step": 6648
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001346256812483455,
+      "loss": 0.1585,
+      "step": 6649
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001346084804739099,
+      "loss": 0.2263,
+      "step": 6650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013459127853611446,
+      "loss": 0.1084,
+      "step": 6651
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013457407543553747,
+      "loss": 0.165,
+      "step": 6652
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001345568711727572,
+      "loss": 0.2473,
+      "step": 6653
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001345396657483519,
+      "loss": 0.069,
+      "step": 6654
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013452245916290004,
+      "loss": 0.2499,
+      "step": 6655
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013450525141697991,
+      "loss": 0.144,
+      "step": 6656
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013448804251117003,
+      "loss": 0.1099,
+      "step": 6657
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013447083244604887,
+      "loss": 0.1081,
+      "step": 6658
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013445362122219487,
+      "loss": 0.1883,
+      "step": 6659
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013443640884018665,
+      "loss": 0.1904,
+      "step": 6660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013441919530060277,
+      "loss": 0.3326,
+      "step": 6661
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001344019806040219,
+      "loss": 0.1362,
+      "step": 6662
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013438476475102263,
+      "loss": 0.1389,
+      "step": 6663
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013436754774218376,
+      "loss": 0.1578,
+      "step": 6664
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013435032957808402,
+      "loss": 0.2423,
+      "step": 6665
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013433311025930214,
+      "loss": 0.1391,
+      "step": 6666
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000134315889786417,
+      "loss": 0.2508,
+      "step": 6667
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013429866816000742,
+      "loss": 0.1342,
+      "step": 6668
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013428144538065235,
+      "loss": 0.0884,
+      "step": 6669
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013426422144893069,
+      "loss": 0.1552,
+      "step": 6670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013424699636542146,
+      "loss": 0.2105,
+      "step": 6671
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00013422977013070368,
+      "loss": 0.1785,
+      "step": 6672
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013421254274535632,
+      "loss": 0.2562,
+      "step": 6673
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001341953142099586,
+      "loss": 0.2014,
+      "step": 6674
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013417808452508958,
+      "loss": 0.1304,
+      "step": 6675
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013416085369132846,
+      "loss": 0.1196,
+      "step": 6676
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013414362170925442,
+      "loss": 0.1804,
+      "step": 6677
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001341263885794468,
+      "loss": 0.2784,
+      "step": 6678
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013410915430248476,
+      "loss": 0.1404,
+      "step": 6679
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001340919188789477,
+      "loss": 0.163,
+      "step": 6680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000134074682309415,
+      "loss": 0.2215,
+      "step": 6681
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013405744459446604,
+      "loss": 0.1141,
+      "step": 6682
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013404020573468025,
+      "loss": 0.0955,
+      "step": 6683
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013402296573063714,
+      "loss": 0.1283,
+      "step": 6684
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013400572458291627,
+      "loss": 0.2813,
+      "step": 6685
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013398848229209708,
+      "loss": 0.1363,
+      "step": 6686
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013397123885875925,
+      "loss": 0.0815,
+      "step": 6687
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013395399428348242,
+      "loss": 0.165,
+      "step": 6688
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013393674856684626,
+      "loss": 0.1131,
+      "step": 6689
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013391950170943046,
+      "loss": 0.1116,
+      "step": 6690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001339022537118148,
+      "loss": 0.1244,
+      "step": 6691
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013388500457457903,
+      "loss": 0.1837,
+      "step": 6692
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000133867754298303,
+      "loss": 0.0978,
+      "step": 6693
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013385050288356655,
+      "loss": 0.1317,
+      "step": 6694
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013383325033094963,
+      "loss": 0.1183,
+      "step": 6695
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013381599664103217,
+      "loss": 0.3493,
+      "step": 6696
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013379874181439417,
+      "loss": 0.2127,
+      "step": 6697
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013378148585161556,
+      "loss": 0.1476,
+      "step": 6698
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001337642287532765,
+      "loss": 0.1395,
+      "step": 6699
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000133746970519957,
+      "loss": 0.2553,
+      "step": 6700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013372971115223727,
+      "loss": 0.1414,
+      "step": 6701
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013371245065069745,
+      "loss": 0.1602,
+      "step": 6702
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013369518901591772,
+      "loss": 0.1207,
+      "step": 6703
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001336779262484784,
+      "loss": 0.2775,
+      "step": 6704
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013366066234895971,
+      "loss": 0.2172,
+      "step": 6705
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013364339731794197,
+      "loss": 0.2704,
+      "step": 6706
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001336261311560056,
+      "loss": 0.2005,
+      "step": 6707
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013360886386373094,
+      "loss": 0.1148,
+      "step": 6708
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013359159544169849,
+      "loss": 0.2424,
+      "step": 6709
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013357432589048866,
+      "loss": 0.1369,
+      "step": 6710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000133557055210682,
+      "loss": 0.3811,
+      "step": 6711
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013353978340285906,
+      "loss": 0.1396,
+      "step": 6712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001335225104676004,
+      "loss": 0.2327,
+      "step": 6713
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013350523640548673,
+      "loss": 0.177,
+      "step": 6714
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013348796121709862,
+      "loss": 0.2002,
+      "step": 6715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013347068490301682,
+      "loss": 0.1471,
+      "step": 6716
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013345340746382205,
+      "loss": 0.1048,
+      "step": 6717
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013343612890009515,
+      "loss": 0.1805,
+      "step": 6718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013341884921241683,
+      "loss": 0.1339,
+      "step": 6719
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013340156840136803,
+      "loss": 0.0745,
+      "step": 6720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013338428646752964,
+      "loss": 0.0793,
+      "step": 6721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001333670034114825,
+      "loss": 0.2259,
+      "step": 6722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001333497192338077,
+      "loss": 0.1476,
+      "step": 6723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001333324339350862,
+      "loss": 0.1186,
+      "step": 6724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013331514751589903,
+      "loss": 0.1955,
+      "step": 6725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001332978599768272,
+      "loss": 0.1468,
+      "step": 6726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013328057131845196,
+      "loss": 0.1789,
+      "step": 6727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001332632815413544,
+      "loss": 0.5303,
+      "step": 6728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00013324599064611575,
+      "loss": 0.132,
+      "step": 6729
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001332286986333172,
+      "loss": 0.0989,
+      "step": 6730
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013321140550354,
+      "loss": 0.1775,
+      "step": 6731
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013319411125736552,
+      "loss": 0.2442,
+      "step": 6732
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013317681589537504,
+      "loss": 0.1735,
+      "step": 6733
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013315951941815,
+      "loss": 0.1201,
+      "step": 6734
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013314222182627178,
+      "loss": 0.1007,
+      "step": 6735
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013312492312032186,
+      "loss": 0.1956,
+      "step": 6736
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013310762330088173,
+      "loss": 0.0881,
+      "step": 6737
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001330903223685329,
+      "loss": 0.3898,
+      "step": 6738
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013307302032385695,
+      "loss": 0.1518,
+      "step": 6739
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001330557171674355,
+      "loss": 0.1312,
+      "step": 6740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013303841289985017,
+      "loss": 0.2093,
+      "step": 6741
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013302110752168265,
+      "loss": 0.1758,
+      "step": 6742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013300380103351472,
+      "loss": 0.1401,
+      "step": 6743
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013298649343592803,
+      "loss": 0.2111,
+      "step": 6744
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001329691847295044,
+      "loss": 0.0751,
+      "step": 6745
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013295187491482572,
+      "loss": 0.174,
+      "step": 6746
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001329345639924738,
+      "loss": 0.2583,
+      "step": 6747
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013291725196303052,
+      "loss": 0.1377,
+      "step": 6748
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013289993882707797,
+      "loss": 0.0931,
+      "step": 6749
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013288262458519793,
+      "loss": 0.1443,
+      "step": 6750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013286530923797255,
+      "loss": 0.1893,
+      "step": 6751
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001328479927859838,
+      "loss": 0.102,
+      "step": 6752
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013283067522981386,
+      "loss": 0.134,
+      "step": 6753
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013281335657004478,
+      "loss": 0.2378,
+      "step": 6754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013279603680725876,
+      "loss": 0.2549,
+      "step": 6755
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013277871594203803,
+      "loss": 0.1015,
+      "step": 6756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013276139397496475,
+      "loss": 0.1532,
+      "step": 6757
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013274407090662127,
+      "loss": 0.1236,
+      "step": 6758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013272674673758984,
+      "loss": 0.2436,
+      "step": 6759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013270942146845286,
+      "loss": 0.1957,
+      "step": 6760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013269209509979272,
+      "loss": 0.1818,
+      "step": 6761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013267476763219183,
+      "loss": 0.1654,
+      "step": 6762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001326574390662326,
+      "loss": 0.1455,
+      "step": 6763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001326401094024976,
+      "loss": 0.3126,
+      "step": 6764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013262277864156934,
+      "loss": 0.1802,
+      "step": 6765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013260544678403034,
+      "loss": 0.1587,
+      "step": 6766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001325881138304633,
+      "loss": 0.1461,
+      "step": 6767
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013257077978145088,
+      "loss": 0.1152,
+      "step": 6768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001325534446375756,
+      "loss": 0.2064,
+      "step": 6769
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013253610839942035,
+      "loss": 0.1537,
+      "step": 6770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013251877106756777,
+      "loss": 0.1231,
+      "step": 6771
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013250143264260074,
+      "loss": 0.1468,
+      "step": 6772
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013248409312510202,
+      "loss": 0.266,
+      "step": 6773
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013246675251565447,
+      "loss": 0.1727,
+      "step": 6774
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013244941081484113,
+      "loss": 0.1829,
+      "step": 6775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013243206802324476,
+      "loss": 0.1396,
+      "step": 6776
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013241472414144843,
+      "loss": 0.0904,
+      "step": 6777
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013239737917003513,
+      "loss": 0.1379,
+      "step": 6778
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013238003310958794,
+      "loss": 0.1643,
+      "step": 6779
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013236268596068988,
+      "loss": 0.1856,
+      "step": 6780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013234533772392415,
+      "loss": 0.094,
+      "step": 6781
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013232798839987384,
+      "loss": 0.1676,
+      "step": 6782
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00013231063798912218,
+      "loss": 0.2173,
+      "step": 6783
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001322932864922524,
+      "loss": 0.1982,
+      "step": 6784
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001322759339098478,
+      "loss": 0.1738,
+      "step": 6785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001322585802424916,
+      "loss": 0.2315,
+      "step": 6786
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001322412254907672,
+      "loss": 0.0937,
+      "step": 6787
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000132223869655258,
+      "loss": 0.1569,
+      "step": 6788
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013220651273654737,
+      "loss": 0.3925,
+      "step": 6789
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013218915473521876,
+      "loss": 0.1086,
+      "step": 6790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001321717956518557,
+      "loss": 0.1786,
+      "step": 6791
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013215443548704166,
+      "loss": 0.1261,
+      "step": 6792
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013213707424136026,
+      "loss": 0.2015,
+      "step": 6793
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013211971191539503,
+      "loss": 0.3031,
+      "step": 6794
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013210234850972964,
+      "loss": 0.083,
+      "step": 6795
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013208498402494777,
+      "loss": 0.2117,
+      "step": 6796
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013206761846163307,
+      "loss": 0.0901,
+      "step": 6797
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013205025182036936,
+      "loss": 0.1586,
+      "step": 6798
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013203288410174036,
+      "loss": 0.069,
+      "step": 6799
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001320155153063299,
+      "loss": 0.2041,
+      "step": 6800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013199814543472183,
+      "loss": 0.1353,
+      "step": 6801
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013198077448750003,
+      "loss": 0.1258,
+      "step": 6802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013196340246524846,
+      "loss": 0.1086,
+      "step": 6803
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000131946029368551,
+      "loss": 0.0949,
+      "step": 6804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013192865519799174,
+      "loss": 0.0655,
+      "step": 6805
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013191127995415463,
+      "loss": 0.1561,
+      "step": 6806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013189390363762377,
+      "loss": 0.1233,
+      "step": 6807
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013187652624898328,
+      "loss": 0.1832,
+      "step": 6808
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013185914778881728,
+      "loss": 0.1649,
+      "step": 6809
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013184176825770992,
+      "loss": 0.2333,
+      "step": 6810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013182438765624546,
+      "loss": 0.1613,
+      "step": 6811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001318070059850081,
+      "loss": 0.1369,
+      "step": 6812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013178962324458216,
+      "loss": 0.1942,
+      "step": 6813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013177223943555195,
+      "loss": 0.1372,
+      "step": 6814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013175485455850183,
+      "loss": 0.1102,
+      "step": 6815
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013173746861401615,
+      "loss": 0.1004,
+      "step": 6816
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013172008160267939,
+      "loss": 0.2374,
+      "step": 6817
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013170269352507597,
+      "loss": 0.1594,
+      "step": 6818
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001316853043817904,
+      "loss": 0.242,
+      "step": 6819
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013166791417340725,
+      "loss": 0.164,
+      "step": 6820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013165052290051103,
+      "loss": 0.1501,
+      "step": 6821
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013163313056368641,
+      "loss": 0.1042,
+      "step": 6822
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013161573716351796,
+      "loss": 0.0697,
+      "step": 6823
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013159834270059042,
+      "loss": 0.1417,
+      "step": 6824
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013158094717548844,
+      "loss": 0.1501,
+      "step": 6825
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013156355058879685,
+      "loss": 0.1898,
+      "step": 6826
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013154615294110037,
+      "loss": 0.2038,
+      "step": 6827
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013152875423298383,
+      "loss": 0.2145,
+      "step": 6828
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013151135446503212,
+      "loss": 0.1303,
+      "step": 6829
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013149395363783005,
+      "loss": 0.1769,
+      "step": 6830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013147655175196264,
+      "loss": 0.1956,
+      "step": 6831
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013145914880801482,
+      "loss": 0.1429,
+      "step": 6832
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013144174480657158,
+      "loss": 0.1638,
+      "step": 6833
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013142433974821794,
+      "loss": 0.1824,
+      "step": 6834
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000131406933633539,
+      "loss": 0.1365,
+      "step": 6835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013138952646311983,
+      "loss": 0.2134,
+      "step": 6836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001313721182375456,
+      "loss": 0.1155,
+      "step": 6837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001313547089574015,
+      "loss": 0.1432,
+      "step": 6838
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013133729862327268,
+      "loss": 0.2178,
+      "step": 6839
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013131988723574442,
+      "loss": 0.1589,
+      "step": 6840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013130247479540202,
+      "loss": 0.0969,
+      "step": 6841
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013128506130283076,
+      "loss": 0.1435,
+      "step": 6842
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013126764675861603,
+      "loss": 0.0972,
+      "step": 6843
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013125023116334318,
+      "loss": 0.1705,
+      "step": 6844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001312328145175977,
+      "loss": 0.1137,
+      "step": 6845
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013121539682196495,
+      "loss": 0.1221,
+      "step": 6846
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001311979780770305,
+      "loss": 0.1256,
+      "step": 6847
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013118055828337984,
+      "loss": 0.1191,
+      "step": 6848
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013116313744159855,
+      "loss": 0.1045,
+      "step": 6849
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013114571555227224,
+      "loss": 0.2469,
+      "step": 6850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001311282926159865,
+      "loss": 0.1269,
+      "step": 6851
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013111086863332706,
+      "loss": 0.2026,
+      "step": 6852
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013109344360487962,
+      "loss": 0.2714,
+      "step": 6853
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013107601753122988,
+      "loss": 0.1247,
+      "step": 6854
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001310585904129636,
+      "loss": 0.159,
+      "step": 6855
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013104116225066666,
+      "loss": 0.1056,
+      "step": 6856
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013102373304492483,
+      "loss": 0.1463,
+      "step": 6857
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013100630279632406,
+      "loss": 0.1199,
+      "step": 6858
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013098887150545023,
+      "loss": 0.1664,
+      "step": 6859
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001309714391728893,
+      "loss": 0.1396,
+      "step": 6860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013095400579922725,
+      "loss": 0.2577,
+      "step": 6861
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013093657138505012,
+      "loss": 0.163,
+      "step": 6862
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001309191359309439,
+      "loss": 0.113,
+      "step": 6863
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 0.0974,
+      "step": 6864
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001308842619052888,
+      "loss": 0.125,
+      "step": 6865
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013086682333491215,
+      "loss": 0.0996,
+      "step": 6866
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013084938372695103,
+      "loss": 0.1039,
+      "step": 6867
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013083194308199166,
+      "loss": 0.1328,
+      "step": 6868
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013081450140062036,
+      "loss": 0.1727,
+      "step": 6869
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013079705868342337,
+      "loss": 0.0839,
+      "step": 6870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013077961493098702,
+      "loss": 0.0618,
+      "step": 6871
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013076217014389773,
+      "loss": 0.2516,
+      "step": 6872
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013074472432274186,
+      "loss": 0.1413,
+      "step": 6873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013072727746810583,
+      "loss": 0.0891,
+      "step": 6874
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013070982958057617,
+      "loss": 0.1665,
+      "step": 6875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013069238066073938,
+      "loss": 0.2825,
+      "step": 6876
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013067493070918198,
+      "loss": 0.102,
+      "step": 6877
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013065747972649055,
+      "loss": 0.1214,
+      "step": 6878
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013064002771325173,
+      "loss": 0.4148,
+      "step": 6879
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013062257467005216,
+      "loss": 0.4274,
+      "step": 6880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013060512059747848,
+      "loss": 0.1612,
+      "step": 6881
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001305876654961174,
+      "loss": 0.093,
+      "step": 6882
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013057020936655577,
+      "loss": 0.1689,
+      "step": 6883
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001305527522093803,
+      "loss": 0.1411,
+      "step": 6884
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013053529402517778,
+      "loss": 0.1008,
+      "step": 6885
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013051783481453515,
+      "loss": 0.1287,
+      "step": 6886
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013050037457803924,
+      "loss": 0.3159,
+      "step": 6887
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013048291331627698,
+      "loss": 0.1449,
+      "step": 6888
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013046545102983535,
+      "loss": 0.1379,
+      "step": 6889
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013044798771930132,
+      "loss": 0.1898,
+      "step": 6890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013043052338526193,
+      "loss": 0.1283,
+      "step": 6891
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013041305802830426,
+      "loss": 0.1682,
+      "step": 6892
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013039559164901537,
+      "loss": 0.1895,
+      "step": 6893
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001303781242479824,
+      "loss": 0.1317,
+      "step": 6894
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001303606558257925,
+      "loss": 0.1902,
+      "step": 6895
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013034318638303294,
+      "loss": 0.0858,
+      "step": 6896
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001303257159202909,
+      "loss": 0.1249,
+      "step": 6897
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001303082444381536,
+      "loss": 0.2142,
+      "step": 6898
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001302907719372084,
+      "loss": 0.2298,
+      "step": 6899
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013027329841804267,
+      "loss": 0.1217,
+      "step": 6900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013025582388124366,
+      "loss": 0.2347,
+      "step": 6901
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013023834832739893,
+      "loss": 0.2437,
+      "step": 6902
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013022087175709577,
+      "loss": 0.1963,
+      "step": 6903
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001302033941709218,
+      "loss": 0.2062,
+      "step": 6904
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001301859155694644,
+      "loss": 0.0795,
+      "step": 6905
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013016843595331116,
+      "loss": 0.217,
+      "step": 6906
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001301509553230497,
+      "loss": 0.2005,
+      "step": 6907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013013347367926752,
+      "loss": 0.1423,
+      "step": 6908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013011599102255236,
+      "loss": 0.1716,
+      "step": 6909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001300985073534919,
+      "loss": 0.0993,
+      "step": 6910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013008102267267376,
+      "loss": 0.1374,
+      "step": 6911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013006353698068575,
+      "loss": 0.1237,
+      "step": 6912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013004605027811564,
+      "loss": 0.1675,
+      "step": 6913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001300285625655513,
+      "loss": 0.1045,
+      "step": 6914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013001107384358047,
+      "loss": 0.1096,
+      "step": 6915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001299935841127911,
+      "loss": 0.1806,
+      "step": 6916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001299760933737711,
+      "loss": 0.1242,
+      "step": 6917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012995860162710838,
+      "loss": 0.1165,
+      "step": 6918
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012994110887339097,
+      "loss": 0.0729,
+      "step": 6919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012992361511320687,
+      "loss": 0.2068,
+      "step": 6920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012990612034714411,
+      "loss": 0.1695,
+      "step": 6921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001298886245757908,
+      "loss": 0.1119,
+      "step": 6922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012987112779973509,
+      "loss": 0.0997,
+      "step": 6923
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012985363001956506,
+      "loss": 0.1837,
+      "step": 6924
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012983613123586891,
+      "loss": 0.1181,
+      "step": 6925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012981863144923488,
+      "loss": 0.1043,
+      "step": 6926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012980113066025123,
+      "loss": 0.1036,
+      "step": 6927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012978362886950626,
+      "loss": 0.2405,
+      "step": 6928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012976612607758825,
+      "loss": 0.1874,
+      "step": 6929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001297486222850856,
+      "loss": 0.1881,
+      "step": 6930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001297311174925866,
+      "loss": 0.2052,
+      "step": 6931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001297136117006798,
+      "loss": 0.118,
+      "step": 6932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012969610490995358,
+      "loss": 0.1441,
+      "step": 6933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012967859712099643,
+      "loss": 0.1525,
+      "step": 6934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012966108833439694,
+      "loss": 0.1459,
+      "step": 6935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001296435785507436,
+      "loss": 0.1678,
+      "step": 6936
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000129626067770625,
+      "loss": 0.1706,
+      "step": 6937
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012960855599462975,
+      "loss": 0.1278,
+      "step": 6938
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012959104322334653,
+      "loss": 0.125,
+      "step": 6939
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012957352945736406,
+      "loss": 0.1462,
+      "step": 6940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000129556014697271,
+      "loss": 0.209,
+      "step": 6941
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001295384989436562,
+      "loss": 0.1978,
+      "step": 6942
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001295209821971084,
+      "loss": 0.0951,
+      "step": 6943
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012950346445821635,
+      "loss": 0.0842,
+      "step": 6944
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012948594572756897,
+      "loss": 0.2312,
+      "step": 6945
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012946842600575518,
+      "loss": 0.1197,
+      "step": 6946
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012945090529336388,
+      "loss": 0.093,
+      "step": 6947
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012943338359098401,
+      "loss": 0.1503,
+      "step": 6948
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001294158608992046,
+      "loss": 0.1653,
+      "step": 6949
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012939833721861457,
+      "loss": 0.1617,
+      "step": 6950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012938081254980312,
+      "loss": 0.099,
+      "step": 6951
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012936328689335924,
+      "loss": 0.1168,
+      "step": 6952
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012934576024987206,
+      "loss": 0.2536,
+      "step": 6953
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012932823261993078,
+      "loss": 0.1554,
+      "step": 6954
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001293107040041246,
+      "loss": 0.1065,
+      "step": 6955
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001292931744030427,
+      "loss": 0.2061,
+      "step": 6956
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012927564381727427,
+      "loss": 0.2927,
+      "step": 6957
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00012925811224740873,
+      "loss": 0.0845,
+      "step": 6958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012924057969403537,
+      "loss": 0.2395,
+      "step": 6959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012922304615774349,
+      "loss": 0.1437,
+      "step": 6960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001292055116391225,
+      "loss": 0.1403,
+      "step": 6961
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012918797613876184,
+      "loss": 0.1823,
+      "step": 6962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012917043965725097,
+      "loss": 0.1266,
+      "step": 6963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012915290219517935,
+      "loss": 0.1631,
+      "step": 6964
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012913536375313647,
+      "loss": 0.1462,
+      "step": 6965
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012911782433171198,
+      "loss": 0.0949,
+      "step": 6966
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012910028393149533,
+      "loss": 0.1009,
+      "step": 6967
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001290827425530763,
+      "loss": 0.2151,
+      "step": 6968
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001290652001970444,
+      "loss": 0.2211,
+      "step": 6969
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012904765686398942,
+      "loss": 0.151,
+      "step": 6970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012903011255450096,
+      "loss": 0.1551,
+      "step": 6971
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012901256726916886,
+      "loss": 0.1173,
+      "step": 6972
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001289950210085829,
+      "loss": 0.1135,
+      "step": 6973
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012897747377333286,
+      "loss": 0.1676,
+      "step": 6974
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012895992556400865,
+      "loss": 0.1217,
+      "step": 6975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012894237638120002,
+      "loss": 0.1525,
+      "step": 6976
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012892482622549705,
+      "loss": 0.2194,
+      "step": 6977
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001289072750974895,
+      "loss": 0.2127,
+      "step": 6978
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012888972299776754,
+      "loss": 0.1353,
+      "step": 6979
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001288721699269211,
+      "loss": 0.1635,
+      "step": 6980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012885461588554018,
+      "loss": 0.2792,
+      "step": 6981
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001288370608742149,
+      "loss": 0.091,
+      "step": 6982
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012881950489353536,
+      "loss": 0.0973,
+      "step": 6983
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012880194794409171,
+      "loss": 0.2374,
+      "step": 6984
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012878439002647412,
+      "loss": 0.1213,
+      "step": 6985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012876683114127282,
+      "loss": 0.1377,
+      "step": 6986
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000128749271289078,
+      "loss": 0.1472,
+      "step": 6987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012873171047048,
+      "loss": 0.0951,
+      "step": 6988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001287141486860691,
+      "loss": 0.1661,
+      "step": 6989
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012869658593643558,
+      "loss": 0.0908,
+      "step": 6990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012867902222216983,
+      "loss": 0.2789,
+      "step": 6991
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012866145754386233,
+      "loss": 0.1722,
+      "step": 6992
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001286438919021034,
+      "loss": 0.119,
+      "step": 6993
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012862632529748365,
+      "loss": 0.1511,
+      "step": 6994
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001286087577305934,
+      "loss": 0.0892,
+      "step": 6995
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012859118920202335,
+      "loss": 0.1478,
+      "step": 6996
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012857361971236396,
+      "loss": 0.1168,
+      "step": 6997
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012855604926220588,
+      "loss": 0.1744,
+      "step": 6998
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012853847785213967,
+      "loss": 0.1462,
+      "step": 6999
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012852090548275607,
+      "loss": 0.1507,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012850333215464577,
+      "loss": 0.1376,
+      "step": 7001
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012848575786839943,
+      "loss": 0.1006,
+      "step": 7002
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012846818262460784,
+      "loss": 0.1113,
+      "step": 7003
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001284506064238618,
+      "loss": 0.1697,
+      "step": 7004
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012843302926675212,
+      "loss": 0.2279,
+      "step": 7005
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012841545115386963,
+      "loss": 0.3073,
+      "step": 7006
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001283978720858053,
+      "loss": 0.2133,
+      "step": 7007
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012838029206314996,
+      "loss": 0.2607,
+      "step": 7008
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001283627110864946,
+      "loss": 0.1211,
+      "step": 7009
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001283451291564302,
+      "loss": 0.1166,
+      "step": 7010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012832754627354774,
+      "loss": 0.0856,
+      "step": 7011
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012830996243843833,
+      "loss": 0.1975,
+      "step": 7012
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000128292377651693,
+      "loss": 0.0775,
+      "step": 7013
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012827479191390291,
+      "loss": 0.0874,
+      "step": 7014
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012825720522565911,
+      "loss": 0.1335,
+      "step": 7015
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012823961758755287,
+      "loss": 0.1256,
+      "step": 7016
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012822202900017536,
+      "loss": 0.165,
+      "step": 7017
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012820443946411777,
+      "loss": 0.1483,
+      "step": 7018
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012818684897997144,
+      "loss": 0.1577,
+      "step": 7019
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001281692575483277,
+      "loss": 0.0938,
+      "step": 7020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012815166516977778,
+      "loss": 0.1649,
+      "step": 7021
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012813407184491312,
+      "loss": 0.1104,
+      "step": 7022
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012811647757432506,
+      "loss": 0.1723,
+      "step": 7023
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012809888235860506,
+      "loss": 0.114,
+      "step": 7024
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012808128619834461,
+      "loss": 0.1248,
+      "step": 7025
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001280636890941352,
+      "loss": 0.1298,
+      "step": 7026
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001280460910465683,
+      "loss": 0.1611,
+      "step": 7027
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001280284920562355,
+      "loss": 0.0797,
+      "step": 7028
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012801089212372842,
+      "loss": 0.1021,
+      "step": 7029
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001279932912496386,
+      "loss": 0.1305,
+      "step": 7030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001279756894345578,
+      "loss": 0.2052,
+      "step": 7031
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012795808667907755,
+      "loss": 0.151,
+      "step": 7032
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012794048298378973,
+      "loss": 0.1476,
+      "step": 7033
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012792287834928598,
+      "loss": 0.1233,
+      "step": 7034
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012790527277615812,
+      "loss": 0.0796,
+      "step": 7035
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012788766626499796,
+      "loss": 0.1871,
+      "step": 7036
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001278700588163973,
+      "loss": 0.1518,
+      "step": 7037
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012785245043094804,
+      "loss": 0.172,
+      "step": 7038
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001278348411092421,
+      "loss": 0.1596,
+      "step": 7039
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012781723085187138,
+      "loss": 0.1349,
+      "step": 7040
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001277996196594279,
+      "loss": 0.1453,
+      "step": 7041
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012778200753250357,
+      "loss": 0.1235,
+      "step": 7042
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012776439447169055,
+      "loss": 0.2325,
+      "step": 7043
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012774678047758076,
+      "loss": 0.2154,
+      "step": 7044
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012772916555076638,
+      "loss": 0.0653,
+      "step": 7045
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012771154969183952,
+      "loss": 0.0927,
+      "step": 7046
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001276939329013923,
+      "loss": 0.1985,
+      "step": 7047
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012767631518001698,
+      "loss": 0.0998,
+      "step": 7048
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012765869652830566,
+      "loss": 0.1511,
+      "step": 7049
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001276410769468507,
+      "loss": 0.1097,
+      "step": 7050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012762345643624432,
+      "loss": 0.2642,
+      "step": 7051
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012760583499707887,
+      "loss": 0.1985,
+      "step": 7052
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012758821262994668,
+      "loss": 0.2914,
+      "step": 7053
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001275705893354401,
+      "loss": 0.123,
+      "step": 7054
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012755296511415153,
+      "loss": 0.1228,
+      "step": 7055
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012753533996667343,
+      "loss": 0.1223,
+      "step": 7056
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001275177138935983,
+      "loss": 0.1846,
+      "step": 7057
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012750008689551857,
+      "loss": 0.3018,
+      "step": 7058
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012748245897302682,
+      "loss": 0.2018,
+      "step": 7059
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001274648301267156,
+      "loss": 0.1506,
+      "step": 7060
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012744720035717745,
+      "loss": 0.1526,
+      "step": 7061
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012742956966500508,
+      "loss": 0.1543,
+      "step": 7062
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012741193805079106,
+      "loss": 0.3304,
+      "step": 7063
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012739430551512814,
+      "loss": 0.1772,
+      "step": 7064
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012737667205860902,
+      "loss": 0.1914,
+      "step": 7065
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001273590376818264,
+      "loss": 0.1746,
+      "step": 7066
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012734140238537312,
+      "loss": 0.1808,
+      "step": 7067
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012732376616984194,
+      "loss": 0.0976,
+      "step": 7068
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012730612903582574,
+      "loss": 0.1033,
+      "step": 7069
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012728849098391737,
+      "loss": 0.1395,
+      "step": 7070
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012727085201470973,
+      "loss": 0.1246,
+      "step": 7071
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012725321212879575,
+      "loss": 0.1108,
+      "step": 7072
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012723557132676838,
+      "loss": 0.188,
+      "step": 7073
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012721792960922068,
+      "loss": 0.1401,
+      "step": 7074
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012720028697674556,
+      "loss": 0.2443,
+      "step": 7075
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012718264342993617,
+      "loss": 0.1309,
+      "step": 7076
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012716499896938557,
+      "loss": 0.2745,
+      "step": 7077
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012714735359568686,
+      "loss": 0.1519,
+      "step": 7078
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012712970730943321,
+      "loss": 0.1022,
+      "step": 7079
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012711206011121776,
+      "loss": 0.0927,
+      "step": 7080
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012709441200163378,
+      "loss": 0.0942,
+      "step": 7081
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012707676298127443,
+      "loss": 0.096,
+      "step": 7082
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012705911305073306,
+      "loss": 0.1048,
+      "step": 7083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001270414622106029,
+      "loss": 0.1217,
+      "step": 7084
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012702381046147734,
+      "loss": 0.1301,
+      "step": 7085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012700615780394971,
+      "loss": 0.1501,
+      "step": 7086
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001269885042386134,
+      "loss": 0.1614,
+      "step": 7087
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012697084976606184,
+      "loss": 0.2041,
+      "step": 7088
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001269531943868885,
+      "loss": 0.1434,
+      "step": 7089
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001269355381016868,
+      "loss": 0.1601,
+      "step": 7090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012691788091105037,
+      "loss": 0.132,
+      "step": 7091
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001269002228155726,
+      "loss": 0.115,
+      "step": 7092
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001268825638158472,
+      "loss": 0.0658,
+      "step": 7093
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001268649039124677,
+      "loss": 0.1156,
+      "step": 7094
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001268472431060278,
+      "loss": 0.2769,
+      "step": 7095
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012682958139712108,
+      "loss": 0.1508,
+      "step": 7096
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001268119187863413,
+      "loss": 0.095,
+      "step": 7097
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001267942552742822,
+      "loss": 0.1363,
+      "step": 7098
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012677659086153745,
+      "loss": 0.0903,
+      "step": 7099
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012675892554870092,
+      "loss": 0.0891,
+      "step": 7100
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012674125933636639,
+      "loss": 0.3045,
+      "step": 7101
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012672359222512772,
+      "loss": 0.1766,
+      "step": 7102
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012670592421557876,
+      "loss": 0.1348,
+      "step": 7103
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012668825530831346,
+      "loss": 0.1175,
+      "step": 7104
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012667058550392576,
+      "loss": 0.2422,
+      "step": 7105
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001266529148030096,
+      "loss": 0.1512,
+      "step": 7106
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012663524320615898,
+      "loss": 0.0837,
+      "step": 7107
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012661757071396793,
+      "loss": 0.1247,
+      "step": 7108
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012659989732703052,
+      "loss": 0.2477,
+      "step": 7109
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012658222304594085,
+      "loss": 0.0963,
+      "step": 7110
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000126564547871293,
+      "loss": 0.2426,
+      "step": 7111
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012654687180368114,
+      "loss": 0.179,
+      "step": 7112
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012652919484369942,
+      "loss": 0.0892,
+      "step": 7113
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012651151699194214,
+      "loss": 0.1548,
+      "step": 7114
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012649383824900342,
+      "loss": 0.1253,
+      "step": 7115
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012647615861547765,
+      "loss": 0.1848,
+      "step": 7116
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000126458478091959,
+      "loss": 0.268,
+      "step": 7117
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001264407966790419,
+      "loss": 0.1443,
+      "step": 7118
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012642311437732068,
+      "loss": 0.1303,
+      "step": 7119
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012640543118738967,
+      "loss": 0.3069,
+      "step": 7120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012638774710984341,
+      "loss": 0.0937,
+      "step": 7121
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001263700621452762,
+      "loss": 0.1587,
+      "step": 7122
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012635237629428264,
+      "loss": 0.1376,
+      "step": 7123
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001263346895574572,
+      "loss": 0.1099,
+      "step": 7124
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012631700193539437,
+      "loss": 0.148,
+      "step": 7125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012629931342868878,
+      "loss": 0.0756,
+      "step": 7126
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000126281624037935,
+      "loss": 0.2076,
+      "step": 7127
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012626393376372768,
+      "loss": 0.2055,
+      "step": 7128
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00012624624260666143,
+      "loss": 0.1087,
+      "step": 7129
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000126228550567331,
+      "loss": 0.188,
+      "step": 7130
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012621085764633102,
+      "loss": 0.1434,
+      "step": 7131
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001261931638442563,
+      "loss": 0.2056,
+      "step": 7132
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012617546916170159,
+      "loss": 0.2183,
+      "step": 7133
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001261577735992617,
+      "loss": 0.1585,
+      "step": 7134
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001261400771575315,
+      "loss": 0.1566,
+      "step": 7135
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012612237983710583,
+      "loss": 0.1063,
+      "step": 7136
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012610468163857955,
+      "loss": 0.1203,
+      "step": 7137
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001260869825625476,
+      "loss": 0.157,
+      "step": 7138
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012606928260960494,
+      "loss": 0.1211,
+      "step": 7139
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012605158178034654,
+      "loss": 0.1121,
+      "step": 7140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012603388007536746,
+      "loss": 0.1684,
+      "step": 7141
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012601617749526266,
+      "loss": 0.2333,
+      "step": 7142
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001259984740406273,
+      "loss": 0.1337,
+      "step": 7143
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001259807697120564,
+      "loss": 0.2755,
+      "step": 7144
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012596306451014515,
+      "loss": 0.0954,
+      "step": 7145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012594535843548862,
+      "loss": 0.1808,
+      "step": 7146
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001259276514886821,
+      "loss": 0.1284,
+      "step": 7147
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012590994367032076,
+      "loss": 0.1113,
+      "step": 7148
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012589223498099986,
+      "loss": 0.1094,
+      "step": 7149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012587452542131466,
+      "loss": 0.1944,
+      "step": 7150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012585681499186047,
+      "loss": 0.1651,
+      "step": 7151
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012583910369323262,
+      "loss": 0.1766,
+      "step": 7152
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012582139152602644,
+      "loss": 0.431,
+      "step": 7153
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012580367849083742,
+      "loss": 0.079,
+      "step": 7154
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001257859645882609,
+      "loss": 0.1351,
+      "step": 7155
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012576824981889236,
+      "loss": 0.1067,
+      "step": 7156
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012575053418332729,
+      "loss": 0.1955,
+      "step": 7157
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001257328176821611,
+      "loss": 0.1448,
+      "step": 7158
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001257151003159895,
+      "loss": 0.1391,
+      "step": 7159
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012569738208540788,
+      "loss": 0.0848,
+      "step": 7160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012567966299101199,
+      "loss": 0.1006,
+      "step": 7161
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012566194303339739,
+      "loss": 0.0885,
+      "step": 7162
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001256442222131597,
+      "loss": 0.1812,
+      "step": 7163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001256265005308946,
+      "loss": 0.0856,
+      "step": 7164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001256087779871979,
+      "loss": 0.0873,
+      "step": 7165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012559105458266525,
+      "loss": 0.1452,
+      "step": 7166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012557333031789245,
+      "loss": 0.1456,
+      "step": 7167
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012555560519347534,
+      "loss": 0.1042,
+      "step": 7168
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012553787921000965,
+      "loss": 0.1985,
+      "step": 7169
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012552015236809127,
+      "loss": 0.0812,
+      "step": 7170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012550242466831616,
+      "loss": 0.1425,
+      "step": 7171
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012548469611128014,
+      "loss": 0.1031,
+      "step": 7172
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001254669666975792,
+      "loss": 0.1647,
+      "step": 7173
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012544923642780935,
+      "loss": 0.1096,
+      "step": 7174
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012543150530256654,
+      "loss": 0.1842,
+      "step": 7175
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012541377332244677,
+      "loss": 0.1865,
+      "step": 7176
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012539604048804613,
+      "loss": 0.1457,
+      "step": 7177
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012537830679996075,
+      "loss": 0.252,
+      "step": 7178
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012536057225878667,
+      "loss": 0.1084,
+      "step": 7179
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001253428368651201,
+      "loss": 0.1429,
+      "step": 7180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001253251006195572,
+      "loss": 0.2412,
+      "step": 7181
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012530736352269414,
+      "loss": 0.1243,
+      "step": 7182
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012528962557512715,
+      "loss": 0.1606,
+      "step": 7183
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001252718867774525,
+      "loss": 0.2073,
+      "step": 7184
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012525414713026654,
+      "loss": 0.1468,
+      "step": 7185
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001252364066341655,
+      "loss": 0.257,
+      "step": 7186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012521866528974576,
+      "loss": 0.076,
+      "step": 7187
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012520092309760368,
+      "loss": 0.065,
+      "step": 7188
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012518318005833566,
+      "loss": 0.0713,
+      "step": 7189
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012516543617253816,
+      "loss": 0.1342,
+      "step": 7190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001251476914408076,
+      "loss": 0.1463,
+      "step": 7191
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012512994586374047,
+      "loss": 0.0702,
+      "step": 7192
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001251121994419333,
+      "loss": 0.1735,
+      "step": 7193
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001250944521759827,
+      "loss": 0.3103,
+      "step": 7194
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012507670406648513,
+      "loss": 0.2245,
+      "step": 7195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012505895511403725,
+      "loss": 0.1056,
+      "step": 7196
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012504120531923565,
+      "loss": 0.1684,
+      "step": 7197
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000125023454682677,
+      "loss": 0.2507,
+      "step": 7198
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012500570320495802,
+      "loss": 0.2081,
+      "step": 7199
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001249879508866754,
+      "loss": 0.1318,
+      "step": 7200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012497019772842596,
+      "loss": 0.1934,
+      "step": 7201
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001249524437308063,
+      "loss": 0.1954,
+      "step": 7202
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012493468889441332,
+      "loss": 0.1386,
+      "step": 7203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012491693321984388,
+      "loss": 0.0793,
+      "step": 7204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012489917670769477,
+      "loss": 0.4155,
+      "step": 7205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001248814193585629,
+      "loss": 0.0975,
+      "step": 7206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012486366117304517,
+      "loss": 0.3619,
+      "step": 7207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012484590215173854,
+      "loss": 0.1411,
+      "step": 7208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012482814229523997,
+      "loss": 0.2289,
+      "step": 7209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012481038160414641,
+      "loss": 0.2671,
+      "step": 7210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012479262007905498,
+      "loss": 0.229,
+      "step": 7211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012477485772056262,
+      "loss": 0.1108,
+      "step": 7212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012475709452926652,
+      "loss": 0.1756,
+      "step": 7213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012473933050576372,
+      "loss": 0.1257,
+      "step": 7214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012472156565065137,
+      "loss": 0.1977,
+      "step": 7215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001247037999645266,
+      "loss": 0.072,
+      "step": 7216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012468603344798664,
+      "loss": 0.3,
+      "step": 7217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012466826610162872,
+      "loss": 0.1649,
+      "step": 7218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012465049792605005,
+      "loss": 0.1733,
+      "step": 7219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012463272892184794,
+      "loss": 0.1202,
+      "step": 7220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012461495908961968,
+      "loss": 0.1568,
+      "step": 7221
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001245971884299626,
+      "loss": 0.0874,
+      "step": 7222
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000124579416943474,
+      "loss": 0.1946,
+      "step": 7223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012456164463075137,
+      "loss": 0.1538,
+      "step": 7224
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012454387149239208,
+      "loss": 0.1495,
+      "step": 7225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012452609752899358,
+      "loss": 0.1899,
+      "step": 7226
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001245083227411533,
+      "loss": 0.0922,
+      "step": 7227
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012449054712946876,
+      "loss": 0.0826,
+      "step": 7228
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012447277069453752,
+      "loss": 0.2057,
+      "step": 7229
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012445499343695707,
+      "loss": 0.0767,
+      "step": 7230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012443721535732505,
+      "loss": 0.1735,
+      "step": 7231
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012441943645623903,
+      "loss": 0.2651,
+      "step": 7232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012440165673429667,
+      "loss": 0.1284,
+      "step": 7233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012438387619209562,
+      "loss": 0.2972,
+      "step": 7234
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012436609483023353,
+      "loss": 0.1614,
+      "step": 7235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001243483126493082,
+      "loss": 0.1696,
+      "step": 7236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012433052964991732,
+      "loss": 0.3836,
+      "step": 7237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012431274583265867,
+      "loss": 0.1331,
+      "step": 7238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001242949611981301,
+      "loss": 0.109,
+      "step": 7239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012427717574692936,
+      "loss": 0.1075,
+      "step": 7240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012425938947965433,
+      "loss": 0.1621,
+      "step": 7241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012424160239690293,
+      "loss": 0.2086,
+      "step": 7242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012422381449927306,
+      "loss": 0.1225,
+      "step": 7243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001242060257873626,
+      "loss": 0.1229,
+      "step": 7244
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001241882362617696,
+      "loss": 0.1032,
+      "step": 7245
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012417044592309202,
+      "loss": 0.1513,
+      "step": 7246
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012415265477192783,
+      "loss": 0.1147,
+      "step": 7247
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012413486280887516,
+      "loss": 0.1171,
+      "step": 7248
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012411707003453202,
+      "loss": 0.2268,
+      "step": 7249
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012409927644949656,
+      "loss": 0.205,
+      "step": 7250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012408148205436684,
+      "loss": 0.0831,
+      "step": 7251
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001240636868497411,
+      "loss": 0.0959,
+      "step": 7252
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012404589083621752,
+      "loss": 0.2023,
+      "step": 7253
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001240280940143942,
+      "loss": 0.0804,
+      "step": 7254
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012401029638486953,
+      "loss": 0.1106,
+      "step": 7255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012399249794824163,
+      "loss": 0.0732,
+      "step": 7256
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001239746987051089,
+      "loss": 0.1411,
+      "step": 7257
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001239568986560696,
+      "loss": 0.1314,
+      "step": 7258
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012393909780172212,
+      "loss": 0.1093,
+      "step": 7259
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012392129614266484,
+      "loss": 0.0775,
+      "step": 7260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012390349367949607,
+      "loss": 0.1651,
+      "step": 7261
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012388569041281436,
+      "loss": 0.205,
+      "step": 7262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012386788634321803,
+      "loss": 0.1098,
+      "step": 7263
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012385008147130567,
+      "loss": 0.0971,
+      "step": 7264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001238322757976758,
+      "loss": 0.0947,
+      "step": 7265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012381446932292685,
+      "loss": 0.2102,
+      "step": 7266
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012379666204765746,
+      "loss": 0.1046,
+      "step": 7267
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012377885397246623,
+      "loss": 0.0934,
+      "step": 7268
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012376104509795172,
+      "loss": 0.1366,
+      "step": 7269
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001237432354247126,
+      "loss": 0.3055,
+      "step": 7270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012372542495334753,
+      "loss": 0.1077,
+      "step": 7271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012370761368445528,
+      "loss": 0.1857,
+      "step": 7272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012368980161863444,
+      "loss": 0.161,
+      "step": 7273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012367198875648386,
+      "loss": 0.0774,
+      "step": 7274
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001236541750986023,
+      "loss": 0.2122,
+      "step": 7275
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012363636064558855,
+      "loss": 0.1521,
+      "step": 7276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012361854539804143,
+      "loss": 0.1341,
+      "step": 7277
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012360072935655982,
+      "loss": 0.2115,
+      "step": 7278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012358291252174261,
+      "loss": 0.1364,
+      "step": 7279
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012356509489418866,
+      "loss": 0.1954,
+      "step": 7280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012354727647449697,
+      "loss": 0.0683,
+      "step": 7281
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012352945726326647,
+      "loss": 0.1733,
+      "step": 7282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012351163726109616,
+      "loss": 0.1839,
+      "step": 7283
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012349381646858503,
+      "loss": 0.098,
+      "step": 7284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001234759948863322,
+      "loss": 0.1757,
+      "step": 7285
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012345817251493664,
+      "loss": 0.1614,
+      "step": 7286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001234403493549975,
+      "loss": 0.0817,
+      "step": 7287
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012342252540711393,
+      "loss": 0.0852,
+      "step": 7288
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012340470067188504,
+      "loss": 0.0775,
+      "step": 7289
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012338687514991002,
+      "loss": 0.1778,
+      "step": 7290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012336904884178805,
+      "loss": 0.1525,
+      "step": 7291
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001233512217481184,
+      "loss": 0.255,
+      "step": 7292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001233333938695003,
+      "loss": 0.1191,
+      "step": 7293
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000123315565206533,
+      "loss": 0.1319,
+      "step": 7294
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012329773575981586,
+      "loss": 0.076,
+      "step": 7295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012327990552994818,
+      "loss": 0.2328,
+      "step": 7296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012326207451752937,
+      "loss": 0.1487,
+      "step": 7297
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012324424272315882,
+      "loss": 0.1617,
+      "step": 7298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012322641014743586,
+      "loss": 0.1175,
+      "step": 7299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012320857679096,
+      "loss": 0.1621,
+      "step": 7300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012319074265433063,
+      "loss": 0.1324,
+      "step": 7301
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012317290773814735,
+      "loss": 0.1365,
+      "step": 7302
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012315507204300962,
+      "loss": 0.1555,
+      "step": 7303
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000123137235569517,
+      "loss": 0.1693,
+      "step": 7304
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012311939831826903,
+      "loss": 0.1143,
+      "step": 7305
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012310156028986534,
+      "loss": 0.1425,
+      "step": 7306
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012308372148490558,
+      "loss": 0.1273,
+      "step": 7307
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001230658819039893,
+      "loss": 0.2196,
+      "step": 7308
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012304804154771625,
+      "loss": 0.1619,
+      "step": 7309
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012303020041668614,
+      "loss": 0.1278,
+      "step": 7310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012301235851149865,
+      "loss": 0.0702,
+      "step": 7311
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001229945158327536,
+      "loss": 0.1489,
+      "step": 7312
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012297667238105067,
+      "loss": 0.2617,
+      "step": 7313
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012295882815698976,
+      "loss": 0.1663,
+      "step": 7314
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012294098316117065,
+      "loss": 0.2185,
+      "step": 7315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012292313739419323,
+      "loss": 0.14,
+      "step": 7316
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012290529085665737,
+      "loss": 0.2148,
+      "step": 7317
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012288744354916294,
+      "loss": 0.1954,
+      "step": 7318
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012286959547230995,
+      "loss": 0.1196,
+      "step": 7319
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012285174662669829,
+      "loss": 0.0897,
+      "step": 7320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000122833897012928,
+      "loss": 0.0928,
+      "step": 7321
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012281604663159904,
+      "loss": 0.1692,
+      "step": 7322
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001227981954833115,
+      "loss": 0.111,
+      "step": 7323
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012278034356866545,
+      "loss": 0.216,
+      "step": 7324
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001227624908882609,
+      "loss": 0.2169,
+      "step": 7325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012274463744269806,
+      "loss": 0.0954,
+      "step": 7326
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000122726783232577,
+      "loss": 0.0622,
+      "step": 7327
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012270892825849795,
+      "loss": 0.1082,
+      "step": 7328
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012269107252106104,
+      "loss": 0.119,
+      "step": 7329
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012267321602086651,
+      "loss": 0.2218,
+      "step": 7330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012265535875851465,
+      "loss": 0.2059,
+      "step": 7331
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012263750073460565,
+      "loss": 0.1413,
+      "step": 7332
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012261964194973987,
+      "loss": 0.1023,
+      "step": 7333
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001226017824045176,
+      "loss": 0.1006,
+      "step": 7334
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012258392209953922,
+      "loss": 0.0685,
+      "step": 7335
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012256606103540503,
+      "loss": 0.0653,
+      "step": 7336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001225481992127155,
+      "loss": 0.1073,
+      "step": 7337
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012253033663207105,
+      "loss": 0.2926,
+      "step": 7338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012251247329407206,
+      "loss": 0.1435,
+      "step": 7339
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012249460919931907,
+      "loss": 0.1846,
+      "step": 7340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012247674434841258,
+      "loss": 0.1602,
+      "step": 7341
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012245887874195305,
+      "loss": 0.224,
+      "step": 7342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001224410123805411,
+      "loss": 0.1492,
+      "step": 7343
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012242314526477726,
+      "loss": 0.2029,
+      "step": 7344
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001224052773952622,
+      "loss": 0.115,
+      "step": 7345
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012238740877259643,
+      "loss": 0.0849,
+      "step": 7346
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001223695393973807,
+      "loss": 0.1711,
+      "step": 7347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012235166927021564,
+      "loss": 0.1607,
+      "step": 7348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012233379839170198,
+      "loss": 0.0793,
+      "step": 7349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012231592676244043,
+      "loss": 0.1706,
+      "step": 7350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012229805438303176,
+      "loss": 0.1995,
+      "step": 7351
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012228018125407672,
+      "loss": 0.1871,
+      "step": 7352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012226230737617612,
+      "loss": 0.1703,
+      "step": 7353
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001222444327499308,
+      "loss": 0.0852,
+      "step": 7354
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012222655737594162,
+      "loss": 0.1156,
+      "step": 7355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012220868125480943,
+      "loss": 0.0567,
+      "step": 7356
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012219080438713518,
+      "loss": 0.0818,
+      "step": 7357
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012217292677351975,
+      "loss": 0.1706,
+      "step": 7358
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012215504841456412,
+      "loss": 0.2722,
+      "step": 7359
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012213716931086923,
+      "loss": 0.129,
+      "step": 7360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012211928946303616,
+      "loss": 0.1228,
+      "step": 7361
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012210140887166588,
+      "loss": 0.1708,
+      "step": 7362
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012208352753735946,
+      "loss": 0.1569,
+      "step": 7363
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012206564546071797,
+      "loss": 0.1546,
+      "step": 7364
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012204776264234253,
+      "loss": 0.1625,
+      "step": 7365
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012202987908283426,
+      "loss": 0.1048,
+      "step": 7366
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012201199478279429,
+      "loss": 0.2062,
+      "step": 7367
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012199410974282386,
+      "loss": 0.2153,
+      "step": 7368
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012197622396352413,
+      "loss": 0.0711,
+      "step": 7369
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001219583374454963,
+      "loss": 0.2333,
+      "step": 7370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012194045018934169,
+      "loss": 0.2198,
+      "step": 7371
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012192256219566151,
+      "loss": 0.1062,
+      "step": 7372
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012190467346505714,
+      "loss": 0.1186,
+      "step": 7373
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012188678399812985,
+      "loss": 0.0985,
+      "step": 7374
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000121868893795481,
+      "loss": 0.1461,
+      "step": 7375
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000121851002857712,
+      "loss": 0.1567,
+      "step": 7376
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001218331111854242,
+      "loss": 0.2383,
+      "step": 7377
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012181521877921906,
+      "loss": 0.1108,
+      "step": 7378
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012179732563969801,
+      "loss": 0.1146,
+      "step": 7379
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012177943176746258,
+      "loss": 0.3909,
+      "step": 7380
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001217615371631142,
+      "loss": 0.1147,
+      "step": 7381
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012174364182725442,
+      "loss": 0.1228,
+      "step": 7382
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012172574576048481,
+      "loss": 0.1149,
+      "step": 7383
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012170784896340692,
+      "loss": 0.1591,
+      "step": 7384
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012168995143662237,
+      "loss": 0.0808,
+      "step": 7385
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012167205318073274,
+      "loss": 0.088,
+      "step": 7386
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012165415419633972,
+      "loss": 0.0715,
+      "step": 7387
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012163625448404499,
+      "loss": 0.1026,
+      "step": 7388
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012161835404445021,
+      "loss": 0.1354,
+      "step": 7389
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012160045287815711,
+      "loss": 0.1586,
+      "step": 7390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012158255098576746,
+      "loss": 0.1192,
+      "step": 7391
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000121564648367883,
+      "loss": 0.1054,
+      "step": 7392
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012154674502510555,
+      "loss": 0.0928,
+      "step": 7393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001215288409580369,
+      "loss": 0.1835,
+      "step": 7394
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012151093616727893,
+      "loss": 0.155,
+      "step": 7395
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012149303065343347,
+      "loss": 0.1557,
+      "step": 7396
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012147512441710243,
+      "loss": 0.1453,
+      "step": 7397
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012145721745888772,
+      "loss": 0.0927,
+      "step": 7398
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012143930977939128,
+      "loss": 0.2014,
+      "step": 7399
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012142140137921509,
+      "loss": 0.1413,
+      "step": 7400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012140349225896112,
+      "loss": 0.0798,
+      "step": 7401
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012138558241923142,
+      "loss": 0.1026,
+      "step": 7402
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012136767186062797,
+      "loss": 0.1772,
+      "step": 7403
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012134976058375287,
+      "loss": 0.113,
+      "step": 7404
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012133184858920816,
+      "loss": 0.1421,
+      "step": 7405
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012131393587759601,
+      "loss": 0.1061,
+      "step": 7406
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012129602244951855,
+      "loss": 0.1552,
+      "step": 7407
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012127810830557786,
+      "loss": 0.1143,
+      "step": 7408
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012126019344637623,
+      "loss": 0.1043,
+      "step": 7409
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012124227787251576,
+      "loss": 0.1963,
+      "step": 7410
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012122436158459877,
+      "loss": 0.099,
+      "step": 7411
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012120644458322746,
+      "loss": 0.1113,
+      "step": 7412
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012118852686900415,
+      "loss": 0.151,
+      "step": 7413
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012117060844253109,
+      "loss": 0.1083,
+      "step": 7414
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012115268930441061,
+      "loss": 0.2718,
+      "step": 7415
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012113476945524513,
+      "loss": 0.2303,
+      "step": 7416
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012111684889563693,
+      "loss": 0.079,
+      "step": 7417
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012109892762618848,
+      "loss": 0.1434,
+      "step": 7418
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012108100564750215,
+      "loss": 0.0911,
+      "step": 7419
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012106308296018045,
+      "loss": 0.113,
+      "step": 7420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012104515956482578,
+      "loss": 0.2437,
+      "step": 7421
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012102723546204064,
+      "loss": 0.3034,
+      "step": 7422
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001210093106524276,
+      "loss": 0.1137,
+      "step": 7423
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012099138513658912,
+      "loss": 0.1288,
+      "step": 7424
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012097345891512786,
+      "loss": 0.162,
+      "step": 7425
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012095553198864636,
+      "loss": 0.3674,
+      "step": 7426
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001209376043577472,
+      "loss": 0.098,
+      "step": 7427
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012091967602303304,
+      "loss": 0.1356,
+      "step": 7428
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012090174698510654,
+      "loss": 0.0707,
+      "step": 7429
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012088381724457041,
+      "loss": 0.096,
+      "step": 7430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012086588680202731,
+      "loss": 0.1577,
+      "step": 7431
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012084795565808003,
+      "loss": 0.134,
+      "step": 7432
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012083002381333126,
+      "loss": 0.1142,
+      "step": 7433
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001208120912683838,
+      "loss": 0.0819,
+      "step": 7434
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012079415802384044,
+      "loss": 0.1649,
+      "step": 7435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012077622408030402,
+      "loss": 0.3783,
+      "step": 7436
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001207582894383774,
+      "loss": 0.081,
+      "step": 7437
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001207403540986634,
+      "loss": 0.1842,
+      "step": 7438
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000120722418061765,
+      "loss": 0.1768,
+      "step": 7439
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012070448132828502,
+      "loss": 0.1598,
+      "step": 7440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012068654389882649,
+      "loss": 0.0907,
+      "step": 7441
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012066860577399228,
+      "loss": 0.1295,
+      "step": 7442
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012065066695438546,
+      "loss": 0.1351,
+      "step": 7443
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000120632727440609,
+      "loss": 0.1381,
+      "step": 7444
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012061478723326594,
+      "loss": 0.1801,
+      "step": 7445
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012059684633295938,
+      "loss": 0.1967,
+      "step": 7446
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012057890474029232,
+      "loss": 0.0653,
+      "step": 7447
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001205609624558679,
+      "loss": 0.2058,
+      "step": 7448
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001205430194802893,
+      "loss": 0.1366,
+      "step": 7449
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001205250758141596,
+      "loss": 0.0963,
+      "step": 7450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012050713145808201,
+      "loss": 0.1158,
+      "step": 7451
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012048918641265971,
+      "loss": 0.1078,
+      "step": 7452
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012047124067849595,
+      "loss": 0.1319,
+      "step": 7453
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012045329425619392,
+      "loss": 0.1926,
+      "step": 7454
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012043534714635692,
+      "loss": 0.1078,
+      "step": 7455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012041739934958827,
+      "loss": 0.1656,
+      "step": 7456
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012039945086649122,
+      "loss": 0.0935,
+      "step": 7457
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012038150169766915,
+      "loss": 0.1469,
+      "step": 7458
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012036355184372543,
+      "loss": 0.129,
+      "step": 7459
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001203456013052634,
+      "loss": 0.0975,
+      "step": 7460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012032765008288646,
+      "loss": 0.1464,
+      "step": 7461
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012030969817719808,
+      "loss": 0.1241,
+      "step": 7462
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012029174558880172,
+      "loss": 0.2717,
+      "step": 7463
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012027379231830078,
+      "loss": 0.1212,
+      "step": 7464
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012025583836629888,
+      "loss": 0.0849,
+      "step": 7465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012023788373339939,
+      "loss": 0.222,
+      "step": 7466
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012021992842020597,
+      "loss": 0.1038,
+      "step": 7467
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012020197242732212,
+      "loss": 0.1373,
+      "step": 7468
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012018401575535144,
+      "loss": 0.107,
+      "step": 7469
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012016605840489759,
+      "loss": 0.1586,
+      "step": 7470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012014810037656412,
+      "loss": 0.093,
+      "step": 7471
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001201301416709548,
+      "loss": 0.1208,
+      "step": 7472
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012011218228867317,
+      "loss": 0.1318,
+      "step": 7473
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012009422223032307,
+      "loss": 0.2382,
+      "step": 7474
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012007626149650813,
+      "loss": 0.1031,
+      "step": 7475
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001200583000878321,
+      "loss": 0.1256,
+      "step": 7476
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012004033800489882,
+      "loss": 0.1703,
+      "step": 7477
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012002237524831203,
+      "loss": 0.0997,
+      "step": 7478
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001200044118186756,
+      "loss": 0.114,
+      "step": 7479
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011998644771659326,
+      "loss": 0.1043,
+      "step": 7480
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011996848294266898,
+      "loss": 0.12,
+      "step": 7481
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011995051749750656,
+      "loss": 0.1358,
+      "step": 7482
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011993255138170997,
+      "loss": 0.1437,
+      "step": 7483
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011991458459588317,
+      "loss": 0.1469,
+      "step": 7484
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011989661714062999,
+      "loss": 0.1263,
+      "step": 7485
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001198786490165545,
+      "loss": 0.1276,
+      "step": 7486
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011986068022426065,
+      "loss": 0.3596,
+      "step": 7487
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001198427107643525,
+      "loss": 0.069,
+      "step": 7488
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011982474063743403,
+      "loss": 0.1059,
+      "step": 7489
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011980676984410934,
+      "loss": 0.1136,
+      "step": 7490
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011978879838498258,
+      "loss": 0.2516,
+      "step": 7491
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011977082626065777,
+      "loss": 0.3088,
+      "step": 7492
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011975285347173906,
+      "loss": 0.3117,
+      "step": 7493
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001197348800188306,
+      "loss": 0.2172,
+      "step": 7494
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001197169059025366,
+      "loss": 0.0852,
+      "step": 7495
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011969893112346118,
+      "loss": 0.2187,
+      "step": 7496
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011968095568220867,
+      "loss": 0.2914,
+      "step": 7497
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011966297957938323,
+      "loss": 0.1552,
+      "step": 7498
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011964500281558915,
+      "loss": 0.1369,
+      "step": 7499
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011962702539143074,
+      "loss": 0.1097,
+      "step": 7500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011960904730751224,
+      "loss": 0.1371,
+      "step": 7501
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011959106856443806,
+      "loss": 0.1808,
+      "step": 7502
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011957308916281248,
+      "loss": 0.2984,
+      "step": 7503
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011955510910323997,
+      "loss": 0.1314,
+      "step": 7504
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011953712838632486,
+      "loss": 0.1045,
+      "step": 7505
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011951914701267157,
+      "loss": 0.1537,
+      "step": 7506
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011950116498288455,
+      "loss": 0.1264,
+      "step": 7507
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011948318229756827,
+      "loss": 0.1777,
+      "step": 7508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001194651989573272,
+      "loss": 0.1242,
+      "step": 7509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011944721496276591,
+      "loss": 0.1777,
+      "step": 7510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011942923031448883,
+      "loss": 0.1704,
+      "step": 7511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011941124501310059,
+      "loss": 0.0632,
+      "step": 7512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011939325905920573,
+      "loss": 0.2096,
+      "step": 7513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011937527245340885,
+      "loss": 0.2354,
+      "step": 7514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011935728519631455,
+      "loss": 0.1101,
+      "step": 7515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011933929728852749,
+      "loss": 0.2485,
+      "step": 7516
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011932130873065239,
+      "loss": 0.1852,
+      "step": 7517
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011930331952329382,
+      "loss": 0.0765,
+      "step": 7518
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011928532966705655,
+      "loss": 0.0733,
+      "step": 7519
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011926733916254527,
+      "loss": 0.1731,
+      "step": 7520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011924934801036478,
+      "loss": 0.2062,
+      "step": 7521
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011923135621111983,
+      "loss": 0.144,
+      "step": 7522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011921336376541518,
+      "loss": 0.1505,
+      "step": 7523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001191953706738557,
+      "loss": 0.1052,
+      "step": 7524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011917737693704618,
+      "loss": 0.1359,
+      "step": 7525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011915938255559151,
+      "loss": 0.113,
+      "step": 7526
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011914138753009652,
+      "loss": 0.1542,
+      "step": 7527
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011912339186116617,
+      "loss": 0.3126,
+      "step": 7528
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00011910539554940534,
+      "loss": 0.1043,
+      "step": 7529
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000119087398595419,
+      "loss": 0.1163,
+      "step": 7530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001190694009998121,
+      "loss": 0.2195,
+      "step": 7531
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011905140276318962,
+      "loss": 0.1526,
+      "step": 7532
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011903340388615659,
+      "loss": 0.1019,
+      "step": 7533
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011901540436931802,
+      "loss": 0.2076,
+      "step": 7534
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011899740421327899,
+      "loss": 0.1264,
+      "step": 7535
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011897940341864453,
+      "loss": 0.0951,
+      "step": 7536
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011896140198601978,
+      "loss": 0.1401,
+      "step": 7537
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001189433999160098,
+      "loss": 0.1341,
+      "step": 7538
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011892539720921978,
+      "loss": 0.0896,
+      "step": 7539
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011890739386625487,
+      "loss": 0.0975,
+      "step": 7540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011888938988772023,
+      "loss": 0.1759,
+      "step": 7541
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011887138527422108,
+      "loss": 0.1794,
+      "step": 7542
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011885338002636264,
+      "loss": 0.1963,
+      "step": 7543
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011883537414475014,
+      "loss": 0.0984,
+      "step": 7544
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011881736762998885,
+      "loss": 0.0899,
+      "step": 7545
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011879936048268405,
+      "loss": 0.1091,
+      "step": 7546
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001187813527034411,
+      "loss": 0.162,
+      "step": 7547
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011876334429286524,
+      "loss": 0.1024,
+      "step": 7548
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001187453352515619,
+      "loss": 0.1046,
+      "step": 7549
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011872732558013644,
+      "loss": 0.1184,
+      "step": 7550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001187093152791942,
+      "loss": 0.1663,
+      "step": 7551
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011869130434934064,
+      "loss": 0.0775,
+      "step": 7552
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011867329279118116,
+      "loss": 0.0954,
+      "step": 7553
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011865528060532127,
+      "loss": 0.2021,
+      "step": 7554
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011863726779236642,
+      "loss": 0.1594,
+      "step": 7555
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011861925435292209,
+      "loss": 0.0918,
+      "step": 7556
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011860124028759383,
+      "loss": 0.1929,
+      "step": 7557
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011858322559698714,
+      "loss": 0.0933,
+      "step": 7558
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011856521028170762,
+      "loss": 0.0684,
+      "step": 7559
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011854719434236081,
+      "loss": 0.0542,
+      "step": 7560
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001185291777795524,
+      "loss": 0.1168,
+      "step": 7561
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011851116059388794,
+      "loss": 0.1677,
+      "step": 7562
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011849314278597306,
+      "loss": 0.1484,
+      "step": 7563
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001184751243564135,
+      "loss": 0.1007,
+      "step": 7564
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011845710530581486,
+      "loss": 0.1589,
+      "step": 7565
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011843908563478294,
+      "loss": 0.1549,
+      "step": 7566
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011842106534392341,
+      "loss": 0.1147,
+      "step": 7567
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011840304443384202,
+      "loss": 0.2454,
+      "step": 7568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011838502290514459,
+      "loss": 0.0933,
+      "step": 7569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011836700075843682,
+      "loss": 0.1231,
+      "step": 7570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011834897799432463,
+      "loss": 0.127,
+      "step": 7571
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011833095461341379,
+      "loss": 0.1564,
+      "step": 7572
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011831293061631017,
+      "loss": 0.2905,
+      "step": 7573
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011829490600361961,
+      "loss": 0.3123,
+      "step": 7574
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011827688077594806,
+      "loss": 0.072,
+      "step": 7575
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001182588549339014,
+      "loss": 0.1039,
+      "step": 7576
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011824082847808558,
+      "loss": 0.1345,
+      "step": 7577
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011822280140910655,
+      "loss": 0.1031,
+      "step": 7578
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011820477372757029,
+      "loss": 0.1206,
+      "step": 7579
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011818674543408281,
+      "loss": 0.0941,
+      "step": 7580
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011816871652925013,
+      "loss": 0.0889,
+      "step": 7581
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011815068701367826,
+      "loss": 0.127,
+      "step": 7582
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011813265688797328,
+      "loss": 0.1879,
+      "step": 7583
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011811462615274127,
+      "loss": 0.1268,
+      "step": 7584
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011809659480858832,
+      "loss": 0.1838,
+      "step": 7585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011807856285612057,
+      "loss": 0.0691,
+      "step": 7586
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00011806053029594418,
+      "loss": 0.1861,
+      "step": 7587
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011804249712866527,
+      "loss": 0.1273,
+      "step": 7588
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011802446335488999,
+      "loss": 0.1003,
+      "step": 7589
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011800642897522465,
+      "loss": 0.1382,
+      "step": 7590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011798839399027536,
+      "loss": 0.1302,
+      "step": 7591
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011797035840064847,
+      "loss": 0.1429,
+      "step": 7592
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011795232220695017,
+      "loss": 0.1799,
+      "step": 7593
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011793428540978679,
+      "loss": 0.1892,
+      "step": 7594
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001179162480097646,
+      "loss": 0.0775,
+      "step": 7595
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001178982100074899,
+      "loss": 0.1834,
+      "step": 7596
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011788017140356913,
+      "loss": 0.0963,
+      "step": 7597
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011786213219860855,
+      "loss": 0.2437,
+      "step": 7598
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011784409239321461,
+      "loss": 0.3263,
+      "step": 7599
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001178260519879937,
+      "loss": 0.1229,
+      "step": 7600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011780801098355225,
+      "loss": 0.1743,
+      "step": 7601
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011778996938049671,
+      "loss": 0.1415,
+      "step": 7602
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011777192717943351,
+      "loss": 0.1822,
+      "step": 7603
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001177538843809692,
+      "loss": 0.09,
+      "step": 7604
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001177358409857102,
+      "loss": 0.1962,
+      "step": 7605
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011771779699426313,
+      "loss": 0.1365,
+      "step": 7606
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011769975240723447,
+      "loss": 0.1261,
+      "step": 7607
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011768170722523083,
+      "loss": 0.272,
+      "step": 7608
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011766366144885877,
+      "loss": 0.0918,
+      "step": 7609
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011764561507872487,
+      "loss": 0.1875,
+      "step": 7610
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011762756811543583,
+      "loss": 0.1234,
+      "step": 7611
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011760952055959821,
+      "loss": 0.1538,
+      "step": 7612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011759147241181878,
+      "loss": 0.4159,
+      "step": 7613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011757342367270415,
+      "loss": 0.0916,
+      "step": 7614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011755537434286101,
+      "loss": 0.0764,
+      "step": 7615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011753732442289615,
+      "loss": 0.3226,
+      "step": 7616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011751927391341628,
+      "loss": 0.0943,
+      "step": 7617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011750122281502817,
+      "loss": 0.0879,
+      "step": 7618
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011748317112833862,
+      "loss": 0.1212,
+      "step": 7619
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011746511885395442,
+      "loss": 0.137,
+      "step": 7620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011744706599248241,
+      "loss": 0.1698,
+      "step": 7621
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011742901254452938,
+      "loss": 0.1501,
+      "step": 7622
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011741095851070228,
+      "loss": 0.0677,
+      "step": 7623
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011739290389160791,
+      "loss": 0.2542,
+      "step": 7624
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011737484868785325,
+      "loss": 0.1073,
+      "step": 7625
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011735679290004517,
+      "loss": 0.1373,
+      "step": 7626
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011733873652879064,
+      "loss": 0.0943,
+      "step": 7627
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011732067957469661,
+      "loss": 0.1026,
+      "step": 7628
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011730262203837005,
+      "loss": 0.0922,
+      "step": 7629
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011728456392041802,
+      "loss": 0.1155,
+      "step": 7630
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011726650522144747,
+      "loss": 0.2061,
+      "step": 7631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011724844594206547,
+      "loss": 0.0833,
+      "step": 7632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001172303860828791,
+      "loss": 0.2005,
+      "step": 7633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011721232564449541,
+      "loss": 0.1013,
+      "step": 7634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001171942646275215,
+      "loss": 0.1037,
+      "step": 7635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011717620303256449,
+      "loss": 0.1085,
+      "step": 7636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011715814086023155,
+      "loss": 0.1436,
+      "step": 7637
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011714007811112981,
+      "loss": 0.094,
+      "step": 7638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011712201478586647,
+      "loss": 0.1454,
+      "step": 7639
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011710395088504869,
+      "loss": 0.3024,
+      "step": 7640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011708588640928368,
+      "loss": 0.2191,
+      "step": 7641
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011706782135917876,
+      "loss": 0.1355,
+      "step": 7642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011704975573534107,
+      "loss": 0.0859,
+      "step": 7643
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00011703168953837796,
+      "loss": 0.0867,
+      "step": 7644
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011701362276889669,
+      "loss": 0.1515,
+      "step": 7645
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001169955554275046,
+      "loss": 0.4242,
+      "step": 7646
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011697748751480899,
+      "loss": 0.2356,
+      "step": 7647
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011695941903141723,
+      "loss": 0.1029,
+      "step": 7648
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011694134997793669,
+      "loss": 0.0639,
+      "step": 7649
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011692328035497473,
+      "loss": 0.0798,
+      "step": 7650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011690521016313882,
+      "loss": 0.0729,
+      "step": 7651
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011688713940303635,
+      "loss": 0.2515,
+      "step": 7652
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011686906807527473,
+      "loss": 0.0955,
+      "step": 7653
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011685099618046149,
+      "loss": 0.118,
+      "step": 7654
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011683292371920407,
+      "loss": 0.101,
+      "step": 7655
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011681485069210997,
+      "loss": 0.2739,
+      "step": 7656
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011679677709978675,
+      "loss": 0.0873,
+      "step": 7657
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011677870294284197,
+      "loss": 0.1382,
+      "step": 7658
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011676062822188311,
+      "loss": 0.1833,
+      "step": 7659
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011674255293751781,
+      "loss": 0.1911,
+      "step": 7660
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011672447709035366,
+      "loss": 0.1298,
+      "step": 7661
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011670640068099824,
+      "loss": 0.1877,
+      "step": 7662
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011668832371005924,
+      "loss": 0.2828,
+      "step": 7663
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011667024617814428,
+      "loss": 0.2233,
+      "step": 7664
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011665216808586107,
+      "loss": 0.1181,
+      "step": 7665
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011663408943381727,
+      "loss": 0.0979,
+      "step": 7666
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011661601022262057,
+      "loss": 0.0864,
+      "step": 7667
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011659793045287875,
+      "loss": 0.2502,
+      "step": 7668
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011657985012519952,
+      "loss": 0.1362,
+      "step": 7669
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011656176924019071,
+      "loss": 0.1135,
+      "step": 7670
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011654368779846002,
+      "loss": 0.1076,
+      "step": 7671
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011652560580061533,
+      "loss": 0.1565,
+      "step": 7672
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011650752324726442,
+      "loss": 0.1941,
+      "step": 7673
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011648944013901512,
+      "loss": 0.1616,
+      "step": 7674
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011647135647647536,
+      "loss": 0.2052,
+      "step": 7675
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011645327226025295,
+      "loss": 0.1203,
+      "step": 7676
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011643518749095586,
+      "loss": 0.1402,
+      "step": 7677
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011641710216919192,
+      "loss": 0.1398,
+      "step": 7678
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011639901629556912,
+      "loss": 0.0808,
+      "step": 7679
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011638092987069542,
+      "loss": 0.2091,
+      "step": 7680
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011636284289517874,
+      "loss": 0.2589,
+      "step": 7681
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011634475536962714,
+      "loss": 0.1674,
+      "step": 7682
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011632666729464858,
+      "loss": 0.1835,
+      "step": 7683
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011630857867085114,
+      "loss": 0.1832,
+      "step": 7684
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001162904894988428,
+      "loss": 0.1484,
+      "step": 7685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001162723997792317,
+      "loss": 0.212,
+      "step": 7686
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011625430951262583,
+      "loss": 0.1742,
+      "step": 7687
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011623621869963334,
+      "loss": 0.1435,
+      "step": 7688
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011621812734086241,
+      "loss": 0.0746,
+      "step": 7689
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011620003543692109,
+      "loss": 0.2185,
+      "step": 7690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011618194298841761,
+      "loss": 0.2066,
+      "step": 7691
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011616384999596006,
+      "loss": 0.3134,
+      "step": 7692
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011614575646015669,
+      "loss": 0.0702,
+      "step": 7693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001161276623816157,
+      "loss": 0.2026,
+      "step": 7694
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011610956776094532,
+      "loss": 0.225,
+      "step": 7695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011609147259875382,
+      "loss": 0.1262,
+      "step": 7696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011607337689564945,
+      "loss": 0.1408,
+      "step": 7697
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001160552806522405,
+      "loss": 0.3158,
+      "step": 7698
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011603718386913522,
+      "loss": 0.1517,
+      "step": 7699
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011601908654694198,
+      "loss": 0.2055,
+      "step": 7700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00011600098868626917,
+      "loss": 0.1513,
+      "step": 7701
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011598289028772504,
+      "loss": 0.1467,
+      "step": 7702
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011596479135191804,
+      "loss": 0.1134,
+      "step": 7703
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011594669187945655,
+      "loss": 0.1073,
+      "step": 7704
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011592859187094897,
+      "loss": 0.1559,
+      "step": 7705
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011591049132700373,
+      "loss": 0.2481,
+      "step": 7706
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011589239024822926,
+      "loss": 0.129,
+      "step": 7707
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011587428863523409,
+      "loss": 0.1179,
+      "step": 7708
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011585618648862662,
+      "loss": 0.1034,
+      "step": 7709
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011583808380901544,
+      "loss": 0.1883,
+      "step": 7710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011581998059700899,
+      "loss": 0.0995,
+      "step": 7711
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011580187685321583,
+      "loss": 0.0934,
+      "step": 7712
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011578377257824452,
+      "loss": 0.0628,
+      "step": 7713
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011576566777270363,
+      "loss": 0.1996,
+      "step": 7714
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001157475624372018,
+      "loss": 0.1906,
+      "step": 7715
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011572945657234755,
+      "loss": 0.1676,
+      "step": 7716
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001157113501787496,
+      "loss": 0.0753,
+      "step": 7717
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011569324325701649,
+      "loss": 0.0907,
+      "step": 7718
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011567513580775697,
+      "loss": 0.1163,
+      "step": 7719
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011565702783157967,
+      "loss": 0.1652,
+      "step": 7720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001156389193290933,
+      "loss": 0.1007,
+      "step": 7721
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011562081030090661,
+      "loss": 0.1938,
+      "step": 7722
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011560270074762831,
+      "loss": 0.3601,
+      "step": 7723
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011558459066986713,
+      "loss": 0.2994,
+      "step": 7724
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011556648006823183,
+      "loss": 0.0934,
+      "step": 7725
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011554836894333123,
+      "loss": 0.2237,
+      "step": 7726
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011553025729577411,
+      "loss": 0.2074,
+      "step": 7727
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001155121451261693,
+      "loss": 0.1482,
+      "step": 7728
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011549403243512566,
+      "loss": 0.1323,
+      "step": 7729
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011547591922325203,
+      "loss": 0.0944,
+      "step": 7730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011545780549115728,
+      "loss": 0.1307,
+      "step": 7731
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011543969123945026,
+      "loss": 0.1517,
+      "step": 7732
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011542157646873996,
+      "loss": 0.142,
+      "step": 7733
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011540346117963523,
+      "loss": 0.2472,
+      "step": 7734
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011538534537274506,
+      "loss": 0.1467,
+      "step": 7735
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011536722904867843,
+      "loss": 0.2728,
+      "step": 7736
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011534911220804426,
+      "loss": 0.0963,
+      "step": 7737
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011533099485145155,
+      "loss": 0.1988,
+      "step": 7738
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011531287697950934,
+      "loss": 0.1304,
+      "step": 7739
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011529475859282668,
+      "loss": 0.1145,
+      "step": 7740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011527663969201256,
+      "loss": 0.2408,
+      "step": 7741
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011525852027767607,
+      "loss": 0.2107,
+      "step": 7742
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011524040035042636,
+      "loss": 0.0797,
+      "step": 7743
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011522227991087242,
+      "loss": 0.1874,
+      "step": 7744
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011520415895962342,
+      "loss": 0.1547,
+      "step": 7745
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011518603749728847,
+      "loss": 0.0835,
+      "step": 7746
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011516791552447676,
+      "loss": 0.1074,
+      "step": 7747
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011514979304179742,
+      "loss": 0.1564,
+      "step": 7748
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011513167004985965,
+      "loss": 0.1133,
+      "step": 7749
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011511354654927268,
+      "loss": 0.2356,
+      "step": 7750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011509542254064567,
+      "loss": 0.1116,
+      "step": 7751
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011507729802458791,
+      "loss": 0.292,
+      "step": 7752
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001150591730017086,
+      "loss": 0.1085,
+      "step": 7753
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011504104747261706,
+      "loss": 0.1271,
+      "step": 7754
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001150229214379226,
+      "loss": 0.0814,
+      "step": 7755
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011500479489823442,
+      "loss": 0.0768,
+      "step": 7756
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011498666785416195,
+      "loss": 0.1869,
+      "step": 7757
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00011496854030631443,
+      "loss": 0.1076,
+      "step": 7758
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011495041225530132,
+      "loss": 0.1086,
+      "step": 7759
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001149322837017319,
+      "loss": 0.0659,
+      "step": 7760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011491415464621562,
+      "loss": 0.1006,
+      "step": 7761
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011489602508936188,
+      "loss": 0.0927,
+      "step": 7762
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011487789503178009,
+      "loss": 0.2002,
+      "step": 7763
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011485976447407967,
+      "loss": 0.0912,
+      "step": 7764
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011484163341687008,
+      "loss": 0.2407,
+      "step": 7765
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011482350186076083,
+      "loss": 0.1185,
+      "step": 7766
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011480536980636138,
+      "loss": 0.1003,
+      "step": 7767
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011478723725428127,
+      "loss": 0.1341,
+      "step": 7768
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011476910420512998,
+      "loss": 0.1819,
+      "step": 7769
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011475097065951705,
+      "loss": 0.0844,
+      "step": 7770
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011473283661805207,
+      "loss": 0.0863,
+      "step": 7771
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001147147020813446,
+      "loss": 0.0737,
+      "step": 7772
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011469656705000424,
+      "loss": 0.127,
+      "step": 7773
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011467843152464057,
+      "loss": 0.1854,
+      "step": 7774
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011466029550586323,
+      "loss": 0.151,
+      "step": 7775
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011464215899428188,
+      "loss": 0.1907,
+      "step": 7776
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011462402199050613,
+      "loss": 0.1717,
+      "step": 7777
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011460588449514571,
+      "loss": 0.1857,
+      "step": 7778
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011458774650881023,
+      "loss": 0.091,
+      "step": 7779
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001145696080321095,
+      "loss": 0.1425,
+      "step": 7780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011455146906565318,
+      "loss": 0.0811,
+      "step": 7781
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011453332961005096,
+      "loss": 0.1176,
+      "step": 7782
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011451518966591272,
+      "loss": 0.0919,
+      "step": 7783
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011449704923384812,
+      "loss": 0.1039,
+      "step": 7784
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011447890831446701,
+      "loss": 0.1238,
+      "step": 7785
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011446076690837917,
+      "loss": 0.2061,
+      "step": 7786
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011444262501619445,
+      "loss": 0.0624,
+      "step": 7787
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011442448263852265,
+      "loss": 0.1635,
+      "step": 7788
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011440633977597363,
+      "loss": 0.1297,
+      "step": 7789
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001143881964291573,
+      "loss": 0.1535,
+      "step": 7790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001143700525986835,
+      "loss": 0.1061,
+      "step": 7791
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011435190828516216,
+      "loss": 0.1888,
+      "step": 7792
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011433376348920315,
+      "loss": 0.2066,
+      "step": 7793
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001143156182114165,
+      "loss": 0.1579,
+      "step": 7794
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011429747245241207,
+      "loss": 0.236,
+      "step": 7795
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011427932621279986,
+      "loss": 0.0723,
+      "step": 7796
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011426117949318987,
+      "loss": 0.104,
+      "step": 7797
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011424303229419207,
+      "loss": 0.1352,
+      "step": 7798
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011422488461641649,
+      "loss": 0.1137,
+      "step": 7799
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011420673646047317,
+      "loss": 0.0956,
+      "step": 7800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011418858782697217,
+      "loss": 0.0675,
+      "step": 7801
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001141704387165235,
+      "loss": 0.0839,
+      "step": 7802
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001141522891297373,
+      "loss": 0.1,
+      "step": 7803
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011413413906722363,
+      "loss": 0.1471,
+      "step": 7804
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011411598852959262,
+      "loss": 0.0615,
+      "step": 7805
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001140978375174544,
+      "loss": 0.1238,
+      "step": 7806
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001140796860314191,
+      "loss": 0.1043,
+      "step": 7807
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011406153407209688,
+      "loss": 0.1063,
+      "step": 7808
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011404338164009791,
+      "loss": 0.2133,
+      "step": 7809
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011402522873603241,
+      "loss": 0.1424,
+      "step": 7810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011400707536051056,
+      "loss": 0.1963,
+      "step": 7811
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001139889215141426,
+      "loss": 0.1172,
+      "step": 7812
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011397076719753878,
+      "loss": 0.198,
+      "step": 7813
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011395261241130932,
+      "loss": 0.0671,
+      "step": 7814
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00011393445715606452,
+      "loss": 0.1166,
+      "step": 7815
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011391630143241466,
+      "loss": 0.2003,
+      "step": 7816
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011389814524097002,
+      "loss": 0.07,
+      "step": 7817
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011387998858234095,
+      "loss": 0.1011,
+      "step": 7818
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011386183145713776,
+      "loss": 0.1267,
+      "step": 7819
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011384367386597081,
+      "loss": 0.1051,
+      "step": 7820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011382551580945049,
+      "loss": 0.2446,
+      "step": 7821
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011380735728818712,
+      "loss": 0.0694,
+      "step": 7822
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011378919830279118,
+      "loss": 0.1628,
+      "step": 7823
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011377103885387299,
+      "loss": 0.1038,
+      "step": 7824
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011375287894204305,
+      "loss": 0.1314,
+      "step": 7825
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011373471856791176,
+      "loss": 0.1392,
+      "step": 7826
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001137165577320896,
+      "loss": 0.1069,
+      "step": 7827
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011369839643518705,
+      "loss": 0.1021,
+      "step": 7828
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011368023467781457,
+      "loss": 0.1421,
+      "step": 7829
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001136620724605827,
+      "loss": 0.1728,
+      "step": 7830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001136439097841019,
+      "loss": 0.0851,
+      "step": 7831
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011362574664898282,
+      "loss": 0.1429,
+      "step": 7832
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001136075830558359,
+      "loss": 0.0975,
+      "step": 7833
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011358941900527176,
+      "loss": 0.1664,
+      "step": 7834
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011357125449790098,
+      "loss": 0.2538,
+      "step": 7835
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011355308953433412,
+      "loss": 0.1516,
+      "step": 7836
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011353492411518185,
+      "loss": 0.0831,
+      "step": 7837
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011351675824105474,
+      "loss": 0.0635,
+      "step": 7838
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011349859191256351,
+      "loss": 0.072,
+      "step": 7839
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011348042513031876,
+      "loss": 0.1184,
+      "step": 7840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011346225789493115,
+      "loss": 0.1044,
+      "step": 7841
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011344409020701141,
+      "loss": 0.2155,
+      "step": 7842
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011342592206717022,
+      "loss": 0.2115,
+      "step": 7843
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011340775347601833,
+      "loss": 0.2,
+      "step": 7844
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011338958443416645,
+      "loss": 0.213,
+      "step": 7845
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011337141494222534,
+      "loss": 0.1928,
+      "step": 7846
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011335324500080577,
+      "loss": 0.0951,
+      "step": 7847
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011333507461051849,
+      "loss": 0.0931,
+      "step": 7848
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011331690377197433,
+      "loss": 0.0885,
+      "step": 7849
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011329873248578406,
+      "loss": 0.1625,
+      "step": 7850
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011328056075255858,
+      "loss": 0.0896,
+      "step": 7851
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011326238857290867,
+      "loss": 0.1484,
+      "step": 7852
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011324421594744516,
+      "loss": 0.1031,
+      "step": 7853
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011322604287677901,
+      "loss": 0.0853,
+      "step": 7854
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000113207869361521,
+      "loss": 0.1993,
+      "step": 7855
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011318969540228215,
+      "loss": 0.091,
+      "step": 7856
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011317152099967327,
+      "loss": 0.0712,
+      "step": 7857
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011315334615430535,
+      "loss": 0.1452,
+      "step": 7858
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011313517086678933,
+      "loss": 0.1514,
+      "step": 7859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011311699513773612,
+      "loss": 0.152,
+      "step": 7860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011309881896775676,
+      "loss": 0.1226,
+      "step": 7861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011308064235746218,
+      "loss": 0.1735,
+      "step": 7862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011306246530746345,
+      "loss": 0.064,
+      "step": 7863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011304428781837154,
+      "loss": 0.0849,
+      "step": 7864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001130261098907975,
+      "loss": 0.1288,
+      "step": 7865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011300793152535239,
+      "loss": 0.0747,
+      "step": 7866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011298975272264722,
+      "loss": 0.1032,
+      "step": 7867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011297157348329315,
+      "loss": 0.1765,
+      "step": 7868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011295339380790122,
+      "loss": 0.1277,
+      "step": 7869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011293521369708256,
+      "loss": 0.2183,
+      "step": 7870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011291703315144826,
+      "loss": 0.1067,
+      "step": 7871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001128988521716095,
+      "loss": 0.1347,
+      "step": 7872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00011288067075817741,
+      "loss": 0.0795,
+      "step": 7873
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011286248891176314,
+      "loss": 0.1232,
+      "step": 7874
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011284430663297791,
+      "loss": 0.1023,
+      "step": 7875
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011282612392243286,
+      "loss": 0.1995,
+      "step": 7876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011280794078073926,
+      "loss": 0.1151,
+      "step": 7877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011278975720850833,
+      "loss": 0.1075,
+      "step": 7878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011277157320635123,
+      "loss": 0.1429,
+      "step": 7879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011275338877487928,
+      "loss": 0.2369,
+      "step": 7880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011273520391470376,
+      "loss": 0.5547,
+      "step": 7881
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011271701862643591,
+      "loss": 0.1319,
+      "step": 7882
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011269883291068703,
+      "loss": 0.1991,
+      "step": 7883
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011268064676806846,
+      "loss": 0.1477,
+      "step": 7884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011266246019919154,
+      "loss": 0.1025,
+      "step": 7885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011264427320466751,
+      "loss": 0.2686,
+      "step": 7886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011262608578510785,
+      "loss": 0.2291,
+      "step": 7887
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011260789794112383,
+      "loss": 0.1571,
+      "step": 7888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011258970967332689,
+      "loss": 0.1162,
+      "step": 7889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001125715209823284,
+      "loss": 0.0828,
+      "step": 7890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011255333186873977,
+      "loss": 0.1289,
+      "step": 7891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011253514233317246,
+      "loss": 0.0772,
+      "step": 7892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011251695237623785,
+      "loss": 0.0733,
+      "step": 7893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011249876199854745,
+      "loss": 0.3056,
+      "step": 7894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011248057120071269,
+      "loss": 0.1053,
+      "step": 7895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011246237998334508,
+      "loss": 0.1159,
+      "step": 7896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001124441883470561,
+      "loss": 0.1723,
+      "step": 7897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011242599629245722,
+      "loss": 0.1135,
+      "step": 7898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011240780382016005,
+      "loss": 0.1461,
+      "step": 7899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011238961093077604,
+      "loss": 0.0808,
+      "step": 7900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011237141762491683,
+      "loss": 0.0878,
+      "step": 7901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011235322390319389,
+      "loss": 0.0925,
+      "step": 7902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011233502976621888,
+      "loss": 0.1411,
+      "step": 7903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011231683521460337,
+      "loss": 0.1302,
+      "step": 7904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011229864024895891,
+      "loss": 0.1096,
+      "step": 7905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011228044486989722,
+      "loss": 0.094,
+      "step": 7906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011226224907802985,
+      "loss": 0.2588,
+      "step": 7907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011224405287396851,
+      "loss": 0.1116,
+      "step": 7908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011222585625832483,
+      "loss": 0.1386,
+      "step": 7909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011220765923171053,
+      "loss": 0.095,
+      "step": 7910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011218946179473722,
+      "loss": 0.0948,
+      "step": 7911
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011217126394801665,
+      "loss": 0.1668,
+      "step": 7912
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011215306569216056,
+      "loss": 0.1291,
+      "step": 7913
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011213486702778062,
+      "loss": 0.1313,
+      "step": 7914
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011211666795548867,
+      "loss": 0.1136,
+      "step": 7915
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011209846847589638,
+      "loss": 0.1317,
+      "step": 7916
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011208026858961562,
+      "loss": 0.1532,
+      "step": 7917
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011206206829725805,
+      "loss": 0.143,
+      "step": 7918
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011204386759943554,
+      "loss": 0.0857,
+      "step": 7919
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011202566649675996,
+      "loss": 0.1409,
+      "step": 7920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011200746498984303,
+      "loss": 0.0953,
+      "step": 7921
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011198926307929664,
+      "loss": 0.2309,
+      "step": 7922
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001119710607657327,
+      "loss": 0.1165,
+      "step": 7923
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011195285804976298,
+      "loss": 0.1942,
+      "step": 7924
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001119346549319994,
+      "loss": 0.1092,
+      "step": 7925
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011191645141305387,
+      "loss": 0.15,
+      "step": 7926
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001118982474935383,
+      "loss": 0.0965,
+      "step": 7927
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001118800431740646,
+      "loss": 0.1113,
+      "step": 7928
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011186183845524472,
+      "loss": 0.2134,
+      "step": 7929
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001118436333376906,
+      "loss": 0.1696,
+      "step": 7930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011182542782201419,
+      "loss": 0.1132,
+      "step": 7931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011180722190882746,
+      "loss": 0.1577,
+      "step": 7932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011178901559874244,
+      "loss": 0.0706,
+      "step": 7933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001117708088923711,
+      "loss": 0.1299,
+      "step": 7934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011175260179032547,
+      "loss": 0.0851,
+      "step": 7935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011173439429321762,
+      "loss": 0.1215,
+      "step": 7936
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011171618640165949,
+      "loss": 0.2689,
+      "step": 7937
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001116979781162632,
+      "loss": 0.1108,
+      "step": 7938
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011167976943764085,
+      "loss": 0.1021,
+      "step": 7939
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011166156036640446,
+      "loss": 0.1722,
+      "step": 7940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011164335090316616,
+      "loss": 0.2344,
+      "step": 7941
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011162514104853803,
+      "loss": 0.0775,
+      "step": 7942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011160693080313227,
+      "loss": 0.1954,
+      "step": 7943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001115887201675609,
+      "loss": 0.0917,
+      "step": 7944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011157050914243614,
+      "loss": 0.1788,
+      "step": 7945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011155229772837015,
+      "loss": 0.107,
+      "step": 7946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011153408592597507,
+      "loss": 0.0664,
+      "step": 7947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011151587373586314,
+      "loss": 0.1544,
+      "step": 7948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011149766115864653,
+      "loss": 0.1576,
+      "step": 7949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011147944819493746,
+      "loss": 0.1077,
+      "step": 7950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011146123484534812,
+      "loss": 0.1126,
+      "step": 7951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011144302111049078,
+      "loss": 0.0977,
+      "step": 7952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011142480699097773,
+      "loss": 0.3797,
+      "step": 7953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011140659248742116,
+      "loss": 0.105,
+      "step": 7954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011138837760043344,
+      "loss": 0.1083,
+      "step": 7955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011137016233062674,
+      "loss": 0.1039,
+      "step": 7956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011135194667861348,
+      "loss": 0.0864,
+      "step": 7957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001113337306450059,
+      "loss": 0.0745,
+      "step": 7958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011131551423041635,
+      "loss": 0.0907,
+      "step": 7959
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001112972974354572,
+      "loss": 0.1442,
+      "step": 7960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011127908026074079,
+      "loss": 0.1891,
+      "step": 7961
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001112608627068795,
+      "loss": 0.1126,
+      "step": 7962
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011124264477448568,
+      "loss": 0.0722,
+      "step": 7963
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011122442646417172,
+      "loss": 0.0858,
+      "step": 7964
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011120620777655007,
+      "loss": 0.1185,
+      "step": 7965
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011118798871223309,
+      "loss": 0.0874,
+      "step": 7966
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011116976927183327,
+      "loss": 0.19,
+      "step": 7967
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011115154945596305,
+      "loss": 0.0934,
+      "step": 7968
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011113332926523484,
+      "loss": 0.1181,
+      "step": 7969
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011111510870026112,
+      "loss": 0.1316,
+      "step": 7970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011109688776165441,
+      "loss": 0.2608,
+      "step": 7971
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011107866645002719,
+      "loss": 0.1263,
+      "step": 7972
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011106044476599194,
+      "loss": 0.1248,
+      "step": 7973
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001110422227101612,
+      "loss": 0.0685,
+      "step": 7974
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011102400028314753,
+      "loss": 0.0769,
+      "step": 7975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011100577748556342,
+      "loss": 0.1082,
+      "step": 7976
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011098755431802146,
+      "loss": 0.1189,
+      "step": 7977
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011096933078113423,
+      "loss": 0.078,
+      "step": 7978
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011095110687551426,
+      "loss": 0.1129,
+      "step": 7979
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011093288260177419,
+      "loss": 0.1099,
+      "step": 7980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011091465796052664,
+      "loss": 0.1137,
+      "step": 7981
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011089643295238419,
+      "loss": 0.0923,
+      "step": 7982
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011087820757795948,
+      "loss": 0.0619,
+      "step": 7983
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011085998183786516,
+      "loss": 0.0924,
+      "step": 7984
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011084175573271391,
+      "loss": 0.4352,
+      "step": 7985
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011082352926311834,
+      "loss": 0.1117,
+      "step": 7986
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011080530242969116,
+      "loss": 0.2546,
+      "step": 7987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011078707523304512,
+      "loss": 0.1638,
+      "step": 7988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011076884767379283,
+      "loss": 0.1419,
+      "step": 7989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011075061975254707,
+      "loss": 0.127,
+      "step": 7990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011073239146992054,
+      "loss": 0.0901,
+      "step": 7991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000110714162826526,
+      "loss": 0.0864,
+      "step": 7992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011069593382297616,
+      "loss": 0.0709,
+      "step": 7993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011067770445988386,
+      "loss": 0.1195,
+      "step": 7994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011065947473786184,
+      "loss": 0.0818,
+      "step": 7995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011064124465752285,
+      "loss": 0.0957,
+      "step": 7996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011062301421947976,
+      "loss": 0.1967,
+      "step": 7997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011060478342434532,
+      "loss": 0.1846,
+      "step": 7998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011058655227273243,
+      "loss": 0.1036,
+      "step": 7999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011056832076525388,
+      "loss": 0.108,
+      "step": 8000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011055008890252251,
+      "loss": 0.1029,
+      "step": 8001
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011053185668515122,
+      "loss": 0.0882,
+      "step": 8002
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011051362411375283,
+      "loss": 0.1669,
+      "step": 8003
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001104953911889403,
+      "loss": 0.0818,
+      "step": 8004
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011047715791132645,
+      "loss": 0.1111,
+      "step": 8005
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011045892428152422,
+      "loss": 0.2693,
+      "step": 8006
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011044069030014659,
+      "loss": 0.1007,
+      "step": 8007
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011042245596780641,
+      "loss": 0.0909,
+      "step": 8008
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011040422128511665,
+      "loss": 0.2545,
+      "step": 8009
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011038598625269026,
+      "loss": 0.1878,
+      "step": 8010
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011036775087114024,
+      "loss": 0.1935,
+      "step": 8011
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011034951514107952,
+      "loss": 0.0776,
+      "step": 8012
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011033127906312115,
+      "loss": 0.2426,
+      "step": 8013
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011031304263787812,
+      "loss": 0.0997,
+      "step": 8014
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001102948058659634,
+      "loss": 0.0599,
+      "step": 8015
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011027656874799004,
+      "loss": 0.1108,
+      "step": 8016
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011025833128457108,
+      "loss": 0.1609,
+      "step": 8017
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011024009347631961,
+      "loss": 0.1261,
+      "step": 8018
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011022185532384863,
+      "loss": 0.0648,
+      "step": 8019
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011020361682777128,
+      "loss": 0.0919,
+      "step": 8020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011018537798870056,
+      "loss": 0.157,
+      "step": 8021
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011016713880724962,
+      "loss": 0.1145,
+      "step": 8022
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011014889928403158,
+      "loss": 0.2488,
+      "step": 8023
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011013065941965952,
+      "loss": 0.0934,
+      "step": 8024
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011011241921474662,
+      "loss": 0.0967,
+      "step": 8025
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011009417866990597,
+      "loss": 0.1458,
+      "step": 8026
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011007593778575074,
+      "loss": 0.1148,
+      "step": 8027
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011005769656289411,
+      "loss": 0.0727,
+      "step": 8028
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011003945500194923,
+      "loss": 0.1301,
+      "step": 8029
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011002121310352933,
+      "loss": 0.0845,
+      "step": 8030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011000297086824755,
+      "loss": 0.0916,
+      "step": 8031
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010998472829671716,
+      "loss": 0.148,
+      "step": 8032
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010996648538955137,
+      "loss": 0.1134,
+      "step": 8033
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010994824214736337,
+      "loss": 0.1621,
+      "step": 8034
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010992999857076645,
+      "loss": 0.2385,
+      "step": 8035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010991175466037382,
+      "loss": 0.1818,
+      "step": 8036
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001098935104167988,
+      "loss": 0.1337,
+      "step": 8037
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010987526584065464,
+      "loss": 0.0822,
+      "step": 8038
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010985702093255466,
+      "loss": 0.1423,
+      "step": 8039
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010983877569311212,
+      "loss": 0.1246,
+      "step": 8040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010982053012294031,
+      "loss": 0.1412,
+      "step": 8041
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010980228422265262,
+      "loss": 0.0894,
+      "step": 8042
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010978403799286235,
+      "loss": 0.0845,
+      "step": 8043
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00010976579143418284,
+      "loss": 0.096,
+      "step": 8044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010974754454722744,
+      "loss": 0.146,
+      "step": 8045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010972929733260955,
+      "loss": 0.1366,
+      "step": 8046
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010971104979094254,
+      "loss": 0.1611,
+      "step": 8047
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010969280192283974,
+      "loss": 0.0829,
+      "step": 8048
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010967455372891464,
+      "loss": 0.1396,
+      "step": 8049
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010965630520978057,
+      "loss": 0.2391,
+      "step": 8050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000109638056366051,
+      "loss": 0.0841,
+      "step": 8051
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010961980719833937,
+      "loss": 0.1646,
+      "step": 8052
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010960155770725909,
+      "loss": 0.0776,
+      "step": 8053
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010958330789342363,
+      "loss": 0.1767,
+      "step": 8054
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010956505775744643,
+      "loss": 0.0906,
+      "step": 8055
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010954680729994102,
+      "loss": 0.1351,
+      "step": 8056
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010952855652152083,
+      "loss": 0.1024,
+      "step": 8057
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010951030542279942,
+      "loss": 0.0745,
+      "step": 8058
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010949205400439026,
+      "loss": 0.1125,
+      "step": 8059
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010947380226690684,
+      "loss": 0.18,
+      "step": 8060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010945555021096274,
+      "loss": 0.0959,
+      "step": 8061
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010943729783717147,
+      "loss": 0.3309,
+      "step": 8062
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010941904514614663,
+      "loss": 0.2018,
+      "step": 8063
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001094007921385017,
+      "loss": 0.1217,
+      "step": 8064
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010938253881485033,
+      "loss": 0.1365,
+      "step": 8065
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010936428517580608,
+      "loss": 0.1926,
+      "step": 8066
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001093460312219825,
+      "loss": 0.0883,
+      "step": 8067
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010932777695399325,
+      "loss": 0.2094,
+      "step": 8068
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010930952237245191,
+      "loss": 0.0948,
+      "step": 8069
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010929126747797214,
+      "loss": 0.0967,
+      "step": 8070
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010927301227116753,
+      "loss": 0.126,
+      "step": 8071
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001092547567526518,
+      "loss": 0.0697,
+      "step": 8072
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010923650092303855,
+      "loss": 0.0885,
+      "step": 8073
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001092182447829414,
+      "loss": 0.1157,
+      "step": 8074
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010919998833297413,
+      "loss": 0.2486,
+      "step": 8075
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010918173157375037,
+      "loss": 0.0887,
+      "step": 8076
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010916347450588387,
+      "loss": 0.1106,
+      "step": 8077
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010914521712998826,
+      "loss": 0.1411,
+      "step": 8078
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001091269594466773,
+      "loss": 0.2161,
+      "step": 8079
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010910870145656475,
+      "loss": 0.1037,
+      "step": 8080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010909044316026429,
+      "loss": 0.1647,
+      "step": 8081
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010907218455838972,
+      "loss": 0.153,
+      "step": 8082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010905392565155477,
+      "loss": 0.209,
+      "step": 8083
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010903566644037324,
+      "loss": 0.0687,
+      "step": 8084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001090174069254589,
+      "loss": 0.1099,
+      "step": 8085
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010899914710742549,
+      "loss": 0.1266,
+      "step": 8086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010898088698688691,
+      "loss": 0.1487,
+      "step": 8087
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010896262656445688,
+      "loss": 0.1648,
+      "step": 8088
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010894436584074927,
+      "loss": 0.1369,
+      "step": 8089
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001089261048163779,
+      "loss": 0.1322,
+      "step": 8090
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010890784349195664,
+      "loss": 0.0759,
+      "step": 8091
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001088895818680993,
+      "loss": 0.0918,
+      "step": 8092
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010887131994541975,
+      "loss": 0.1127,
+      "step": 8093
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001088530577245319,
+      "loss": 0.0886,
+      "step": 8094
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010883479520604955,
+      "loss": 0.0949,
+      "step": 8095
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010881653239058669,
+      "loss": 0.0787,
+      "step": 8096
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010879826927875718,
+      "loss": 0.111,
+      "step": 8097
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001087800058711749,
+      "loss": 0.1529,
+      "step": 8098
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010876174216845383,
+      "loss": 0.1526,
+      "step": 8099
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010874347817120784,
+      "loss": 0.1704,
+      "step": 8100
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00010872521388005094,
+      "loss": 0.0793,
+      "step": 8101
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010870694929559702,
+      "loss": 0.2575,
+      "step": 8102
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001086886844184601,
+      "loss": 0.1271,
+      "step": 8103
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010867041924925412,
+      "loss": 0.1562,
+      "step": 8104
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010865215378859304,
+      "loss": 0.0903,
+      "step": 8105
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010863388803709089,
+      "loss": 0.0776,
+      "step": 8106
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010861562199536163,
+      "loss": 0.0865,
+      "step": 8107
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010859735566401934,
+      "loss": 0.0645,
+      "step": 8108
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010857908904367797,
+      "loss": 0.2245,
+      "step": 8109
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010856082213495158,
+      "loss": 0.158,
+      "step": 8110
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010854255493845422,
+      "loss": 0.1093,
+      "step": 8111
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001085242874547999,
+      "loss": 0.144,
+      "step": 8112
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010850601968460275,
+      "loss": 0.1444,
+      "step": 8113
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010848775162847675,
+      "loss": 0.137,
+      "step": 8114
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010846948328703604,
+      "loss": 0.2837,
+      "step": 8115
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010845121466089467,
+      "loss": 0.309,
+      "step": 8116
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010843294575066681,
+      "loss": 0.0833,
+      "step": 8117
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010841467655696649,
+      "loss": 0.2649,
+      "step": 8118
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010839640708040782,
+      "loss": 0.1073,
+      "step": 8119
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000108378137321605,
+      "loss": 0.0876,
+      "step": 8120
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010835986728117209,
+      "loss": 0.08,
+      "step": 8121
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001083415969597233,
+      "loss": 0.1077,
+      "step": 8122
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010832332635787277,
+      "loss": 0.0771,
+      "step": 8123
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001083050554762346,
+      "loss": 0.1727,
+      "step": 8124
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010828678431542303,
+      "loss": 0.1292,
+      "step": 8125
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010826851287605222,
+      "loss": 0.0805,
+      "step": 8126
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010825024115873639,
+      "loss": 0.1755,
+      "step": 8127
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010823196916408968,
+      "loss": 0.2585,
+      "step": 8128
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010821369689272638,
+      "loss": 0.0752,
+      "step": 8129
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010819542434526065,
+      "loss": 0.2632,
+      "step": 8130
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010817715152230672,
+      "loss": 0.0791,
+      "step": 8131
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010815887842447887,
+      "loss": 0.0847,
+      "step": 8132
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010814060505239132,
+      "loss": 0.0762,
+      "step": 8133
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010812233140665835,
+      "loss": 0.1401,
+      "step": 8134
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010810405748789416,
+      "loss": 0.1284,
+      "step": 8135
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010808578329671313,
+      "loss": 0.2026,
+      "step": 8136
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010806750883372947,
+      "loss": 0.1859,
+      "step": 8137
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010804923409955747,
+      "loss": 0.0594,
+      "step": 8138
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010803095909481149,
+      "loss": 0.185,
+      "step": 8139
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010801268382010579,
+      "loss": 0.1238,
+      "step": 8140
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010799440827605471,
+      "loss": 0.0739,
+      "step": 8141
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010797613246327257,
+      "loss": 0.115,
+      "step": 8142
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010795785638237372,
+      "loss": 0.1697,
+      "step": 8143
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010793958003397254,
+      "loss": 0.0689,
+      "step": 8144
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010792130341868331,
+      "loss": 0.1219,
+      "step": 8145
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010790302653712046,
+      "loss": 0.0688,
+      "step": 8146
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010788474938989835,
+      "loss": 0.2258,
+      "step": 8147
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010786647197763136,
+      "loss": 0.2743,
+      "step": 8148
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010784819430093386,
+      "loss": 0.1559,
+      "step": 8149
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010782991636042028,
+      "loss": 0.0684,
+      "step": 8150
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010781163815670505,
+      "loss": 0.073,
+      "step": 8151
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010779335969040252,
+      "loss": 0.1736,
+      "step": 8152
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010777508096212721,
+      "loss": 0.1186,
+      "step": 8153
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010775680197249348,
+      "loss": 0.0806,
+      "step": 8154
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010773852272211584,
+      "loss": 0.1876,
+      "step": 8155
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010772024321160867,
+      "loss": 0.1403,
+      "step": 8156
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010770196344158647,
+      "loss": 0.0778,
+      "step": 8157
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00010768368341266375,
+      "loss": 0.1552,
+      "step": 8158
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010766540312545493,
+      "loss": 0.2062,
+      "step": 8159
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010764712258057455,
+      "loss": 0.0703,
+      "step": 8160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010762884177863704,
+      "loss": 0.1184,
+      "step": 8161
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010761056072025702,
+      "loss": 0.0921,
+      "step": 8162
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010759227940604888,
+      "loss": 0.0754,
+      "step": 8163
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010757399783662723,
+      "loss": 0.1577,
+      "step": 8164
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010755571601260657,
+      "loss": 0.0809,
+      "step": 8165
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010753743393460142,
+      "loss": 0.0899,
+      "step": 8166
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010751915160322639,
+      "loss": 0.1608,
+      "step": 8167
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010750086901909599,
+      "loss": 0.2206,
+      "step": 8168
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010748258618282483,
+      "loss": 0.1652,
+      "step": 8169
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010746430309502742,
+      "loss": 0.0831,
+      "step": 8170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010744601975631837,
+      "loss": 0.1307,
+      "step": 8171
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001074277361673123,
+      "loss": 0.3662,
+      "step": 8172
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010740945232862378,
+      "loss": 0.1329,
+      "step": 8173
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010739116824086746,
+      "loss": 0.2514,
+      "step": 8174
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010737288390465792,
+      "loss": 0.1065,
+      "step": 8175
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010735459932060978,
+      "loss": 0.1619,
+      "step": 8176
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010733631448933769,
+      "loss": 0.1331,
+      "step": 8177
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010731802941145628,
+      "loss": 0.1085,
+      "step": 8178
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010729974408758024,
+      "loss": 0.0596,
+      "step": 8179
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010728145851832416,
+      "loss": 0.0943,
+      "step": 8180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010726317270430281,
+      "loss": 0.0938,
+      "step": 8181
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010724488664613077,
+      "loss": 0.0837,
+      "step": 8182
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010722660034442274,
+      "loss": 0.1539,
+      "step": 8183
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010720831379979344,
+      "loss": 0.149,
+      "step": 8184
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010719002701285754,
+      "loss": 0.0759,
+      "step": 8185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001071717399842298,
+      "loss": 0.0827,
+      "step": 8186
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010715345271452488,
+      "loss": 0.0973,
+      "step": 8187
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010713516520435756,
+      "loss": 0.1115,
+      "step": 8188
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010711687745434251,
+      "loss": 0.0449,
+      "step": 8189
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010709858946509448,
+      "loss": 0.0996,
+      "step": 8190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010708030123722828,
+      "loss": 0.1988,
+      "step": 8191
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010706201277135858,
+      "loss": 0.1333,
+      "step": 8192
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010704372406810023,
+      "loss": 0.1424,
+      "step": 8193
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010702543512806795,
+      "loss": 0.1136,
+      "step": 8194
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010700714595187655,
+      "loss": 0.1561,
+      "step": 8195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010698885654014078,
+      "loss": 0.0518,
+      "step": 8196
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010697056689347544,
+      "loss": 0.0885,
+      "step": 8197
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010695227701249537,
+      "loss": 0.1944,
+      "step": 8198
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010693398689781536,
+      "loss": 0.2166,
+      "step": 8199
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010691569655005023,
+      "loss": 0.2013,
+      "step": 8200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010689740596981483,
+      "loss": 0.1539,
+      "step": 8201
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010687911515772398,
+      "loss": 0.0709,
+      "step": 8202
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001068608241143925,
+      "loss": 0.1282,
+      "step": 8203
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010684253284043524,
+      "loss": 0.2058,
+      "step": 8204
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001068242413364671,
+      "loss": 0.1461,
+      "step": 8205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010680594960310293,
+      "loss": 0.2283,
+      "step": 8206
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010678765764095762,
+      "loss": 0.0736,
+      "step": 8207
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000106769365450646,
+      "loss": 0.1282,
+      "step": 8208
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010675107303278302,
+      "loss": 0.1819,
+      "step": 8209
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010673278038798353,
+      "loss": 0.145,
+      "step": 8210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010671448751686246,
+      "loss": 0.1104,
+      "step": 8211
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010669619442003471,
+      "loss": 0.0737,
+      "step": 8212
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010667790109811524,
+      "loss": 0.1418,
+      "step": 8213
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010665960755171893,
+      "loss": 0.0743,
+      "step": 8214
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00010664131378146072,
+      "loss": 0.0602,
+      "step": 8215
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010662301978795557,
+      "loss": 0.3726,
+      "step": 8216
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010660472557181842,
+      "loss": 0.174,
+      "step": 8217
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010658643113366422,
+      "loss": 0.1404,
+      "step": 8218
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010656813647410798,
+      "loss": 0.0773,
+      "step": 8219
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010654984159376465,
+      "loss": 0.0766,
+      "step": 8220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010653154649324917,
+      "loss": 0.137,
+      "step": 8221
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010651325117317654,
+      "loss": 0.1388,
+      "step": 8222
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010649495563416181,
+      "loss": 0.2117,
+      "step": 8223
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010647665987681992,
+      "loss": 0.2254,
+      "step": 8224
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010645836390176591,
+      "loss": 0.179,
+      "step": 8225
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010644006770961483,
+      "loss": 0.0799,
+      "step": 8226
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010642177130098161,
+      "loss": 0.0993,
+      "step": 8227
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010640347467648138,
+      "loss": 0.1494,
+      "step": 8228
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001063851778367291,
+      "loss": 0.2094,
+      "step": 8229
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010636688078233988,
+      "loss": 0.0714,
+      "step": 8230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010634858351392871,
+      "loss": 0.1916,
+      "step": 8231
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010633028603211071,
+      "loss": 0.2007,
+      "step": 8232
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010631198833750096,
+      "loss": 0.2486,
+      "step": 8233
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010629369043071444,
+      "loss": 0.136,
+      "step": 8234
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010627539231236632,
+      "loss": 0.0877,
+      "step": 8235
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010625709398307165,
+      "loss": 0.1264,
+      "step": 8236
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010623879544344555,
+      "loss": 0.1591,
+      "step": 8237
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010622049669410308,
+      "loss": 0.1808,
+      "step": 8238
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001062021977356594,
+      "loss": 0.1138,
+      "step": 8239
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001061838985687296,
+      "loss": 0.1016,
+      "step": 8240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001061655991939288,
+      "loss": 0.1025,
+      "step": 8241
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010614729961187218,
+      "loss": 0.0809,
+      "step": 8242
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010612899982317478,
+      "loss": 0.074,
+      "step": 8243
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010611069982845183,
+      "loss": 0.0967,
+      "step": 8244
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010609239962831847,
+      "loss": 0.1375,
+      "step": 8245
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010607409922338987,
+      "loss": 0.1216,
+      "step": 8246
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010605579861428114,
+      "loss": 0.1549,
+      "step": 8247
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010603749780160747,
+      "loss": 0.1046,
+      "step": 8248
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010601919678598409,
+      "loss": 0.0902,
+      "step": 8249
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001060008955680261,
+      "loss": 0.0913,
+      "step": 8250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010598259414834877,
+      "loss": 0.1138,
+      "step": 8251
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001059642925275673,
+      "loss": 0.08,
+      "step": 8252
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010594599070629683,
+      "loss": 0.1085,
+      "step": 8253
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010592768868515264,
+      "loss": 0.0855,
+      "step": 8254
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010590938646474991,
+      "loss": 0.0707,
+      "step": 8255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010589108404570389,
+      "loss": 0.098,
+      "step": 8256
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001058727814286298,
+      "loss": 0.0742,
+      "step": 8257
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010585447861414286,
+      "loss": 0.103,
+      "step": 8258
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001058361756028584,
+      "loss": 0.083,
+      "step": 8259
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010581787239539156,
+      "loss": 0.1298,
+      "step": 8260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010579956899235767,
+      "loss": 0.1615,
+      "step": 8261
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010578126539437199,
+      "loss": 0.172,
+      "step": 8262
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001057629616020498,
+      "loss": 0.1003,
+      "step": 8263
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010574465761600634,
+      "loss": 0.122,
+      "step": 8264
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010572635343685693,
+      "loss": 0.0857,
+      "step": 8265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010570804906521685,
+      "loss": 0.1229,
+      "step": 8266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010568974450170139,
+      "loss": 0.1682,
+      "step": 8267
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010567143974692591,
+      "loss": 0.0967,
+      "step": 8268
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010565313480150562,
+      "loss": 0.1041,
+      "step": 8269
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010563482966605593,
+      "loss": 0.096,
+      "step": 8270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010561652434119214,
+      "loss": 0.1202,
+      "step": 8271
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010559821882752958,
+      "loss": 0.088,
+      "step": 8272
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00010557991312568359,
+      "loss": 0.0902,
+      "step": 8273
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010556160723626946,
+      "loss": 0.0988,
+      "step": 8274
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010554330115990263,
+      "loss": 0.2315,
+      "step": 8275
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010552499489719839,
+      "loss": 0.1199,
+      "step": 8276
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010550668844877214,
+      "loss": 0.2577,
+      "step": 8277
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010548838181523923,
+      "loss": 0.1628,
+      "step": 8278
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010547007499721504,
+      "loss": 0.2991,
+      "step": 8279
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010545176799531496,
+      "loss": 0.1024,
+      "step": 8280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010543346081015435,
+      "loss": 0.0836,
+      "step": 8281
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010541515344234862,
+      "loss": 0.0625,
+      "step": 8282
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001053968458925132,
+      "loss": 0.1317,
+      "step": 8283
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010537853816126346,
+      "loss": 0.1397,
+      "step": 8284
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010536023024921482,
+      "loss": 0.0886,
+      "step": 8285
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010534192215698268,
+      "loss": 0.121,
+      "step": 8286
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010532361388518252,
+      "loss": 0.1119,
+      "step": 8287
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001053053054344297,
+      "loss": 0.0766,
+      "step": 8288
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010528699680533971,
+      "loss": 0.0886,
+      "step": 8289
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010526868799852796,
+      "loss": 0.1254,
+      "step": 8290
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010525037901460992,
+      "loss": 0.29,
+      "step": 8291
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010523206985420105,
+      "loss": 0.0994,
+      "step": 8292
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010521376051791676,
+      "loss": 0.2068,
+      "step": 8293
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010519545100637259,
+      "loss": 0.1552,
+      "step": 8294
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010517714132018392,
+      "loss": 0.1062,
+      "step": 8295
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010515883145996635,
+      "loss": 0.0559,
+      "step": 8296
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010514052142633526,
+      "loss": 0.137,
+      "step": 8297
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010512221121990615,
+      "loss": 0.0735,
+      "step": 8298
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010510390084129457,
+      "loss": 0.1203,
+      "step": 8299
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010508559029111598,
+      "loss": 0.1219,
+      "step": 8300
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010506727956998592,
+      "loss": 0.2158,
+      "step": 8301
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010504896867851986,
+      "loss": 0.0766,
+      "step": 8302
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010503065761733333,
+      "loss": 0.0634,
+      "step": 8303
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001050123463870419,
+      "loss": 0.1128,
+      "step": 8304
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010499403498826102,
+      "loss": 0.188,
+      "step": 8305
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010497572342160629,
+      "loss": 0.1739,
+      "step": 8306
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010495741168769322,
+      "loss": 0.0784,
+      "step": 8307
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010493909978713738,
+      "loss": 0.0879,
+      "step": 8308
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010492078772055426,
+      "loss": 0.056,
+      "step": 8309
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010490247548855954,
+      "loss": 0.2806,
+      "step": 8310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010488416309176867,
+      "loss": 0.3131,
+      "step": 8311
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010486585053079725,
+      "loss": 0.1326,
+      "step": 8312
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010484753780626089,
+      "loss": 0.0955,
+      "step": 8313
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010482922491877511,
+      "loss": 0.1125,
+      "step": 8314
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010481091186895556,
+      "loss": 0.3799,
+      "step": 8315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010479259865741777,
+      "loss": 0.1637,
+      "step": 8316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010477428528477739,
+      "loss": 0.1117,
+      "step": 8317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010475597175165,
+      "loss": 0.0726,
+      "step": 8318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010473765805865119,
+      "loss": 0.1297,
+      "step": 8319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010471934420639659,
+      "loss": 0.1018,
+      "step": 8320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010470103019550182,
+      "loss": 0.0896,
+      "step": 8321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010468271602658248,
+      "loss": 0.1217,
+      "step": 8322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010466440170025425,
+      "loss": 0.0642,
+      "step": 8323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001046460872171327,
+      "loss": 0.1095,
+      "step": 8324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010462777257783352,
+      "loss": 0.1566,
+      "step": 8325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010460945778297231,
+      "loss": 0.1297,
+      "step": 8326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010459114283316476,
+      "loss": 0.0853,
+      "step": 8327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001045728277290265,
+      "loss": 0.1793,
+      "step": 8328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001045545124711732,
+      "loss": 0.0726,
+      "step": 8329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00010453619706022055,
+      "loss": 0.1494,
+      "step": 8330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010451788149678417,
+      "loss": 0.068,
+      "step": 8331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010449956578147975,
+      "loss": 0.2605,
+      "step": 8332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010448124991492299,
+      "loss": 0.1565,
+      "step": 8333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010446293389772957,
+      "loss": 0.1027,
+      "step": 8334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010444461773051514,
+      "loss": 0.1121,
+      "step": 8335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010442630141389549,
+      "loss": 0.1333,
+      "step": 8336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010440798494848625,
+      "loss": 0.0827,
+      "step": 8337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010438966833490309,
+      "loss": 0.1953,
+      "step": 8338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010437135157376181,
+      "loss": 0.0976,
+      "step": 8339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010435303466567808,
+      "loss": 0.1985,
+      "step": 8340
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010433471761126763,
+      "loss": 0.1646,
+      "step": 8341
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010431640041114617,
+      "loss": 0.0774,
+      "step": 8342
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010429808306592943,
+      "loss": 0.4017,
+      "step": 8343
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010427976557623318,
+      "loss": 0.1067,
+      "step": 8344
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010426144794267312,
+      "loss": 0.2413,
+      "step": 8345
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010424313016586502,
+      "loss": 0.0912,
+      "step": 8346
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010422481224642462,
+      "loss": 0.091,
+      "step": 8347
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010420649418496769,
+      "loss": 0.1644,
+      "step": 8348
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010418817598210995,
+      "loss": 0.1621,
+      "step": 8349
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010416985763846719,
+      "loss": 0.0705,
+      "step": 8350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001041515391546552,
+      "loss": 0.143,
+      "step": 8351
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010413322053128971,
+      "loss": 0.1221,
+      "step": 8352
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010411490176898652,
+      "loss": 0.0842,
+      "step": 8353
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010409658286836143,
+      "loss": 0.129,
+      "step": 8354
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010407826383003021,
+      "loss": 0.1046,
+      "step": 8355
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010405994465460866,
+      "loss": 0.1774,
+      "step": 8356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010404162534271254,
+      "loss": 0.0588,
+      "step": 8357
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001040233058949577,
+      "loss": 0.2128,
+      "step": 8358
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010400498631195992,
+      "loss": 0.0865,
+      "step": 8359
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010398666659433502,
+      "loss": 0.0942,
+      "step": 8360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010396834674269882,
+      "loss": 0.1417,
+      "step": 8361
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010395002675766712,
+      "loss": 0.176,
+      "step": 8362
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010393170663985579,
+      "loss": 0.0831,
+      "step": 8363
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010391338638988057,
+      "loss": 0.2019,
+      "step": 8364
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010389506600835738,
+      "loss": 0.2015,
+      "step": 8365
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010387674549590202,
+      "loss": 0.0937,
+      "step": 8366
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010385842485313036,
+      "loss": 0.1108,
+      "step": 8367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001038401040806582,
+      "loss": 0.0764,
+      "step": 8368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010382178317910141,
+      "loss": 0.0939,
+      "step": 8369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010380346214907585,
+      "loss": 0.1339,
+      "step": 8370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010378514099119738,
+      "loss": 0.1726,
+      "step": 8371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010376681970608187,
+      "loss": 0.1351,
+      "step": 8372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010374849829434517,
+      "loss": 0.0805,
+      "step": 8373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010373017675660315,
+      "loss": 0.1066,
+      "step": 8374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010371185509347172,
+      "loss": 0.1321,
+      "step": 8375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010369353330556673,
+      "loss": 0.1567,
+      "step": 8376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010367521139350407,
+      "loss": 0.1126,
+      "step": 8377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001036568893578996,
+      "loss": 0.1039,
+      "step": 8378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010363856719936931,
+      "loss": 0.0997,
+      "step": 8379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010362024491852898,
+      "loss": 0.0876,
+      "step": 8380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010360192251599458,
+      "loss": 0.1025,
+      "step": 8381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000103583599992382,
+      "loss": 0.0831,
+      "step": 8382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010356527734830712,
+      "loss": 0.0711,
+      "step": 8383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010354695458438593,
+      "loss": 0.0674,
+      "step": 8384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010352863170123425,
+      "loss": 0.0968,
+      "step": 8385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00010351030869946808,
+      "loss": 0.0885,
+      "step": 8386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001034919855797033,
+      "loss": 0.1594,
+      "step": 8387
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010347366234255586,
+      "loss": 0.0547,
+      "step": 8388
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001034553389886417,
+      "loss": 0.0975,
+      "step": 8389
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010343701551857672,
+      "loss": 0.1086,
+      "step": 8390
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001034186919329769,
+      "loss": 0.0648,
+      "step": 8391
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010340036823245815,
+      "loss": 0.1561,
+      "step": 8392
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010338204441763649,
+      "loss": 0.0925,
+      "step": 8393
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001033637204891278,
+      "loss": 0.2123,
+      "step": 8394
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010334539644754802,
+      "loss": 0.1248,
+      "step": 8395
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001033270722935132,
+      "loss": 0.0938,
+      "step": 8396
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010330874802763921,
+      "loss": 0.0814,
+      "step": 8397
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010329042365054209,
+      "loss": 0.2373,
+      "step": 8398
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010327209916283778,
+      "loss": 0.1054,
+      "step": 8399
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010325377456514227,
+      "loss": 0.1125,
+      "step": 8400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010323544985807153,
+      "loss": 0.1646,
+      "step": 8401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001032171250422415,
+      "loss": 0.1108,
+      "step": 8402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010319880011826825,
+      "loss": 0.1179,
+      "step": 8403
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001031804750867677,
+      "loss": 0.0733,
+      "step": 8404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001031621499483559,
+      "loss": 0.1794,
+      "step": 8405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010314382470364878,
+      "loss": 0.193,
+      "step": 8406
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010312549935326243,
+      "loss": 0.1158,
+      "step": 8407
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010310717389781278,
+      "loss": 0.1912,
+      "step": 8408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010308884833791584,
+      "loss": 0.0891,
+      "step": 8409
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010307052267418766,
+      "loss": 0.1115,
+      "step": 8410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010305219690724424,
+      "loss": 0.1939,
+      "step": 8411
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010303387103770159,
+      "loss": 0.1674,
+      "step": 8412
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010301554506617575,
+      "loss": 0.0948,
+      "step": 8413
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010299721899328274,
+      "loss": 0.1624,
+      "step": 8414
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010297889281963855,
+      "loss": 0.1008,
+      "step": 8415
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010296056654585926,
+      "loss": 0.121,
+      "step": 8416
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001029422401725609,
+      "loss": 0.0665,
+      "step": 8417
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010292391370035948,
+      "loss": 0.0869,
+      "step": 8418
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010290558712987107,
+      "loss": 0.0607,
+      "step": 8419
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010288726046171171,
+      "loss": 0.139,
+      "step": 8420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010286893369649746,
+      "loss": 0.0503,
+      "step": 8421
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010285060683484433,
+      "loss": 0.085,
+      "step": 8422
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010283227987736839,
+      "loss": 0.0857,
+      "step": 8423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010281395282468574,
+      "loss": 0.0939,
+      "step": 8424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001027956256774124,
+      "loss": 0.0832,
+      "step": 8425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010277729843616448,
+      "loss": 0.2217,
+      "step": 8426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010275897110155797,
+      "loss": 0.0786,
+      "step": 8427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010274064367420897,
+      "loss": 0.0809,
+      "step": 8428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010272231615473363,
+      "loss": 0.2643,
+      "step": 8429
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010270398854374791,
+      "loss": 0.1419,
+      "step": 8430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010268566084186799,
+      "loss": 0.0986,
+      "step": 8431
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010266733304970988,
+      "loss": 0.0977,
+      "step": 8432
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010264900516788973,
+      "loss": 0.1479,
+      "step": 8433
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010263067719702357,
+      "loss": 0.1862,
+      "step": 8434
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010261234913772753,
+      "loss": 0.1407,
+      "step": 8435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001025940209906177,
+      "loss": 0.1732,
+      "step": 8436
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010257569275631016,
+      "loss": 0.0898,
+      "step": 8437
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010255736443542104,
+      "loss": 0.0709,
+      "step": 8438
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010253903602856643,
+      "loss": 0.1089,
+      "step": 8439
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010252070753636245,
+      "loss": 0.0647,
+      "step": 8440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010250237895942514,
+      "loss": 0.0825,
+      "step": 8441
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001024840502983707,
+      "loss": 0.0657,
+      "step": 8442
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010246572155381522,
+      "loss": 0.0704,
+      "step": 8443
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010244739272637481,
+      "loss": 0.1131,
+      "step": 8444
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001024290638166656,
+      "loss": 0.1418,
+      "step": 8445
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010241073482530372,
+      "loss": 0.0738,
+      "step": 8446
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010239240575290528,
+      "loss": 0.0607,
+      "step": 8447
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010237407660008637,
+      "loss": 0.0737,
+      "step": 8448
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010235574736746321,
+      "loss": 0.1545,
+      "step": 8449
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010233741805565188,
+      "loss": 0.2665,
+      "step": 8450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010231908866526851,
+      "loss": 0.136,
+      "step": 8451
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010230075919692931,
+      "loss": 0.099,
+      "step": 8452
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010228242965125035,
+      "loss": 0.0842,
+      "step": 8453
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001022641000288478,
+      "loss": 0.0908,
+      "step": 8454
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001022457703303378,
+      "loss": 0.0851,
+      "step": 8455
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010222744055633648,
+      "loss": 0.0812,
+      "step": 8456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010220911070746007,
+      "loss": 0.1344,
+      "step": 8457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010219078078432463,
+      "loss": 0.0857,
+      "step": 8458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010217245078754641,
+      "loss": 0.1005,
+      "step": 8459
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010215412071774149,
+      "loss": 0.1141,
+      "step": 8460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010213579057552609,
+      "loss": 0.1091,
+      "step": 8461
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010211746036151632,
+      "loss": 0.1236,
+      "step": 8462
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010209913007632838,
+      "loss": 0.0812,
+      "step": 8463
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010208079972057847,
+      "loss": 0.0608,
+      "step": 8464
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010206246929488273,
+      "loss": 0.1866,
+      "step": 8465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010204413879985735,
+      "loss": 0.2804,
+      "step": 8466
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010202580823611845,
+      "loss": 0.1128,
+      "step": 8467
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010200747760428228,
+      "loss": 0.077,
+      "step": 8468
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010198914690496499,
+      "loss": 0.0658,
+      "step": 8469
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010197081613878276,
+      "loss": 0.1654,
+      "step": 8470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010195248530635184,
+      "loss": 0.2563,
+      "step": 8471
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010193415440828832,
+      "loss": 0.0866,
+      "step": 8472
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010191582344520847,
+      "loss": 0.1892,
+      "step": 8473
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010189749241772844,
+      "loss": 0.1356,
+      "step": 8474
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010187916132646443,
+      "loss": 0.1131,
+      "step": 8475
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010186083017203263,
+      "loss": 0.0694,
+      "step": 8476
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010184249895504925,
+      "loss": 0.1133,
+      "step": 8477
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010182416767613053,
+      "loss": 0.111,
+      "step": 8478
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010180583633589261,
+      "loss": 0.1556,
+      "step": 8479
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010178750493495173,
+      "loss": 0.0925,
+      "step": 8480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010176917347392408,
+      "loss": 0.113,
+      "step": 8481
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001017508419534259,
+      "loss": 0.0999,
+      "step": 8482
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010173251037407336,
+      "loss": 0.133,
+      "step": 8483
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010171417873648272,
+      "loss": 0.1445,
+      "step": 8484
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010169584704127018,
+      "loss": 0.1222,
+      "step": 8485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010167751528905192,
+      "loss": 0.0616,
+      "step": 8486
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010165918348044421,
+      "loss": 0.1045,
+      "step": 8487
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010164085161606325,
+      "loss": 0.0682,
+      "step": 8488
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010162251969652523,
+      "loss": 0.0842,
+      "step": 8489
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010160418772244646,
+      "loss": 0.2319,
+      "step": 8490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001015858556944431,
+      "loss": 0.0805,
+      "step": 8491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001015675236131314,
+      "loss": 0.0865,
+      "step": 8492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010154919147912758,
+      "loss": 0.2074,
+      "step": 8493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010153085929304789,
+      "loss": 0.2117,
+      "step": 8494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010151252705550853,
+      "loss": 0.0849,
+      "step": 8495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010149419476712575,
+      "loss": 0.0629,
+      "step": 8496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010147586242851585,
+      "loss": 0.0936,
+      "step": 8497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010145753004029497,
+      "loss": 0.0871,
+      "step": 8498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010143919760307942,
+      "loss": 0.1399,
+      "step": 8499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010142086511748541,
+      "loss": 0.2272,
+      "step": 8500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001014025325841292,
+      "loss": 0.1694,
+      "step": 8501
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000101384200003627,
+      "loss": 0.1809,
+      "step": 8502
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001013658673765951,
+      "loss": 0.1525,
+      "step": 8503
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010134753470364977,
+      "loss": 0.0904,
+      "step": 8504
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001013292019854072,
+      "loss": 0.0925,
+      "step": 8505
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001013108692224837,
+      "loss": 0.1112,
+      "step": 8506
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010129253641549543,
+      "loss": 0.1432,
+      "step": 8507
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010127420356505876,
+      "loss": 0.0809,
+      "step": 8508
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010125587067178985,
+      "loss": 0.125,
+      "step": 8509
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010123753773630505,
+      "loss": 0.1092,
+      "step": 8510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010121920475922056,
+      "loss": 0.1381,
+      "step": 8511
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010120087174115262,
+      "loss": 0.1025,
+      "step": 8512
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010118253868271757,
+      "loss": 0.0828,
+      "step": 8513
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010116420558453159,
+      "loss": 0.0564,
+      "step": 8514
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010114587244721099,
+      "loss": 0.0989,
+      "step": 8515
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010112753927137202,
+      "loss": 0.0624,
+      "step": 8516
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010110920605763098,
+      "loss": 0.3422,
+      "step": 8517
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010109087280660411,
+      "loss": 0.2403,
+      "step": 8518
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010107253951890767,
+      "loss": 0.105,
+      "step": 8519
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010105420619515798,
+      "loss": 0.0729,
+      "step": 8520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010103587283597122,
+      "loss": 0.084,
+      "step": 8521
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010101753944196377,
+      "loss": 0.071,
+      "step": 8522
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010099920601375186,
+      "loss": 0.0762,
+      "step": 8523
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010098087255195174,
+      "loss": 0.0989,
+      "step": 8524
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010096253905717971,
+      "loss": 0.059,
+      "step": 8525
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010094420553005204,
+      "loss": 0.0726,
+      "step": 8526
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010092587197118504,
+      "loss": 0.0812,
+      "step": 8527
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010090753838119496,
+      "loss": 0.1448,
+      "step": 8528
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010088920476069809,
+      "loss": 0.1123,
+      "step": 8529
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010087087111031073,
+      "loss": 0.2157,
+      "step": 8530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010085253743064911,
+      "loss": 0.0664,
+      "step": 8531
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010083420372232956,
+      "loss": 0.0975,
+      "step": 8532
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010081586998596835,
+      "loss": 0.0912,
+      "step": 8533
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010079753622218177,
+      "loss": 0.1776,
+      "step": 8534
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010077920243158612,
+      "loss": 0.1371,
+      "step": 8535
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010076086861479767,
+      "loss": 0.2278,
+      "step": 8536
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010074253477243274,
+      "loss": 0.0698,
+      "step": 8537
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010072420090510755,
+      "loss": 0.1111,
+      "step": 8538
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010070586701343847,
+      "loss": 0.0662,
+      "step": 8539
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010068753309804174,
+      "loss": 0.1309,
+      "step": 8540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010066919915953368,
+      "loss": 0.0655,
+      "step": 8541
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010065086519853056,
+      "loss": 0.0698,
+      "step": 8542
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010063253121564868,
+      "loss": 0.0986,
+      "step": 8543
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010061419721150435,
+      "loss": 0.0613,
+      "step": 8544
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010059586318671386,
+      "loss": 0.0905,
+      "step": 8545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001005775291418935,
+      "loss": 0.0663,
+      "step": 8546
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010055919507765959,
+      "loss": 0.0791,
+      "step": 8547
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010054086099462839,
+      "loss": 0.1672,
+      "step": 8548
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010052252689341621,
+      "loss": 0.1003,
+      "step": 8549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010050419277463934,
+      "loss": 0.1632,
+      "step": 8550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001004858586389141,
+      "loss": 0.0698,
+      "step": 8551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010046752448685679,
+      "loss": 0.1622,
+      "step": 8552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010044919031908371,
+      "loss": 0.0796,
+      "step": 8553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010043085613621113,
+      "loss": 0.0778,
+      "step": 8554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001004125219388554,
+      "loss": 0.0764,
+      "step": 8555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010039418772763278,
+      "loss": 0.0947,
+      "step": 8556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010037585350315959,
+      "loss": 0.098,
+      "step": 8557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010035751926605214,
+      "loss": 0.09,
+      "step": 8558
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001003391850169267,
+      "loss": 0.2137,
+      "step": 8559
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010032085075639963,
+      "loss": 0.1723,
+      "step": 8560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010030251648508718,
+      "loss": 0.1201,
+      "step": 8561
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010028418220360567,
+      "loss": 0.0729,
+      "step": 8562
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010026584791257143,
+      "loss": 0.1373,
+      "step": 8563
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010024751361260072,
+      "loss": 0.1222,
+      "step": 8564
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010022917930430989,
+      "loss": 0.1028,
+      "step": 8565
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010021084498831522,
+      "loss": 0.1718,
+      "step": 8566
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010019251066523303,
+      "loss": 0.2165,
+      "step": 8567
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001001741763356796,
+      "loss": 0.1845,
+      "step": 8568
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010015584200027126,
+      "loss": 0.104,
+      "step": 8569
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010013750765962433,
+      "loss": 0.0838,
+      "step": 8570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010011917331435508,
+      "loss": 0.0965,
+      "step": 8571
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010010083896507983,
+      "loss": 0.1031,
+      "step": 8572
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010008250461241489,
+      "loss": 0.0867,
+      "step": 8573
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010006417025697656,
+      "loss": 0.1056,
+      "step": 8574
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010004583589938119,
+      "loss": 0.1058,
+      "step": 8575
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010002750154024503,
+      "loss": 0.1571,
+      "step": 8576
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010000916718018441,
+      "loss": 0.0715,
+      "step": 8577
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.999083281981562e-05,
+      "loss": 0.1081,
+      "step": 8578
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.997249845975501e-05,
+      "loss": 0.0933,
+      "step": 8579
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.995416410061883e-05,
+      "loss": 0.1787,
+      "step": 8580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.993582974302345e-05,
+      "loss": 0.1161,
+      "step": 8581
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.991749538758512e-05,
+      "loss": 0.0801,
+      "step": 8582
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.98991610349202e-05,
+      "loss": 0.1001,
+      "step": 8583
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.988082668564497e-05,
+      "loss": 0.2611,
+      "step": 8584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.986249234037572e-05,
+      "loss": 0.126,
+      "step": 8585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.984415799972877e-05,
+      "loss": 0.14,
+      "step": 8586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.982582366432042e-05,
+      "loss": 0.0961,
+      "step": 8587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.9807489334767e-05,
+      "loss": 0.0901,
+      "step": 8588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.97891550116848e-05,
+      "loss": 0.1496,
+      "step": 8589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.977082069569012e-05,
+      "loss": 0.0605,
+      "step": 8590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.97524863873993e-05,
+      "loss": 0.0726,
+      "step": 8591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.973415208742859e-05,
+      "loss": 0.1706,
+      "step": 8592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.971581779639435e-05,
+      "loss": 0.1517,
+      "step": 8593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.969748351491283e-05,
+      "loss": 0.145,
+      "step": 8594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.96791492436004e-05,
+      "loss": 0.1176,
+      "step": 8595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.96608149830733e-05,
+      "loss": 0.0906,
+      "step": 8596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.964248073394791e-05,
+      "loss": 0.1443,
+      "step": 8597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.962414649684045e-05,
+      "loss": 0.1,
+      "step": 8598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.960581227236725e-05,
+      "loss": 0.105,
+      "step": 8599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.958747806114462e-05,
+      "loss": 0.1061,
+      "step": 8600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.956914386378888e-05,
+      "loss": 0.1391,
+      "step": 8601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.955080968091632e-05,
+      "loss": 0.1204,
+      "step": 8602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.953247551314323e-05,
+      "loss": 0.1045,
+      "step": 8603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.95141413610859e-05,
+      "loss": 0.1483,
+      "step": 8604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.949580722536069e-05,
+      "loss": 0.0891,
+      "step": 8605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.947747310658381e-05,
+      "loss": 0.1005,
+      "step": 8606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.945913900537164e-05,
+      "loss": 0.0812,
+      "step": 8607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.944080492234042e-05,
+      "loss": 0.0475,
+      "step": 8608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.94224708581065e-05,
+      "loss": 0.0933,
+      "step": 8609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.940413681328618e-05,
+      "loss": 0.1236,
+      "step": 8610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.938580278849567e-05,
+      "loss": 0.0642,
+      "step": 8611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.936746878435136e-05,
+      "loss": 0.0681,
+      "step": 8612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.934913480146948e-05,
+      "loss": 0.1148,
+      "step": 8613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.933080084046636e-05,
+      "loss": 0.0926,
+      "step": 8614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.93124669019583e-05,
+      "loss": 0.178,
+      "step": 8615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.929413298656156e-05,
+      "loss": 0.1091,
+      "step": 8616
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.927579909489247e-05,
+      "loss": 0.0713,
+      "step": 8617
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.925746522756729e-05,
+      "loss": 0.0868,
+      "step": 8618
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.923913138520235e-05,
+      "loss": 0.218,
+      "step": 8619
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.922079756841389e-05,
+      "loss": 0.1016,
+      "step": 8620
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.920246377781825e-05,
+      "loss": 0.0706,
+      "step": 8621
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.918413001403165e-05,
+      "loss": 0.1783,
+      "step": 8622
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.916579627767049e-05,
+      "loss": 0.0963,
+      "step": 8623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.914746256935095e-05,
+      "loss": 0.0842,
+      "step": 8624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.912912888968933e-05,
+      "loss": 0.0774,
+      "step": 8625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.911079523930193e-05,
+      "loss": 0.2544,
+      "step": 8626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.909246161880506e-05,
+      "loss": 0.1056,
+      "step": 8627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.907412802881497e-05,
+      "loss": 0.0896,
+      "step": 8628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.905579446994797e-05,
+      "loss": 0.0978,
+      "step": 8629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.90374609428203e-05,
+      "loss": 0.1203,
+      "step": 8630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.901912744804829e-05,
+      "loss": 0.0524,
+      "step": 8631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.900079398624816e-05,
+      "loss": 0.1028,
+      "step": 8632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.898246055803624e-05,
+      "loss": 0.1167,
+      "step": 8633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.896412716402876e-05,
+      "loss": 0.1091,
+      "step": 8634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.894579380484204e-05,
+      "loss": 0.2553,
+      "step": 8635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.892746048109237e-05,
+      "loss": 0.1197,
+      "step": 8636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.890912719339592e-05,
+      "loss": 0.0661,
+      "step": 8637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.889079394236905e-05,
+      "loss": 0.0945,
+      "step": 8638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.887246072862799e-05,
+      "loss": 0.2347,
+      "step": 8639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.885412755278902e-05,
+      "loss": 0.141,
+      "step": 8640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.883579441546844e-05,
+      "loss": 0.0716,
+      "step": 8641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.881746131728246e-05,
+      "loss": 0.1173,
+      "step": 8642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.879912825884739e-05,
+      "loss": 0.0826,
+      "step": 8643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.878079524077946e-05,
+      "loss": 0.1091,
+      "step": 8644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.876246226369497e-05,
+      "loss": 0.0767,
+      "step": 8645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.874412932821013e-05,
+      "loss": 0.0732,
+      "step": 8646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.872579643494126e-05,
+      "loss": 0.0743,
+      "step": 8647
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.870746358450457e-05,
+      "loss": 0.1303,
+      "step": 8648
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.868913077751636e-05,
+      "loss": 0.0687,
+      "step": 8649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.867079801459284e-05,
+      "loss": 0.0695,
+      "step": 8650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.865246529635025e-05,
+      "loss": 0.1452,
+      "step": 8651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.863413262340491e-05,
+      "loss": 0.109,
+      "step": 8652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.861579999637301e-05,
+      "loss": 0.1069,
+      "step": 8653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.859746741587082e-05,
+      "loss": 0.1184,
+      "step": 8654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.857913488251463e-05,
+      "loss": 0.0779,
+      "step": 8655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.85608023969206e-05,
+      "loss": 0.1566,
+      "step": 8656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.854246995970506e-05,
+      "loss": 0.1488,
+      "step": 8657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.852413757148417e-05,
+      "loss": 0.1264,
+      "step": 8658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.850580523287425e-05,
+      "loss": 0.0909,
+      "step": 8659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.848747294449148e-05,
+      "loss": 0.1045,
+      "step": 8660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.846914070695213e-05,
+      "loss": 0.1788,
+      "step": 8661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.845080852087247e-05,
+      "loss": 0.1177,
+      "step": 8662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.843247638686863e-05,
+      "loss": 0.0612,
+      "step": 8663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.841414430555695e-05,
+      "loss": 0.17,
+      "step": 8664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.839581227755356e-05,
+      "loss": 0.0514,
+      "step": 8665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.837748030347479e-05,
+      "loss": 0.1495,
+      "step": 8666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.835914838393678e-05,
+      "loss": 0.1322,
+      "step": 8667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.83408165195558e-05,
+      "loss": 0.1678,
+      "step": 8668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.832248471094811e-05,
+      "loss": 0.0848,
+      "step": 8669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.830415295872984e-05,
+      "loss": 0.0872,
+      "step": 8670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.82858212635173e-05,
+      "loss": 0.1147,
+      "step": 8671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.826748962592664e-05,
+      "loss": 0.0847,
+      "step": 8672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.824915804657412e-05,
+      "loss": 0.0741,
+      "step": 8673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.823082652607592e-05,
+      "loss": 0.1167,
+      "step": 8674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.821249506504831e-05,
+      "loss": 0.0971,
+      "step": 8675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.819416366410744e-05,
+      "loss": 0.1008,
+      "step": 8676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.817583232386951e-05,
+      "loss": 0.085,
+      "step": 8677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.815750104495078e-05,
+      "loss": 0.0696,
+      "step": 8678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.813916982796741e-05,
+      "loss": 0.0632,
+      "step": 8679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.81208386735356e-05,
+      "loss": 0.0925,
+      "step": 8680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.81025075822716e-05,
+      "loss": 0.1219,
+      "step": 8681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.808417655479154e-05,
+      "loss": 0.0762,
+      "step": 8682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.806584559171169e-05,
+      "loss": 0.0752,
+      "step": 8683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.804751469364817e-05,
+      "loss": 0.0942,
+      "step": 8684
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.802918386121722e-05,
+      "loss": 0.1111,
+      "step": 8685
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.801085309503501e-05,
+      "loss": 0.1156,
+      "step": 8686
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.799252239571773e-05,
+      "loss": 0.1629,
+      "step": 8687
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.79741917638816e-05,
+      "loss": 0.0738,
+      "step": 8688
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.795586120014271e-05,
+      "loss": 0.0822,
+      "step": 8689
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.793753070511731e-05,
+      "loss": 0.1339,
+      "step": 8690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.791920027942156e-05,
+      "loss": 0.0842,
+      "step": 8691
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.790086992367164e-05,
+      "loss": 0.1058,
+      "step": 8692
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.78825396384837e-05,
+      "loss": 0.0839,
+      "step": 8693
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.786420942447393e-05,
+      "loss": 0.1223,
+      "step": 8694
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.784587928225853e-05,
+      "loss": 0.104,
+      "step": 8695
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.78275492124536e-05,
+      "loss": 0.1306,
+      "step": 8696
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.780921921567537e-05,
+      "loss": 0.1432,
+      "step": 8697
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.779088929253994e-05,
+      "loss": 0.0776,
+      "step": 8698
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.777255944366351e-05,
+      "loss": 0.1502,
+      "step": 8699
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.775422966966221e-05,
+      "loss": 0.0571,
+      "step": 8700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.773589997115225e-05,
+      "loss": 0.266,
+      "step": 8701
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.771757034874969e-05,
+      "loss": 0.1498,
+      "step": 8702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.769924080307071e-05,
+      "loss": 0.19,
+      "step": 8703
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.76809113347315e-05,
+      "loss": 0.0913,
+      "step": 8704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.766258194434814e-05,
+      "loss": 0.0733,
+      "step": 8705
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.764425263253681e-05,
+      "loss": 0.126,
+      "step": 8706
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.762592339991365e-05,
+      "loss": 0.0652,
+      "step": 8707
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.760759424709474e-05,
+      "loss": 0.1008,
+      "step": 8708
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.75892651746963e-05,
+      "loss": 0.1839,
+      "step": 8709
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.75709361833344e-05,
+      "loss": 0.0966,
+      "step": 8710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.75526072736252e-05,
+      "loss": 0.2475,
+      "step": 8711
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.753427844618476e-05,
+      "loss": 0.1846,
+      "step": 8712
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.751594970162933e-05,
+      "loss": 0.0843,
+      "step": 8713
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.749762104057489e-05,
+      "loss": 0.1402,
+      "step": 8714
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.747929246363761e-05,
+      "loss": 0.1776,
+      "step": 8715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.746096397143362e-05,
+      "loss": 0.1088,
+      "step": 8716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.744263556457899e-05,
+      "loss": 0.1605,
+      "step": 8717
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.742430724368986e-05,
+      "loss": 0.1164,
+      "step": 8718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.740597900938232e-05,
+      "loss": 0.1189,
+      "step": 8719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.738765086227249e-05,
+      "loss": 0.1248,
+      "step": 8720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.736932280297644e-05,
+      "loss": 0.062,
+      "step": 8721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.735099483211028e-05,
+      "loss": 0.0894,
+      "step": 8722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.733266695029012e-05,
+      "loss": 0.0946,
+      "step": 8723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.731433915813202e-05,
+      "loss": 0.1386,
+      "step": 8724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.72960114562521e-05,
+      "loss": 0.1099,
+      "step": 8725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.727768384526642e-05,
+      "loss": 0.1106,
+      "step": 8726
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.725935632579104e-05,
+      "loss": 0.0818,
+      "step": 8727
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.724102889844208e-05,
+      "loss": 0.0943,
+      "step": 8728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.722270156383558e-05,
+      "loss": 0.1225,
+      "step": 8729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.720437432258763e-05,
+      "loss": 0.0904,
+      "step": 8730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.718604717531427e-05,
+      "loss": 0.1443,
+      "step": 8731
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.716772012263164e-05,
+      "loss": 0.2754,
+      "step": 8732
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.71493931651557e-05,
+      "loss": 0.0822,
+      "step": 8733
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.713106630350257e-05,
+      "loss": 0.0823,
+      "step": 8734
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.71127395382883e-05,
+      "loss": 0.1907,
+      "step": 8735
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.709441287012893e-05,
+      "loss": 0.3021,
+      "step": 8736
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.707608629964052e-05,
+      "loss": 0.1204,
+      "step": 8737
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.705775982743912e-05,
+      "loss": 0.0734,
+      "step": 8738
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.703943345414078e-05,
+      "loss": 0.1415,
+      "step": 8739
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.702110718036149e-05,
+      "loss": 0.0784,
+      "step": 8740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.70027810067173e-05,
+      "loss": 0.2107,
+      "step": 8741
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698445493382429e-05,
+      "loss": 0.1478,
+      "step": 8742
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.696612896229843e-05,
+      "loss": 0.1723,
+      "step": 8743
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.69478030927558e-05,
+      "loss": 0.1266,
+      "step": 8744
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.692947732581236e-05,
+      "loss": 0.1265,
+      "step": 8745
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.691115166208418e-05,
+      "loss": 0.1823,
+      "step": 8746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.689282610218725e-05,
+      "loss": 0.1134,
+      "step": 8747
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.687450064673759e-05,
+      "loss": 0.252,
+      "step": 8748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.685617529635122e-05,
+      "loss": 0.0772,
+      "step": 8749
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.683785005164411e-05,
+      "loss": 0.1162,
+      "step": 8750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.68195249132323e-05,
+      "loss": 0.09,
+      "step": 8751
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.68011998817318e-05,
+      "loss": 0.1614,
+      "step": 8752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.678287495775852e-05,
+      "loss": 0.1318,
+      "step": 8753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.676455014192852e-05,
+      "loss": 0.1491,
+      "step": 8754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.674622543485776e-05,
+      "loss": 0.0749,
+      "step": 8755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.672790083716224e-05,
+      "loss": 0.215,
+      "step": 8756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.670957634945792e-05,
+      "loss": 0.0908,
+      "step": 8757
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.669125197236081e-05,
+      "loss": 0.1743,
+      "step": 8758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.667292770648684e-05,
+      "loss": 0.1052,
+      "step": 8759
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.665460355245201e-05,
+      "loss": 0.1232,
+      "step": 8760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.663627951087224e-05,
+      "loss": 0.0643,
+      "step": 8761
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.661795558236352e-05,
+      "loss": 0.1288,
+      "step": 8762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.659963176754185e-05,
+      "loss": 0.0928,
+      "step": 8763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.65813080670231e-05,
+      "loss": 0.1286,
+      "step": 8764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.656298448142332e-05,
+      "loss": 0.0671,
+      "step": 8765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.654466101135835e-05,
+      "loss": 0.1177,
+      "step": 8766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.652633765744416e-05,
+      "loss": 0.104,
+      "step": 8767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.650801442029674e-05,
+      "loss": 0.1696,
+      "step": 8768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.648969130053195e-05,
+      "loss": 0.1274,
+      "step": 8769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.647136829876577e-05,
+      "loss": 0.0716,
+      "step": 8770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.645304541561409e-05,
+      "loss": 0.0813,
+      "step": 8771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.643472265169289e-05,
+      "loss": 0.106,
+      "step": 8772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.641640000761802e-05,
+      "loss": 0.1384,
+      "step": 8773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.639807748400544e-05,
+      "loss": 0.0949,
+      "step": 8774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.637975508147103e-05,
+      "loss": 0.0715,
+      "step": 8775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.63614328006307e-05,
+      "loss": 0.0667,
+      "step": 8776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.634311064210039e-05,
+      "loss": 0.1452,
+      "step": 8777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.632478860649598e-05,
+      "loss": 0.135,
+      "step": 8778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.630646669443331e-05,
+      "loss": 0.0555,
+      "step": 8779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.628814490652832e-05,
+      "loss": 0.3401,
+      "step": 8780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.626982324339686e-05,
+      "loss": 0.2179,
+      "step": 8781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.625150170565487e-05,
+      "loss": 0.0998,
+      "step": 8782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.623318029391815e-05,
+      "loss": 0.0679,
+      "step": 8783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.621485900880264e-05,
+      "loss": 0.0838,
+      "step": 8784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.619653785092416e-05,
+      "loss": 0.1033,
+      "step": 8785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.617821682089861e-05,
+      "loss": 0.104,
+      "step": 8786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.615989591934182e-05,
+      "loss": 0.1343,
+      "step": 8787
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.614157514686965e-05,
+      "loss": 0.0658,
+      "step": 8788
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.612325450409799e-05,
+      "loss": 0.1114,
+      "step": 8789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.610493399164261e-05,
+      "loss": 0.1424,
+      "step": 8790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.608661361011945e-05,
+      "loss": 0.101,
+      "step": 8791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.606829336014427e-05,
+      "loss": 0.194,
+      "step": 8792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.604997324233289e-05,
+      "loss": 0.0953,
+      "step": 8793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.603165325730122e-05,
+      "loss": 0.0896,
+      "step": 8794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.601333340566499e-05,
+      "loss": 0.0846,
+      "step": 8795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.59950136880401e-05,
+      "loss": 0.1711,
+      "step": 8796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.597669410504231e-05,
+      "loss": 0.0571,
+      "step": 8797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.595837465728749e-05,
+      "loss": 0.0766,
+      "step": 8798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.594005534539136e-05,
+      "loss": 0.1239,
+      "step": 8799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.592173616996981e-05,
+      "loss": 0.0599,
+      "step": 8800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.590341713163858e-05,
+      "loss": 0.1223,
+      "step": 8801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.588509823101346e-05,
+      "loss": 0.1441,
+      "step": 8802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.586677946871029e-05,
+      "loss": 0.1516,
+      "step": 8803
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.584846084534484e-05,
+      "loss": 0.1868,
+      "step": 8804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.583014236153282e-05,
+      "loss": 0.0702,
+      "step": 8805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.581182401789009e-05,
+      "loss": 0.0868,
+      "step": 8806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.579350581503235e-05,
+      "loss": 0.183,
+      "step": 8807
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.57751877535754e-05,
+      "loss": 0.0697,
+      "step": 8808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.5756869834135e-05,
+      "loss": 0.1052,
+      "step": 8809
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.57385520573269e-05,
+      "loss": 0.2479,
+      "step": 8810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.572023442376683e-05,
+      "loss": 0.1241,
+      "step": 8811
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.570191693407058e-05,
+      "loss": 0.1334,
+      "step": 8812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.568359958885384e-05,
+      "loss": 0.121,
+      "step": 8813
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.566528238873239e-05,
+      "loss": 0.109,
+      "step": 8814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.564696533432193e-05,
+      "loss": 0.125,
+      "step": 8815
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.562864842623817e-05,
+      "loss": 0.17,
+      "step": 8816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.561033166509692e-05,
+      "loss": 0.0613,
+      "step": 8817
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.55920150515138e-05,
+      "loss": 0.1035,
+      "step": 8818
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.557369858610453e-05,
+      "loss": 0.0889,
+      "step": 8819
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.555538226948487e-05,
+      "loss": 0.0906,
+      "step": 8820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.553706610227046e-05,
+      "loss": 0.0724,
+      "step": 8821
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.551875008507705e-05,
+      "loss": 0.127,
+      "step": 8822
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.550043421852026e-05,
+      "loss": 0.0728,
+      "step": 8823
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.548211850321587e-05,
+      "loss": 0.0868,
+      "step": 8824
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.546380293977947e-05,
+      "loss": 0.0659,
+      "step": 8825
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.544548752882682e-05,
+      "loss": 0.0877,
+      "step": 8826
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.542717227097351e-05,
+      "loss": 0.0912,
+      "step": 8827
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.540885716683526e-05,
+      "loss": 0.1194,
+      "step": 8828
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.539054221702769e-05,
+      "loss": 0.0872,
+      "step": 8829
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.537222742216653e-05,
+      "loss": 0.3531,
+      "step": 8830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.535391278286732e-05,
+      "loss": 0.1684,
+      "step": 8831
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.53355982997458e-05,
+      "loss": 0.0736,
+      "step": 8832
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.531728397341754e-05,
+      "loss": 0.0754,
+      "step": 8833
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.529896980449822e-05,
+      "loss": 0.145,
+      "step": 8834
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.528065579360342e-05,
+      "loss": 0.1119,
+      "step": 8835
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.526234194134885e-05,
+      "loss": 0.0687,
+      "step": 8836
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.524402824835001e-05,
+      "loss": 0.1429,
+      "step": 8837
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.522571471522263e-05,
+      "loss": 0.1305,
+      "step": 8838
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.520740134258223e-05,
+      "loss": 0.093,
+      "step": 8839
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.518908813104445e-05,
+      "loss": 0.1664,
+      "step": 8840
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.517077508122488e-05,
+      "loss": 0.0962,
+      "step": 8841
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.51524621937391e-05,
+      "loss": 0.2272,
+      "step": 8842
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.513414946920278e-05,
+      "loss": 0.1224,
+      "step": 8843
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.511583690823137e-05,
+      "loss": 0.0694,
+      "step": 8844
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.509752451144048e-05,
+      "loss": 0.1152,
+      "step": 8845
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.507921227944575e-05,
+      "loss": 0.0576,
+      "step": 8846
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.506090021286264e-05,
+      "loss": 0.0832,
+      "step": 8847
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.504258831230682e-05,
+      "loss": 0.1034,
+      "step": 8848
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.502427657839374e-05,
+      "loss": 0.0708,
+      "step": 8849
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.5005965011739e-05,
+      "loss": 0.1574,
+      "step": 8850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.498765361295813e-05,
+      "loss": 0.2193,
+      "step": 8851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.496934238266668e-05,
+      "loss": 0.0865,
+      "step": 8852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.495103132148015e-05,
+      "loss": 0.0596,
+      "step": 8853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.493272043001411e-05,
+      "loss": 0.0907,
+      "step": 8854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.491440970888405e-05,
+      "loss": 0.2374,
+      "step": 8855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.489609915870546e-05,
+      "loss": 0.0842,
+      "step": 8856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.487778878009389e-05,
+      "loss": 0.1122,
+      "step": 8857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.485947857366478e-05,
+      "loss": 0.1062,
+      "step": 8858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.484116854003369e-05,
+      "loss": 0.0912,
+      "step": 8859
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.482285867981609e-05,
+      "loss": 0.0927,
+      "step": 8860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.480454899362744e-05,
+      "loss": 0.2007,
+      "step": 8861
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.478623948208326e-05,
+      "loss": 0.1233,
+      "step": 8862
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.476793014579897e-05,
+      "loss": 0.0715,
+      "step": 8863
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.47496209853901e-05,
+      "loss": 0.1243,
+      "step": 8864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.473131200147205e-05,
+      "loss": 0.0827,
+      "step": 8865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.471300319466032e-05,
+      "loss": 0.0698,
+      "step": 8866
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.469469456557031e-05,
+      "loss": 0.1112,
+      "step": 8867
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.467638611481753e-05,
+      "loss": 0.1093,
+      "step": 8868
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.465807784301735e-05,
+      "loss": 0.2154,
+      "step": 8869
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.463976975078522e-05,
+      "loss": 0.0615,
+      "step": 8870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.462146183873657e-05,
+      "loss": 0.0784,
+      "step": 8871
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.460315410748684e-05,
+      "loss": 0.1813,
+      "step": 8872
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.458484655765139e-05,
+      "loss": 0.0745,
+      "step": 8873
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.456653918984569e-05,
+      "loss": 0.1084,
+      "step": 8874
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.454823200468507e-05,
+      "loss": 0.1126,
+      "step": 8875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.4529925002785e-05,
+      "loss": 0.0944,
+      "step": 8876
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.451161818476079e-05,
+      "loss": 0.0791,
+      "step": 8877
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.44933115512279e-05,
+      "loss": 0.0835,
+      "step": 8878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.447500510280162e-05,
+      "loss": 0.1001,
+      "step": 8879
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.44566988400974e-05,
+      "loss": 0.1654,
+      "step": 8880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.443839276373059e-05,
+      "loss": 0.0854,
+      "step": 8881
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.442008687431648e-05,
+      "loss": 0.1431,
+      "step": 8882
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.440178117247047e-05,
+      "loss": 0.1344,
+      "step": 8883
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.438347565880788e-05,
+      "loss": 0.1097,
+      "step": 8884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.436517033394407e-05,
+      "loss": 0.1363,
+      "step": 8885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.43468651984944e-05,
+      "loss": 0.1045,
+      "step": 8886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.432856025307413e-05,
+      "loss": 0.0889,
+      "step": 8887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.431025549829862e-05,
+      "loss": 0.156,
+      "step": 8888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.429195093478317e-05,
+      "loss": 0.1075,
+      "step": 8889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.427364656314311e-05,
+      "loss": 0.0714,
+      "step": 8890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.425534238399367e-05,
+      "loss": 0.1274,
+      "step": 8891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.423703839795022e-05,
+      "loss": 0.0596,
+      "step": 8892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.421873460562801e-05,
+      "loss": 0.0869,
+      "step": 8893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.420043100764235e-05,
+      "loss": 0.0863,
+      "step": 8894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.418212760460848e-05,
+      "loss": 0.0991,
+      "step": 8895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.416382439714164e-05,
+      "loss": 0.0856,
+      "step": 8896
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.414552138585716e-05,
+      "loss": 0.1518,
+      "step": 8897
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.412721857137023e-05,
+      "loss": 0.1325,
+      "step": 8898
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.410891595429613e-05,
+      "loss": 0.0606,
+      "step": 8899
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.409061353525011e-05,
+      "loss": 0.0856,
+      "step": 8900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.407231131484737e-05,
+      "loss": 0.0719,
+      "step": 8901
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.405400929370318e-05,
+      "loss": 0.1077,
+      "step": 8902
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.403570747243272e-05,
+      "loss": 0.1001,
+      "step": 8903
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.401740585165123e-05,
+      "loss": 0.0854,
+      "step": 8904
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.39991044319739e-05,
+      "loss": 0.0708,
+      "step": 8905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.398080321401595e-05,
+      "loss": 0.056,
+      "step": 8906
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.396250219839258e-05,
+      "loss": 0.1151,
+      "step": 8907
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.394420138571891e-05,
+      "loss": 0.1027,
+      "step": 8908
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.392590077661018e-05,
+      "loss": 0.2169,
+      "step": 8909
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.390760037168155e-05,
+      "loss": 0.0871,
+      "step": 8910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.388930017154819e-05,
+      "loss": 0.1046,
+      "step": 8911
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.387100017682523e-05,
+      "loss": 0.1108,
+      "step": 8912
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.385270038812785e-05,
+      "loss": 0.1095,
+      "step": 8913
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.38344008060712e-05,
+      "loss": 0.1329,
+      "step": 8914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.38161014312704e-05,
+      "loss": 0.1566,
+      "step": 8915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.379780226434062e-05,
+      "loss": 0.1642,
+      "step": 8916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.377950330589692e-05,
+      "loss": 0.0754,
+      "step": 8917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.376120455655449e-05,
+      "loss": 0.0792,
+      "step": 8918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.374290601692835e-05,
+      "loss": 0.1046,
+      "step": 8919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.372460768763372e-05,
+      "loss": 0.0853,
+      "step": 8920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.37063095692856e-05,
+      "loss": 0.1719,
+      "step": 8921
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.368801166249909e-05,
+      "loss": 0.1467,
+      "step": 8922
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.366971396788931e-05,
+      "loss": 0.0746,
+      "step": 8923
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.36514164860713e-05,
+      "loss": 0.0711,
+      "step": 8924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.363311921766015e-05,
+      "loss": 0.0792,
+      "step": 8925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.361482216327092e-05,
+      "loss": 0.1958,
+      "step": 8926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.359652532351864e-05,
+      "loss": 0.0881,
+      "step": 8927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.357822869901841e-05,
+      "loss": 0.1188,
+      "step": 8928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.35599322903852e-05,
+      "loss": 0.117,
+      "step": 8929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.35416360982341e-05,
+      "loss": 0.07,
+      "step": 8930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.352334012318009e-05,
+      "loss": 0.0761,
+      "step": 8931
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.350504436583821e-05,
+      "loss": 0.1654,
+      "step": 8932
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.34867488268235e-05,
+      "loss": 0.0551,
+      "step": 8933
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.346845350675088e-05,
+      "loss": 0.092,
+      "step": 8934
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.345015840623542e-05,
+      "loss": 0.0925,
+      "step": 8935
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.343186352589203e-05,
+      "loss": 0.0972,
+      "step": 8936
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.34135688663358e-05,
+      "loss": 0.0618,
+      "step": 8937
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.33952744281816e-05,
+      "loss": 0.0999,
+      "step": 8938
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.337698021204443e-05,
+      "loss": 0.1117,
+      "step": 8939
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.33586862185393e-05,
+      "loss": 0.0757,
+      "step": 8940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.334039244828109e-05,
+      "loss": 0.2453,
+      "step": 8941
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.332209890188478e-05,
+      "loss": 0.0578,
+      "step": 8942
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.330380557996528e-05,
+      "loss": 0.1781,
+      "step": 8943
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.328551248313755e-05,
+      "loss": 0.0777,
+      "step": 8944
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.326721961201647e-05,
+      "loss": 0.1586,
+      "step": 8945
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.324892696721701e-05,
+      "loss": 0.0821,
+      "step": 8946
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.323063454935403e-05,
+      "loss": 0.0742,
+      "step": 8947
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.321234235904242e-05,
+      "loss": 0.0749,
+      "step": 8948
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.319405039689711e-05,
+      "loss": 0.1154,
+      "step": 8949
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.317575866353292e-05,
+      "loss": 0.2164,
+      "step": 8950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.315746715956479e-05,
+      "loss": 0.1461,
+      "step": 8951
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.313917588560754e-05,
+      "loss": 0.0781,
+      "step": 8952
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.312088484227605e-05,
+      "loss": 0.0597,
+      "step": 8953
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.31025940301852e-05,
+      "loss": 0.0826,
+      "step": 8954
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.308430344994976e-05,
+      "loss": 0.1037,
+      "step": 8955
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.306601310218464e-05,
+      "loss": 0.1417,
+      "step": 8956
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.304772298750463e-05,
+      "loss": 0.1361,
+      "step": 8957
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.302943310652457e-05,
+      "loss": 0.1142,
+      "step": 8958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.301114345985927e-05,
+      "loss": 0.0896,
+      "step": 8959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.299285404812349e-05,
+      "loss": 0.1089,
+      "step": 8960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.297456487193207e-05,
+      "loss": 0.0775,
+      "step": 8961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.295627593189979e-05,
+      "loss": 0.1425,
+      "step": 8962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.293798722864143e-05,
+      "loss": 0.113,
+      "step": 8963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.291969876277175e-05,
+      "loss": 0.2574,
+      "step": 8964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.290141053490553e-05,
+      "loss": 0.1472,
+      "step": 8965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.288312254565753e-05,
+      "loss": 0.0985,
+      "step": 8966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.286483479564245e-05,
+      "loss": 0.186,
+      "step": 8967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.284654728547512e-05,
+      "loss": 0.0634,
+      "step": 8968
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.28282600157702e-05,
+      "loss": 0.0665,
+      "step": 8969
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.280997298714245e-05,
+      "loss": 0.0819,
+      "step": 8970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.279168620020655e-05,
+      "loss": 0.0676,
+      "step": 8971
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.277339965557728e-05,
+      "loss": 0.1164,
+      "step": 8972
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.275511335386929e-05,
+      "loss": 0.0847,
+      "step": 8973
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.273682729569723e-05,
+      "loss": 0.0699,
+      "step": 8974
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.271854148167585e-05,
+      "loss": 0.1159,
+      "step": 8975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.270025591241977e-05,
+      "loss": 0.0966,
+      "step": 8976
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.268197058854373e-05,
+      "loss": 0.0669,
+      "step": 8977
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.266368551066233e-05,
+      "loss": 0.2769,
+      "step": 8978
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.264540067939022e-05,
+      "loss": 0.1119,
+      "step": 8979
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.26271160953421e-05,
+      "loss": 0.1476,
+      "step": 8980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.260883175913254e-05,
+      "loss": 0.1266,
+      "step": 8981
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.259054767137622e-05,
+      "loss": 0.0838,
+      "step": 8982
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.257226383268769e-05,
+      "loss": 0.1513,
+      "step": 8983
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.255398024368164e-05,
+      "loss": 0.1784,
+      "step": 8984
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.253569690497263e-05,
+      "loss": 0.129,
+      "step": 8985
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.251741381717521e-05,
+      "loss": 0.1112,
+      "step": 8986
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.249913098090403e-05,
+      "loss": 0.0957,
+      "step": 8987
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.248084839677362e-05,
+      "loss": 0.0824,
+      "step": 8988
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.24625660653986e-05,
+      "loss": 0.0924,
+      "step": 8989
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.244428398739345e-05,
+      "loss": 0.1903,
+      "step": 8990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.24260021633728e-05,
+      "loss": 0.055,
+      "step": 8991
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.240772059395112e-05,
+      "loss": 0.0693,
+      "step": 8992
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.238943927974299e-05,
+      "loss": 0.1149,
+      "step": 8993
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.237115822136295e-05,
+      "loss": 0.2335,
+      "step": 8994
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.235287741942546e-05,
+      "loss": 0.1095,
+      "step": 8995
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.233459687454508e-05,
+      "loss": 0.2078,
+      "step": 8996
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.23163165873363e-05,
+      "loss": 0.073,
+      "step": 8997
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.229803655841355e-05,
+      "loss": 0.094,
+      "step": 8998
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.227975678839138e-05,
+      "loss": 0.1089,
+      "step": 8999
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.226147727788421e-05,
+      "loss": 0.0719,
+      "step": 9000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.224319802750656e-05,
+      "loss": 0.252,
+      "step": 9001
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.222491903787283e-05,
+      "loss": 0.0835,
+      "step": 9002
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.220664030959749e-05,
+      "loss": 0.2254,
+      "step": 9003
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.218836184329498e-05,
+      "loss": 0.1263,
+      "step": 9004
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.217008363957974e-05,
+      "loss": 0.0971,
+      "step": 9005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.215180569906615e-05,
+      "loss": 0.0778,
+      "step": 9006
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.213352802236865e-05,
+      "loss": 0.2469,
+      "step": 9007
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.211525061010166e-05,
+      "loss": 0.0898,
+      "step": 9008
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.209697346287953e-05,
+      "loss": 0.0778,
+      "step": 9009
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.207869658131671e-05,
+      "loss": 0.0642,
+      "step": 9010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.206041996602751e-05,
+      "loss": 0.1573,
+      "step": 9011
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.204214361762629e-05,
+      "loss": 0.0558,
+      "step": 9012
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.202386753672746e-05,
+      "loss": 0.1345,
+      "step": 9013
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.200559172394531e-05,
+      "loss": 0.1114,
+      "step": 9014
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.198731617989425e-05,
+      "loss": 0.1103,
+      "step": 9015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.196904090518852e-05,
+      "loss": 0.0786,
+      "step": 9016
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.195076590044255e-05,
+      "loss": 0.0922,
+      "step": 9017
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.193249116627055e-05,
+      "loss": 0.0625,
+      "step": 9018
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.19142167032869e-05,
+      "loss": 0.0807,
+      "step": 9019
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.189594251210582e-05,
+      "loss": 0.0791,
+      "step": 9020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.187766859334166e-05,
+      "loss": 0.0677,
+      "step": 9021
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.185939494760868e-05,
+      "loss": 0.093,
+      "step": 9022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.184112157552117e-05,
+      "loss": 0.1034,
+      "step": 9023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.182284847769329e-05,
+      "loss": 0.159,
+      "step": 9024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.180457565473939e-05,
+      "loss": 0.0549,
+      "step": 9025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.178630310727365e-05,
+      "loss": 0.0845,
+      "step": 9026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.176803083591033e-05,
+      "loss": 0.1557,
+      "step": 9027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.174975884126365e-05,
+      "loss": 0.1681,
+      "step": 9028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.173148712394781e-05,
+      "loss": 0.1868,
+      "step": 9029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.171321568457698e-05,
+      "loss": 0.0991,
+      "step": 9030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.169494452376543e-05,
+      "loss": 0.1225,
+      "step": 9031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.167667364212726e-05,
+      "loss": 0.0779,
+      "step": 9032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.16584030402767e-05,
+      "loss": 0.1117,
+      "step": 9033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.16401327188279e-05,
+      "loss": 0.0566,
+      "step": 9034
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.1621862678395e-05,
+      "loss": 0.1379,
+      "step": 9035
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.160359291959219e-05,
+      "loss": 0.2571,
+      "step": 9036
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.158532344303357e-05,
+      "loss": 0.1754,
+      "step": 9037
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.156705424933323e-05,
+      "loss": 0.1996,
+      "step": 9038
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.154878533910534e-05,
+      "loss": 0.1805,
+      "step": 9039
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.153051671296397e-05,
+      "loss": 0.0707,
+      "step": 9040
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.151224837152327e-05,
+      "loss": 0.1297,
+      "step": 9041
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.149398031539727e-05,
+      "loss": 0.0939,
+      "step": 9042
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.14757125452001e-05,
+      "loss": 0.0622,
+      "step": 9043
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.145744506154579e-05,
+      "loss": 0.1219,
+      "step": 9044
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.143917786504843e-05,
+      "loss": 0.0764,
+      "step": 9045
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.142091095632204e-05,
+      "loss": 0.0591,
+      "step": 9046
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.140264433598065e-05,
+      "loss": 0.0607,
+      "step": 9047
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.138437800463835e-05,
+      "loss": 0.0858,
+      "step": 9048
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.136611196290915e-05,
+      "loss": 0.0876,
+      "step": 9049
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.134784621140698e-05,
+      "loss": 0.1341,
+      "step": 9050
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.132958075074592e-05,
+      "loss": 0.0979,
+      "step": 9051
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.131131558153992e-05,
+      "loss": 0.0592,
+      "step": 9052
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.129305070440299e-05,
+      "loss": 0.0632,
+      "step": 9053
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.127478611994908e-05,
+      "loss": 0.0698,
+      "step": 9054
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.125652182879218e-05,
+      "loss": 0.0823,
+      "step": 9055
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.12382578315462e-05,
+      "loss": 0.1521,
+      "step": 9056
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.121999412882513e-05,
+      "loss": 0.0853,
+      "step": 9057
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.120173072124286e-05,
+      "loss": 0.1549,
+      "step": 9058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.118346760941333e-05,
+      "loss": 0.068,
+      "step": 9059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.116520479395045e-05,
+      "loss": 0.0772,
+      "step": 9060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.114694227546811e-05,
+      "loss": 0.1484,
+      "step": 9061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.112868005458027e-05,
+      "loss": 0.1218,
+      "step": 9062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.111041813190074e-05,
+      "loss": 0.1526,
+      "step": 9063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.10921565080434e-05,
+      "loss": 0.0986,
+      "step": 9064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.107389518362212e-05,
+      "loss": 0.0659,
+      "step": 9065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.105563415925073e-05,
+      "loss": 0.0879,
+      "step": 9066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.103737343554315e-05,
+      "loss": 0.1431,
+      "step": 9067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.101911301311313e-05,
+      "loss": 0.0961,
+      "step": 9068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.100085289257452e-05,
+      "loss": 0.2529,
+      "step": 9069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.098259307454113e-05,
+      "loss": 0.0789,
+      "step": 9070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.096433355962678e-05,
+      "loss": 0.165,
+      "step": 9071
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.094607434844523e-05,
+      "loss": 0.0582,
+      "step": 9072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.092781544161029e-05,
+      "loss": 0.0576,
+      "step": 9073
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.09095568397357e-05,
+      "loss": 0.0856,
+      "step": 9074
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.08912985434353e-05,
+      "loss": 0.1331,
+      "step": 9075
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.087304055332271e-05,
+      "loss": 0.0923,
+      "step": 9076
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.085478287001177e-05,
+      "loss": 0.0881,
+      "step": 9077
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.083652549411617e-05,
+      "loss": 0.1236,
+      "step": 9078
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.081826842624964e-05,
+      "loss": 0.1212,
+      "step": 9079
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.080001166702588e-05,
+      "loss": 0.1964,
+      "step": 9080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.078175521705862e-05,
+      "loss": 0.0874,
+      "step": 9081
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.076349907696149e-05,
+      "loss": 0.0705,
+      "step": 9082
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.074524324734823e-05,
+      "loss": 0.237,
+      "step": 9083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.072698772883245e-05,
+      "loss": 0.065,
+      "step": 9084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.070873252202787e-05,
+      "loss": 0.0891,
+      "step": 9085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.069047762754808e-05,
+      "loss": 0.0668,
+      "step": 9086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.067222304600674e-05,
+      "loss": 0.0828,
+      "step": 9087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.065396877801755e-05,
+      "loss": 0.0546,
+      "step": 9088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.063571482419398e-05,
+      "loss": 0.0984,
+      "step": 9089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.061746118514968e-05,
+      "loss": 0.0792,
+      "step": 9090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.059920786149831e-05,
+      "loss": 0.1501,
+      "step": 9091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.058095485385339e-05,
+      "loss": 0.0932,
+      "step": 9092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.056270216282854e-05,
+      "loss": 0.1609,
+      "step": 9093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.054444978903727e-05,
+      "loss": 0.0922,
+      "step": 9094
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.052619773309317e-05,
+      "loss": 0.1207,
+      "step": 9095
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.050794599560976e-05,
+      "loss": 0.0962,
+      "step": 9096
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.04896945772006e-05,
+      "loss": 0.0916,
+      "step": 9097
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.047144347847915e-05,
+      "loss": 0.0564,
+      "step": 9098
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.0453192700059e-05,
+      "loss": 0.2075,
+      "step": 9099
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.043494224255356e-05,
+      "loss": 0.2487,
+      "step": 9100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.041669210657642e-05,
+      "loss": 0.0632,
+      "step": 9101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.039844229274096e-05,
+      "loss": 0.0942,
+      "step": 9102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.038019280166066e-05,
+      "loss": 0.1265,
+      "step": 9103
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.036194363394901e-05,
+      "loss": 0.1166,
+      "step": 9104
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.034369479021946e-05,
+      "loss": 0.0849,
+      "step": 9105
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.03254462710854e-05,
+      "loss": 0.0749,
+      "step": 9106
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.030719807716028e-05,
+      "loss": 0.0943,
+      "step": 9107
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.028895020905749e-05,
+      "loss": 0.1118,
+      "step": 9108
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.027070266739047e-05,
+      "loss": 0.0976,
+      "step": 9109
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.025245545277256e-05,
+      "loss": 0.0981,
+      "step": 9110
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.023420856581718e-05,
+      "loss": 0.0737,
+      "step": 9111
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.021596200713765e-05,
+      "loss": 0.0611,
+      "step": 9112
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.019771577734742e-05,
+      "loss": 0.0458,
+      "step": 9113
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.017946987705972e-05,
+      "loss": 0.0672,
+      "step": 9114
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.016122430688794e-05,
+      "loss": 0.075,
+      "step": 9115
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.014297906744537e-05,
+      "loss": 0.0569,
+      "step": 9116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.012473415934538e-05,
+      "loss": 0.1081,
+      "step": 9117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.01064895832012e-05,
+      "loss": 0.1554,
+      "step": 9118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.00882453396262e-05,
+      "loss": 0.1008,
+      "step": 9119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.007000142923357e-05,
+      "loss": 0.0647,
+      "step": 9120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.005175785263667e-05,
+      "loss": 0.1199,
+      "step": 9121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.003351461044867e-05,
+      "loss": 0.0931,
+      "step": 9122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.001527170328285e-05,
+      "loss": 0.0819,
+      "step": 9123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.999702913175246e-05,
+      "loss": 0.0918,
+      "step": 9124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.99787868964707e-05,
+      "loss": 0.0662,
+      "step": 9125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.996054499805082e-05,
+      "loss": 0.1121,
+      "step": 9126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.994230343710594e-05,
+      "loss": 0.1084,
+      "step": 9127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.992406221424931e-05,
+      "loss": 0.0689,
+      "step": 9128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.990582133009408e-05,
+      "loss": 0.1126,
+      "step": 9129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.988758078525342e-05,
+      "loss": 0.2697,
+      "step": 9130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.98693405803405e-05,
+      "loss": 0.1146,
+      "step": 9131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.985110071596843e-05,
+      "loss": 0.0988,
+      "step": 9132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.983286119275039e-05,
+      "loss": 0.1972,
+      "step": 9133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.981462201129945e-05,
+      "loss": 0.1393,
+      "step": 9134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.979638317222876e-05,
+      "loss": 0.0595,
+      "step": 9135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.977814467615137e-05,
+      "loss": 0.1084,
+      "step": 9136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.97599065236804e-05,
+      "loss": 0.0755,
+      "step": 9137
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.97416687154289e-05,
+      "loss": 0.0838,
+      "step": 9138
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.972343125201e-05,
+      "loss": 0.0652,
+      "step": 9139
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.970519413403667e-05,
+      "loss": 0.1225,
+      "step": 9140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.968695736212193e-05,
+      "loss": 0.0757,
+      "step": 9141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.966872093687889e-05,
+      "loss": 0.0992,
+      "step": 9142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.965048485892051e-05,
+      "loss": 0.1074,
+      "step": 9143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.963224912885979e-05,
+      "loss": 0.1206,
+      "step": 9144
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.961401374730977e-05,
+      "loss": 0.1014,
+      "step": 9145
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.959577871488339e-05,
+      "loss": 0.0835,
+      "step": 9146
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.957754403219363e-05,
+      "loss": 0.1093,
+      "step": 9147
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.955930969985344e-05,
+      "loss": 0.075,
+      "step": 9148
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.954107571847578e-05,
+      "loss": 0.1707,
+      "step": 9149
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.952284208867356e-05,
+      "loss": 0.1111,
+      "step": 9150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.950460881105973e-05,
+      "loss": 0.1167,
+      "step": 9151
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.948637588624721e-05,
+      "loss": 0.0677,
+      "step": 9152
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.946814331484883e-05,
+      "loss": 0.126,
+      "step": 9153
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.944991109747754e-05,
+      "loss": 0.0756,
+      "step": 9154
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.943167923474616e-05,
+      "loss": 0.0712,
+      "step": 9155
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.94134477272676e-05,
+      "loss": 0.2516,
+      "step": 9156
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.939521657565469e-05,
+      "loss": 0.0833,
+      "step": 9157
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.937698578052026e-05,
+      "loss": 0.0805,
+      "step": 9158
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.935875534247717e-05,
+      "loss": 0.0951,
+      "step": 9159
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.934052526213818e-05,
+      "loss": 0.0933,
+      "step": 9160
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.932229554011617e-05,
+      "loss": 0.1268,
+      "step": 9161
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.930406617702383e-05,
+      "loss": 0.1856,
+      "step": 9162
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.928583717347403e-05,
+      "loss": 0.0747,
+      "step": 9163
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.926760853007946e-05,
+      "loss": 0.1297,
+      "step": 9164
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.924938024745296e-05,
+      "loss": 0.123,
+      "step": 9165
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.92311523262072e-05,
+      "loss": 0.0532,
+      "step": 9166
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.921292476695492e-05,
+      "loss": 0.1409,
+      "step": 9167
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.919469757030885e-05,
+      "loss": 0.0823,
+      "step": 9168
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.917647073688167e-05,
+      "loss": 0.2914,
+      "step": 9169
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.915824426728612e-05,
+      "loss": 0.0932,
+      "step": 9170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.914001816213486e-05,
+      "loss": 0.0689,
+      "step": 9171
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.912179242204053e-05,
+      "loss": 0.0614,
+      "step": 9172
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.910356704761583e-05,
+      "loss": 0.1228,
+      "step": 9173
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.908534203947337e-05,
+      "loss": 0.1158,
+      "step": 9174
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.906711739822582e-05,
+      "loss": 0.0606,
+      "step": 9175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.904889312448575e-05,
+      "loss": 0.2669,
+      "step": 9176
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.90306692188658e-05,
+      "loss": 0.1002,
+      "step": 9177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.901244568197858e-05,
+      "loss": 0.082,
+      "step": 9178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.899422251443661e-05,
+      "loss": 0.0915,
+      "step": 9179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.897599971685252e-05,
+      "loss": 0.096,
+      "step": 9180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.895777728983881e-05,
+      "loss": 0.0616,
+      "step": 9181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.89395552340081e-05,
+      "loss": 0.0687,
+      "step": 9182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.892133354997284e-05,
+      "loss": 0.0558,
+      "step": 9183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.89031122383456e-05,
+      "loss": 0.0772,
+      "step": 9184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.888489129973889e-05,
+      "loss": 0.3406,
+      "step": 9185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.886667073476517e-05,
+      "loss": 0.0823,
+      "step": 9186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.884845054403699e-05,
+      "loss": 0.0659,
+      "step": 9187
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.883023072816673e-05,
+      "loss": 0.1353,
+      "step": 9188
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.881201128776692e-05,
+      "loss": 0.0811,
+      "step": 9189
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.879379222344994e-05,
+      "loss": 0.2258,
+      "step": 9190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.877557353582832e-05,
+      "loss": 0.099,
+      "step": 9191
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.875735522551439e-05,
+      "loss": 0.2553,
+      "step": 9192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.873913729312055e-05,
+      "loss": 0.0715,
+      "step": 9193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.872091973925924e-05,
+      "loss": 0.0786,
+      "step": 9194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.870270256454282e-05,
+      "loss": 0.1618,
+      "step": 9195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.868448576958366e-05,
+      "loss": 0.0762,
+      "step": 9196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.866626935499413e-05,
+      "loss": 0.0557,
+      "step": 9197
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.864805332138655e-05,
+      "loss": 0.0612,
+      "step": 9198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.862983766937327e-05,
+      "loss": 0.129,
+      "step": 9199
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.86116223995666e-05,
+      "loss": 0.0886,
+      "step": 9200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.859340751257884e-05,
+      "loss": 0.1244,
+      "step": 9201
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.857519300902228e-05,
+      "loss": 0.1165,
+      "step": 9202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.855697888950922e-05,
+      "loss": 0.0843,
+      "step": 9203
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.853876515465193e-05,
+      "loss": 0.0819,
+      "step": 9204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.852055180506258e-05,
+      "loss": 0.1123,
+      "step": 9205
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.850233884135351e-05,
+      "loss": 0.0699,
+      "step": 9206
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.848412626413688e-05,
+      "loss": 0.0704,
+      "step": 9207
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.846591407402494e-05,
+      "loss": 0.1131,
+      "step": 9208
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.844770227162988e-05,
+      "loss": 0.1453,
+      "step": 9209
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.84294908575639e-05,
+      "loss": 0.0761,
+      "step": 9210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.841127983243913e-05,
+      "loss": 0.0984,
+      "step": 9211
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.839306919686776e-05,
+      "loss": 0.0841,
+      "step": 9212
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.837485895146196e-05,
+      "loss": 0.0906,
+      "step": 9213
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.835664909683385e-05,
+      "loss": 0.069,
+      "step": 9214
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.833843963359555e-05,
+      "loss": 0.0858,
+      "step": 9215
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.832023056235916e-05,
+      "loss": 0.1227,
+      "step": 9216
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.830202188373682e-05,
+      "loss": 0.0645,
+      "step": 9217
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.828381359834056e-05,
+      "loss": 0.071,
+      "step": 9218
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.826560570678243e-05,
+      "loss": 0.0768,
+      "step": 9219
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.824739820967456e-05,
+      "loss": 0.069,
+      "step": 9220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.822919110762892e-05,
+      "loss": 0.0746,
+      "step": 9221
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.82109844012576e-05,
+      "loss": 0.0673,
+      "step": 9222
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.819277809117256e-05,
+      "loss": 0.0702,
+      "step": 9223
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.817457217798583e-05,
+      "loss": 0.0645,
+      "step": 9224
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.815636666230943e-05,
+      "loss": 0.2049,
+      "step": 9225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.813816154475529e-05,
+      "loss": 0.0668,
+      "step": 9226
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.81199568259354e-05,
+      "loss": 0.0925,
+      "step": 9227
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.81017525064617e-05,
+      "loss": 0.0818,
+      "step": 9228
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.808354858694612e-05,
+      "loss": 0.1276,
+      "step": 9229
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.806534506800064e-05,
+      "loss": 0.1236,
+      "step": 9230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.804714195023705e-05,
+      "loss": 0.2597,
+      "step": 9231
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.802893923426737e-05,
+      "loss": 0.0732,
+      "step": 9232
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.801073692070337e-05,
+      "loss": 0.06,
+      "step": 9233
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.799253501015699e-05,
+      "loss": 0.1508,
+      "step": 9234
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.797433350324007e-05,
+      "loss": 0.0771,
+      "step": 9235
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.795613240056447e-05,
+      "loss": 0.0618,
+      "step": 9236
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.793793170274196e-05,
+      "loss": 0.0898,
+      "step": 9237
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.79197314103844e-05,
+      "loss": 0.1312,
+      "step": 9238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.79015315241036e-05,
+      "loss": 0.0935,
+      "step": 9239
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.788333204451134e-05,
+      "loss": 0.1135,
+      "step": 9240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.786513297221936e-05,
+      "loss": 0.0705,
+      "step": 9241
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.78469343078395e-05,
+      "loss": 0.0608,
+      "step": 9242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.782873605198339e-05,
+      "loss": 0.11,
+      "step": 9243
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.781053820526284e-05,
+      "loss": 0.0881,
+      "step": 9244
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.779234076828952e-05,
+      "loss": 0.0683,
+      "step": 9245
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.777414374167521e-05,
+      "loss": 0.0554,
+      "step": 9246
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.775594712603151e-05,
+      "loss": 0.0749,
+      "step": 9247
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.773775092197017e-05,
+      "loss": 0.0938,
+      "step": 9248
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.771955513010281e-05,
+      "loss": 0.0868,
+      "step": 9249
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.77013597510411e-05,
+      "loss": 0.0856,
+      "step": 9250
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.768316478539667e-05,
+      "loss": 0.0635,
+      "step": 9251
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.766497023378111e-05,
+      "loss": 0.1352,
+      "step": 9252
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.76467760968061e-05,
+      "loss": 0.0629,
+      "step": 9253
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.762858237508318e-05,
+      "loss": 0.0635,
+      "step": 9254
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.761038906922398e-05,
+      "loss": 0.2722,
+      "step": 9255
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.759219617983999e-05,
+      "loss": 0.118,
+      "step": 9256
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.757400370754279e-05,
+      "loss": 0.1152,
+      "step": 9257
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.755581165294395e-05,
+      "loss": 0.2085,
+      "step": 9258
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.753762001665495e-05,
+      "loss": 0.1285,
+      "step": 9259
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.751942879928732e-05,
+      "loss": 0.0682,
+      "step": 9260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.750123800145257e-05,
+      "loss": 0.3587,
+      "step": 9261
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.748304762376215e-05,
+      "loss": 0.0801,
+      "step": 9262
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.746485766682756e-05,
+      "loss": 0.119,
+      "step": 9263
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.744666813126024e-05,
+      "loss": 0.0872,
+      "step": 9264
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.742847901767162e-05,
+      "loss": 0.0825,
+      "step": 9265
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.741029032667312e-05,
+      "loss": 0.1134,
+      "step": 9266
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.739210205887618e-05,
+      "loss": 0.0953,
+      "step": 9267
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.73739142148922e-05,
+      "loss": 0.0617,
+      "step": 9268
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.73557267953325e-05,
+      "loss": 0.1372,
+      "step": 9269
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.733753980080852e-05,
+      "loss": 0.061,
+      "step": 9270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.731935323193156e-05,
+      "loss": 0.0848,
+      "step": 9271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.730116708931299e-05,
+      "loss": 0.0778,
+      "step": 9272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.728298137356411e-05,
+      "loss": 0.0679,
+      "step": 9273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.726479608529627e-05,
+      "loss": 0.066,
+      "step": 9274
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.724661122512073e-05,
+      "loss": 0.1337,
+      "step": 9275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.72284267936488e-05,
+      "loss": 0.1116,
+      "step": 9276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.72102427914917e-05,
+      "loss": 0.1234,
+      "step": 9277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.719205921926072e-05,
+      "loss": 0.1017,
+      "step": 9278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.717387607756713e-05,
+      "loss": 0.11,
+      "step": 9279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.71556933670221e-05,
+      "loss": 0.0975,
+      "step": 9280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.713751108823688e-05,
+      "loss": 0.0529,
+      "step": 9281
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.711932924182264e-05,
+      "loss": 0.0961,
+      "step": 9282
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.710114782839051e-05,
+      "loss": 0.0629,
+      "step": 9283
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.708296684855176e-05,
+      "loss": 0.0891,
+      "step": 9284
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.706478630291746e-05,
+      "loss": 0.084,
+      "step": 9285
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.704660619209882e-05,
+      "loss": 0.1064,
+      "step": 9286
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.702842651670686e-05,
+      "loss": 0.1376,
+      "step": 9287
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.701024727735279e-05,
+      "loss": 0.1249,
+      "step": 9288
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.699206847464765e-05,
+      "loss": 0.2599,
+      "step": 9289
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.697389010920253e-05,
+      "loss": 0.1159,
+      "step": 9290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.695571218162847e-05,
+      "loss": 0.2181,
+      "step": 9291
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.693753469253656e-05,
+      "loss": 0.2851,
+      "step": 9292
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.691935764253783e-05,
+      "loss": 0.0736,
+      "step": 9293
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.69011810322433e-05,
+      "loss": 0.0977,
+      "step": 9294
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.68830048622639e-05,
+      "loss": 0.119,
+      "step": 9295
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.686482913321073e-05,
+      "loss": 0.0921,
+      "step": 9296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.684665384569467e-05,
+      "loss": 0.107,
+      "step": 9297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.682847900032675e-05,
+      "loss": 0.0883,
+      "step": 9298
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.681030459771787e-05,
+      "loss": 0.1241,
+      "step": 9299
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.6792130638479e-05,
+      "loss": 0.0741,
+      "step": 9300
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.6773957123221e-05,
+      "loss": 0.0524,
+      "step": 9301
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.675578405255485e-05,
+      "loss": 0.0694,
+      "step": 9302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.673761142709136e-05,
+      "loss": 0.1475,
+      "step": 9303
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.671943924744145e-05,
+      "loss": 0.0722,
+      "step": 9304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.670126751421592e-05,
+      "loss": 0.1425,
+      "step": 9305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.668309622802567e-05,
+      "loss": 0.064,
+      "step": 9306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.666492538948154e-05,
+      "loss": 0.0899,
+      "step": 9307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.664675499919428e-05,
+      "loss": 0.0638,
+      "step": 9308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.662858505777469e-05,
+      "loss": 0.0871,
+      "step": 9309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.661041556583357e-05,
+      "loss": 0.0998,
+      "step": 9310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.659224652398168e-05,
+      "loss": 0.1302,
+      "step": 9311
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.657407793282979e-05,
+      "loss": 0.0569,
+      "step": 9312
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.655590979298861e-05,
+      "loss": 0.0746,
+      "step": 9313
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.653774210506887e-05,
+      "loss": 0.1015,
+      "step": 9314
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.651957486968128e-05,
+      "loss": 0.0641,
+      "step": 9315
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.650140808743651e-05,
+      "loss": 0.0649,
+      "step": 9316
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.648324175894525e-05,
+      "loss": 0.0657,
+      "step": 9317
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.646507588481817e-05,
+      "loss": 0.0854,
+      "step": 9318
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.644691046566588e-05,
+      "loss": 0.0801,
+      "step": 9319
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.642874550209908e-05,
+      "loss": 0.117,
+      "step": 9320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.641058099472826e-05,
+      "loss": 0.1266,
+      "step": 9321
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.639241694416414e-05,
+      "loss": 0.0845,
+      "step": 9322
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.63742533510172e-05,
+      "loss": 0.1115,
+      "step": 9323
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.63560902158981e-05,
+      "loss": 0.0652,
+      "step": 9324
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.633792753941733e-05,
+      "loss": 0.1208,
+      "step": 9325
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.631976532218547e-05,
+      "loss": 0.0693,
+      "step": 9326
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.630160356481297e-05,
+      "loss": 0.2034,
+      "step": 9327
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.628344226791043e-05,
+      "loss": 0.1001,
+      "step": 9328
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.626528143208825e-05,
+      "loss": 0.1449,
+      "step": 9329
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.624712105795698e-05,
+      "loss": 0.0792,
+      "step": 9330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.622896114612702e-05,
+      "loss": 0.1235,
+      "step": 9331
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.621080169720883e-05,
+      "loss": 0.081,
+      "step": 9332
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.61926427118129e-05,
+      "loss": 0.0737,
+      "step": 9333
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.617448419054956e-05,
+      "loss": 0.0549,
+      "step": 9334
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.61563261340292e-05,
+      "loss": 0.1277,
+      "step": 9335
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.613816854286228e-05,
+      "loss": 0.079,
+      "step": 9336
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.612001141765908e-05,
+      "loss": 0.1319,
+      "step": 9337
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.610185475903002e-05,
+      "loss": 0.1819,
+      "step": 9338
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.608369856758537e-05,
+      "loss": 0.0971,
+      "step": 9339
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.60655428439355e-05,
+      "loss": 0.1148,
+      "step": 9340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.604738758869069e-05,
+      "loss": 0.0873,
+      "step": 9341
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.602923280246124e-05,
+      "loss": 0.0739,
+      "step": 9342
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.60110784858574e-05,
+      "loss": 0.1225,
+      "step": 9343
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.599292463948945e-05,
+      "loss": 0.081,
+      "step": 9344
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.59747712639676e-05,
+      "loss": 0.0729,
+      "step": 9345
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.595661835990212e-05,
+      "loss": 0.0986,
+      "step": 9346
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.593846592790318e-05,
+      "loss": 0.063,
+      "step": 9347
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.592031396858093e-05,
+      "loss": 0.1169,
+      "step": 9348
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.590216248254563e-05,
+      "loss": 0.1303,
+      "step": 9349
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.588401147040741e-05,
+      "loss": 0.1599,
+      "step": 9350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.586586093277638e-05,
+      "loss": 0.0924,
+      "step": 9351
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.584771087026273e-05,
+      "loss": 0.0858,
+      "step": 9352
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.58295612834765e-05,
+      "loss": 0.0595,
+      "step": 9353
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.581141217302785e-05,
+      "loss": 0.1131,
+      "step": 9354
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.579326353952684e-05,
+      "loss": 0.0917,
+      "step": 9355
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.577511538358353e-05,
+      "loss": 0.1569,
+      "step": 9356
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.575696770580794e-05,
+      "loss": 0.0639,
+      "step": 9357
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.573882050681016e-05,
+      "loss": 0.052,
+      "step": 9358
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.572067378720019e-05,
+      "loss": 0.0777,
+      "step": 9359
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.570252754758797e-05,
+      "loss": 0.0871,
+      "step": 9360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.568438178858354e-05,
+      "loss": 0.0928,
+      "step": 9361
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.566623651079686e-05,
+      "loss": 0.0699,
+      "step": 9362
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.564809171483788e-05,
+      "loss": 0.1426,
+      "step": 9363
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.562994740131654e-05,
+      "loss": 0.068,
+      "step": 9364
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.561180357084272e-05,
+      "loss": 0.1184,
+      "step": 9365
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.559366022402637e-05,
+      "loss": 0.163,
+      "step": 9366
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.557551736147736e-05,
+      "loss": 0.115,
+      "step": 9367
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.555737498380557e-05,
+      "loss": 0.0747,
+      "step": 9368
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.553923309162083e-05,
+      "loss": 0.0846,
+      "step": 9369
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.552109168553301e-05,
+      "loss": 0.15,
+      "step": 9370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.550295076615188e-05,
+      "loss": 0.1161,
+      "step": 9371
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.548481033408733e-05,
+      "loss": 0.0851,
+      "step": 9372
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.546667038994906e-05,
+      "loss": 0.062,
+      "step": 9373
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.544853093434689e-05,
+      "loss": 0.0711,
+      "step": 9374
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.543039196789053e-05,
+      "loss": 0.0732,
+      "step": 9375
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.541225349118979e-05,
+      "loss": 0.1158,
+      "step": 9376
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.539411550485433e-05,
+      "loss": 0.0883,
+      "step": 9377
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.537597800949388e-05,
+      "loss": 0.1068,
+      "step": 9378
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.535784100571814e-05,
+      "loss": 0.0872,
+      "step": 9379
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.533970449413679e-05,
+      "loss": 0.052,
+      "step": 9380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.532156847535944e-05,
+      "loss": 0.0938,
+      "step": 9381
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.53034329499958e-05,
+      "loss": 0.0559,
+      "step": 9382
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.528529791865541e-05,
+      "loss": 0.0733,
+      "step": 9383
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.526716338194795e-05,
+      "loss": 0.1635,
+      "step": 9384
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.5249029340483e-05,
+      "loss": 0.067,
+      "step": 9385
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.523089579487007e-05,
+      "loss": 0.0547,
+      "step": 9386
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.521276274571878e-05,
+      "loss": 0.0844,
+      "step": 9387
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.519463019363864e-05,
+      "loss": 0.0632,
+      "step": 9388
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.517649813923918e-05,
+      "loss": 0.0739,
+      "step": 9389
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.515836658312994e-05,
+      "loss": 0.0716,
+      "step": 9390
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.514023552592036e-05,
+      "loss": 0.0901,
+      "step": 9391
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.512210496821995e-05,
+      "loss": 0.0925,
+      "step": 9392
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.510397491063813e-05,
+      "loss": 0.0688,
+      "step": 9393
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.508584535378439e-05,
+      "loss": 0.0765,
+      "step": 9394
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.50677162982681e-05,
+      "loss": 0.0576,
+      "step": 9395
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.504958774469871e-05,
+      "loss": 0.0744,
+      "step": 9396
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.503145969368562e-05,
+      "loss": 0.0688,
+      "step": 9397
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.50133321458381e-05,
+      "loss": 0.0531,
+      "step": 9398
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.499520510176563e-05,
+      "loss": 0.0896,
+      "step": 9399
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.497707856207747e-05,
+      "loss": 0.0686,
+      "step": 9400
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.495895252738297e-05,
+      "loss": 0.0885,
+      "step": 9401
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.494082699829142e-05,
+      "loss": 0.0626,
+      "step": 9402
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.492270197541211e-05,
+      "loss": 0.0732,
+      "step": 9403
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.490457745935435e-05,
+      "loss": 0.1996,
+      "step": 9404
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.488645345072734e-05,
+      "loss": 0.0776,
+      "step": 9405
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.486832995014036e-05,
+      "loss": 0.0794,
+      "step": 9406
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.485020695820259e-05,
+      "loss": 0.1416,
+      "step": 9407
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.483208447552326e-05,
+      "loss": 0.1622,
+      "step": 9408
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.481396250271153e-05,
+      "loss": 0.0837,
+      "step": 9409
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.479584104037663e-05,
+      "loss": 0.1014,
+      "step": 9410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.477772008912763e-05,
+      "loss": 0.1109,
+      "step": 9411
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.475959964957368e-05,
+      "loss": 0.0745,
+      "step": 9412
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.474147972232394e-05,
+      "loss": 0.0883,
+      "step": 9413
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.472336030798745e-05,
+      "loss": 0.0613,
+      "step": 9414
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.470524140717334e-05,
+      "loss": 0.1299,
+      "step": 9415
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.468712302049068e-05,
+      "loss": 0.09,
+      "step": 9416
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.466900514854847e-05,
+      "loss": 0.0584,
+      "step": 9417
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.465088779195578e-05,
+      "loss": 0.0648,
+      "step": 9418
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.46327709513216e-05,
+      "loss": 0.0652,
+      "step": 9419
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.461465462725494e-05,
+      "loss": 0.0537,
+      "step": 9420
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.459653882036475e-05,
+      "loss": 0.0614,
+      "step": 9421
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.457842353126006e-05,
+      "loss": 0.0645,
+      "step": 9422
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.456030876054978e-05,
+      "loss": 0.0611,
+      "step": 9423
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.454219450884277e-05,
+      "loss": 0.058,
+      "step": 9424
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.452408077674801e-05,
+      "loss": 0.0614,
+      "step": 9425
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.450596756487435e-05,
+      "loss": 0.0572,
+      "step": 9426
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.448785487383072e-05,
+      "loss": 0.0568,
+      "step": 9427
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.446974270422591e-05,
+      "loss": 0.0868,
+      "step": 9428
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.445163105666878e-05,
+      "loss": 0.0821,
+      "step": 9429
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.44335199317682e-05,
+      "loss": 0.0849,
+      "step": 9430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.44154093301329e-05,
+      "loss": 0.1409,
+      "step": 9431
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.439729925237171e-05,
+      "loss": 0.0611,
+      "step": 9432
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.437918969909339e-05,
+      "loss": 0.0541,
+      "step": 9433
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.436108067090668e-05,
+      "loss": 0.0576,
+      "step": 9434
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.434297216842032e-05,
+      "loss": 0.073,
+      "step": 9435
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.432486419224306e-05,
+      "loss": 0.1042,
+      "step": 9436
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.430675674298355e-05,
+      "loss": 0.0808,
+      "step": 9437
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.428864982125046e-05,
+      "loss": 0.0645,
+      "step": 9438
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.427054342765248e-05,
+      "loss": 0.0587,
+      "step": 9439
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.425243756279824e-05,
+      "loss": 0.0841,
+      "step": 9440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.423433222729638e-05,
+      "loss": 0.0825,
+      "step": 9441
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.42162274217555e-05,
+      "loss": 0.1634,
+      "step": 9442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.419812314678419e-05,
+      "loss": 0.1473,
+      "step": 9443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.418001940299105e-05,
+      "loss": 0.1115,
+      "step": 9444
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.416191619098458e-05,
+      "loss": 0.0603,
+      "step": 9445
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.414381351137337e-05,
+      "loss": 0.0883,
+      "step": 9446
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.412571136476592e-05,
+      "loss": 0.102,
+      "step": 9447
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.410760975177072e-05,
+      "loss": 0.0659,
+      "step": 9448
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.40895086729963e-05,
+      "loss": 0.0587,
+      "step": 9449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.407140812905105e-05,
+      "loss": 0.1057,
+      "step": 9450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.405330812054349e-05,
+      "loss": 0.0664,
+      "step": 9451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.403520864808197e-05,
+      "loss": 0.0862,
+      "step": 9452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.401710971227498e-05,
+      "loss": 0.0957,
+      "step": 9453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.399901131373087e-05,
+      "loss": 0.0598,
+      "step": 9454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.398091345305803e-05,
+      "loss": 0.0765,
+      "step": 9455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.396281613086481e-05,
+      "loss": 0.0551,
+      "step": 9456
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.394471934775954e-05,
+      "loss": 0.0731,
+      "step": 9457
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.392662310435058e-05,
+      "loss": 0.1103,
+      "step": 9458
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.390852740124618e-05,
+      "loss": 0.1006,
+      "step": 9459
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.389043223905469e-05,
+      "loss": 0.0797,
+      "step": 9460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.38723376183843e-05,
+      "loss": 0.2746,
+      "step": 9461
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.385424353984335e-05,
+      "loss": 0.0615,
+      "step": 9462
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.383615000404e-05,
+      "loss": 0.1134,
+      "step": 9463
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.381805701158244e-05,
+      "loss": 0.0769,
+      "step": 9464
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.379996456307893e-05,
+      "loss": 0.1071,
+      "step": 9465
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.378187265913762e-05,
+      "loss": 0.234,
+      "step": 9466
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.376378130036667e-05,
+      "loss": 0.0655,
+      "step": 9467
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.374569048737419e-05,
+      "loss": 0.0567,
+      "step": 9468
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.372760022076833e-05,
+      "loss": 0.0631,
+      "step": 9469
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.370951050115721e-05,
+      "loss": 0.0698,
+      "step": 9470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.369142132914887e-05,
+      "loss": 0.1268,
+      "step": 9471
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.367333270535141e-05,
+      "loss": 0.0994,
+      "step": 9472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.365524463037285e-05,
+      "loss": 0.1125,
+      "step": 9473
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.363715710482126e-05,
+      "loss": 0.1357,
+      "step": 9474
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.361907012930464e-05,
+      "loss": 0.0687,
+      "step": 9475
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.36009837044309e-05,
+      "loss": 0.0571,
+      "step": 9476
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.358289783080811e-05,
+      "loss": 0.0545,
+      "step": 9477
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.356481250904418e-05,
+      "loss": 0.0863,
+      "step": 9478
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.354672773974706e-05,
+      "loss": 0.0754,
+      "step": 9479
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.352864352352465e-05,
+      "loss": 0.0668,
+      "step": 9480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.351055986098489e-05,
+      "loss": 0.0558,
+      "step": 9481
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.34924767527356e-05,
+      "loss": 0.0812,
+      "step": 9482
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.347439419938468e-05,
+      "loss": 0.0843,
+      "step": 9483
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.345631220153999e-05,
+      "loss": 0.0769,
+      "step": 9484
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.34382307598093e-05,
+      "loss": 0.0721,
+      "step": 9485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.342014987480047e-05,
+      "loss": 0.0716,
+      "step": 9486
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.340206954712125e-05,
+      "loss": 0.2017,
+      "step": 9487
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.338398977737945e-05,
+      "loss": 0.0649,
+      "step": 9488
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.336591056618278e-05,
+      "loss": 0.1017,
+      "step": 9489
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.334783191413896e-05,
+      "loss": 0.0904,
+      "step": 9490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.332975382185574e-05,
+      "loss": 0.1496,
+      "step": 9491
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.331167628994077e-05,
+      "loss": 0.0833,
+      "step": 9492
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.329359931900178e-05,
+      "loss": 0.1834,
+      "step": 9493
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.327552290964636e-05,
+      "loss": 0.0678,
+      "step": 9494
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.325744706248221e-05,
+      "loss": 0.069,
+      "step": 9495
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.32393717781169e-05,
+      "loss": 0.1566,
+      "step": 9496
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.322129705715804e-05,
+      "loss": 0.1467,
+      "step": 9497
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.320322290021324e-05,
+      "loss": 0.0547,
+      "step": 9498
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.318514930789001e-05,
+      "loss": 0.104,
+      "step": 9499
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.316707628079594e-05,
+      "loss": 0.0684,
+      "step": 9500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.314900381953856e-05,
+      "loss": 0.0732,
+      "step": 9501
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.31309319247253e-05,
+      "loss": 0.0784,
+      "step": 9502
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.31128605969637e-05,
+      "loss": 0.0985,
+      "step": 9503
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.30947898368612e-05,
+      "loss": 0.0534,
+      "step": 9504
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.307671964502528e-05,
+      "loss": 0.0632,
+      "step": 9505
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.305865002206332e-05,
+      "loss": 0.0995,
+      "step": 9506
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.304058096858279e-05,
+      "loss": 0.0875,
+      "step": 9507
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.302251248519102e-05,
+      "loss": 0.073,
+      "step": 9508
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.300444457249543e-05,
+      "loss": 0.058,
+      "step": 9509
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.298637723110332e-05,
+      "loss": 0.0701,
+      "step": 9510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.296831046162204e-05,
+      "loss": 0.0959,
+      "step": 9511
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.295024426465894e-05,
+      "loss": 0.0634,
+      "step": 9512
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.29321786408213e-05,
+      "loss": 0.1353,
+      "step": 9513
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.291411359071633e-05,
+      "loss": 0.0641,
+      "step": 9514
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.289604911495135e-05,
+      "loss": 0.1182,
+      "step": 9515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.287798521413357e-05,
+      "loss": 0.0625,
+      "step": 9516
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.285992188887022e-05,
+      "loss": 0.1413,
+      "step": 9517
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.284185913976846e-05,
+      "loss": 0.0918,
+      "step": 9518
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.282379696743552e-05,
+      "loss": 0.1081,
+      "step": 9519
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.280573537247852e-05,
+      "loss": 0.0849,
+      "step": 9520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.278767435550463e-05,
+      "loss": 0.0702,
+      "step": 9521
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.276961391712092e-05,
+      "loss": 0.1315,
+      "step": 9522
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.275155405793452e-05,
+      "loss": 0.1283,
+      "step": 9523
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.273349477855254e-05,
+      "loss": 0.0617,
+      "step": 9524
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.271543607958199e-05,
+      "loss": 0.081,
+      "step": 9525
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.269737796162997e-05,
+      "loss": 0.0702,
+      "step": 9526
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.267932042530342e-05,
+      "loss": 0.0889,
+      "step": 9527
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.266126347120938e-05,
+      "loss": 0.0688,
+      "step": 9528
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.264320709995486e-05,
+      "loss": 0.0658,
+      "step": 9529
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.262515131214677e-05,
+      "loss": 0.0932,
+      "step": 9530
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.260709610839211e-05,
+      "loss": 0.0701,
+      "step": 9531
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.258904148929775e-05,
+      "loss": 0.1015,
+      "step": 9532
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.257098745547064e-05,
+      "loss": 0.2167,
+      "step": 9533
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.255293400751763e-05,
+      "loss": 0.0745,
+      "step": 9534
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.25348811460456e-05,
+      "loss": 0.0727,
+      "step": 9535
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.251682887166139e-05,
+      "loss": 0.0645,
+      "step": 9536
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.249877718497182e-05,
+      "loss": 0.0896,
+      "step": 9537
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.248072608658373e-05,
+      "loss": 0.145,
+      "step": 9538
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.246267557710389e-05,
+      "loss": 0.0545,
+      "step": 9539
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.244462565713901e-05,
+      "loss": 0.0578,
+      "step": 9540
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.24265763272959e-05,
+      "loss": 0.0753,
+      "step": 9541
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.240852758818125e-05,
+      "loss": 0.0555,
+      "step": 9542
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.23904794404018e-05,
+      "loss": 0.0543,
+      "step": 9543
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.237243188456421e-05,
+      "loss": 0.0734,
+      "step": 9544
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.235438492127516e-05,
+      "loss": 0.0911,
+      "step": 9545
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.233633855114127e-05,
+      "loss": 0.062,
+      "step": 9546
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.231829277476921e-05,
+      "loss": 0.0785,
+      "step": 9547
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.230024759276554e-05,
+      "loss": 0.0862,
+      "step": 9548
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.22822030057369e-05,
+      "loss": 0.0946,
+      "step": 9549
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.22641590142898e-05,
+      "loss": 0.05,
+      "step": 9550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.224611561903081e-05,
+      "loss": 0.1067,
+      "step": 9551
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.222807282056651e-05,
+      "loss": 0.12,
+      "step": 9552
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.221003061950333e-05,
+      "loss": 0.076,
+      "step": 9553
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.219198901644777e-05,
+      "loss": 0.0637,
+      "step": 9554
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.217394801200631e-05,
+      "loss": 0.1038,
+      "step": 9555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.21559076067854e-05,
+      "loss": 0.0739,
+      "step": 9556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.213786780139147e-05,
+      "loss": 0.1242,
+      "step": 9557
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.211982859643091e-05,
+      "loss": 0.0653,
+      "step": 9558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.210178999251011e-05,
+      "loss": 0.1048,
+      "step": 9559
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.208375199023543e-05,
+      "loss": 0.0712,
+      "step": 9560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.206571459021324e-05,
+      "loss": 0.1283,
+      "step": 9561
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.204767779304983e-05,
+      "loss": 0.0679,
+      "step": 9562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.202964159935154e-05,
+      "loss": 0.0875,
+      "step": 9563
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.201160600972462e-05,
+      "loss": 0.0923,
+      "step": 9564
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.19935710247754e-05,
+      "loss": 0.0709,
+      "step": 9565
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.197553664511002e-05,
+      "loss": 0.2509,
+      "step": 9566
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.195750287133478e-05,
+      "loss": 0.0867,
+      "step": 9567
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.193946970405586e-05,
+      "loss": 0.1972,
+      "step": 9568
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.192143714387945e-05,
+      "loss": 0.1659,
+      "step": 9569
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.190340519141169e-05,
+      "loss": 0.0645,
+      "step": 9570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.188537384725877e-05,
+      "loss": 0.0857,
+      "step": 9571
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.186734311202674e-05,
+      "loss": 0.0587,
+      "step": 9572
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.184931298632176e-05,
+      "loss": 0.1331,
+      "step": 9573
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.183128347074988e-05,
+      "loss": 0.0584,
+      "step": 9574
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.18132545659172e-05,
+      "loss": 0.1182,
+      "step": 9575
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.17952262724297e-05,
+      "loss": 0.208,
+      "step": 9576
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.177719859089343e-05,
+      "loss": 0.058,
+      "step": 9577
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.175917152191447e-05,
+      "loss": 0.1116,
+      "step": 9578
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.174114506609864e-05,
+      "loss": 0.0883,
+      "step": 9579
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.172311922405197e-05,
+      "loss": 0.0847,
+      "step": 9580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.170509399638043e-05,
+      "loss": 0.055,
+      "step": 9581
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.168706938368987e-05,
+      "loss": 0.0981,
+      "step": 9582
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.166904538658625e-05,
+      "loss": 0.2488,
+      "step": 9583
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.165102200567538e-05,
+      "loss": 0.0603,
+      "step": 9584
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.163299924156318e-05,
+      "loss": 0.0689,
+      "step": 9585
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.161497709485543e-05,
+      "loss": 0.0519,
+      "step": 9586
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.159695556615799e-05,
+      "loss": 0.0922,
+      "step": 9587
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.15789346560766e-05,
+      "loss": 0.0683,
+      "step": 9588
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.156091436521707e-05,
+      "loss": 0.1457,
+      "step": 9589
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.154289469418514e-05,
+      "loss": 0.0463,
+      "step": 9590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.152487564358656e-05,
+      "loss": 0.0541,
+      "step": 9591
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.150685721402698e-05,
+      "loss": 0.091,
+      "step": 9592
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.148883940611211e-05,
+      "loss": 0.1083,
+      "step": 9593
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.147082222044763e-05,
+      "loss": 0.0764,
+      "step": 9594
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.14528056576392e-05,
+      "loss": 0.1128,
+      "step": 9595
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.143478971829241e-05,
+      "loss": 0.0643,
+      "step": 9596
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.14167744030129e-05,
+      "loss": 0.0597,
+      "step": 9597
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.13987597124062e-05,
+      "loss": 0.1069,
+      "step": 9598
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.138074564707793e-05,
+      "loss": 0.2456,
+      "step": 9599
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.13627322076336e-05,
+      "loss": 0.0728,
+      "step": 9600
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.134471939467874e-05,
+      "loss": 0.0864,
+      "step": 9601
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.132670720881884e-05,
+      "loss": 0.111,
+      "step": 9602
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.130869565065938e-05,
+      "loss": 0.0502,
+      "step": 9603
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.129068472080585e-05,
+      "loss": 0.0597,
+      "step": 9604
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.12726744198636e-05,
+      "loss": 0.0599,
+      "step": 9605
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.12546647484381e-05,
+      "loss": 0.0683,
+      "step": 9606
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.123665570713477e-05,
+      "loss": 0.0797,
+      "step": 9607
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.121864729655893e-05,
+      "loss": 0.0988,
+      "step": 9608
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.120063951731596e-05,
+      "loss": 0.0667,
+      "step": 9609
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.118263237001116e-05,
+      "loss": 0.0794,
+      "step": 9610
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.116462585524989e-05,
+      "loss": 0.292,
+      "step": 9611
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.114661997363737e-05,
+      "loss": 0.0851,
+      "step": 9612
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.112861472577893e-05,
+      "loss": 0.0573,
+      "step": 9613
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.111061011227978e-05,
+      "loss": 0.0996,
+      "step": 9614
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.109260613374514e-05,
+      "loss": 0.0635,
+      "step": 9615
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.107460279078021e-05,
+      "loss": 0.0619,
+      "step": 9616
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.105660008399023e-05,
+      "loss": 0.279,
+      "step": 9617
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.103859801398029e-05,
+      "loss": 0.0846,
+      "step": 9618
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.102059658135552e-05,
+      "loss": 0.066,
+      "step": 9619
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.100259578672105e-05,
+      "loss": 0.244,
+      "step": 9620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.098459563068202e-05,
+      "loss": 0.0714,
+      "step": 9621
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.096659611384342e-05,
+      "loss": 0.0818,
+      "step": 9622
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.09485972368104e-05,
+      "loss": 0.0654,
+      "step": 9623
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.093059900018792e-05,
+      "loss": 0.0597,
+      "step": 9624
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.091260140458103e-05,
+      "loss": 0.1444,
+      "step": 9625
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.089460445059467e-05,
+      "loss": 0.1085,
+      "step": 9626
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.087660813883385e-05,
+      "loss": 0.1496,
+      "step": 9627
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.085861246990347e-05,
+      "loss": 0.1366,
+      "step": 9628
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.084061744440851e-05,
+      "loss": 0.0848,
+      "step": 9629
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.082262306295386e-05,
+      "loss": 0.0672,
+      "step": 9630
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.080462932614433e-05,
+      "loss": 0.1044,
+      "step": 9631
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.078663623458484e-05,
+      "loss": 0.0706,
+      "step": 9632
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.076864378888021e-05,
+      "loss": 0.0866,
+      "step": 9633
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.075065198963523e-05,
+      "loss": 0.0922,
+      "step": 9634
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.073266083745474e-05,
+      "loss": 0.094,
+      "step": 9635
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.071467033294348e-05,
+      "loss": 0.0763,
+      "step": 9636
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.069668047670621e-05,
+      "loss": 0.0832,
+      "step": 9637
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.067869126934763e-05,
+      "loss": 0.13,
+      "step": 9638
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.06607027114725e-05,
+      "loss": 0.0734,
+      "step": 9639
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.064271480368544e-05,
+      "loss": 0.0538,
+      "step": 9640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.062472754659118e-05,
+      "loss": 0.0781,
+      "step": 9641
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.060674094079432e-05,
+      "loss": 0.0962,
+      "step": 9642
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.058875498689945e-05,
+      "loss": 0.0622,
+      "step": 9643
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.05707696855112e-05,
+      "loss": 0.1167,
+      "step": 9644
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.055278503723414e-05,
+      "loss": 0.056,
+      "step": 9645
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.053480104267281e-05,
+      "loss": 0.0529,
+      "step": 9646
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.051681770243175e-05,
+      "loss": 0.0809,
+      "step": 9647
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.049883501711546e-05,
+      "loss": 0.068,
+      "step": 9648
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.048085298732846e-05,
+      "loss": 0.0791,
+      "step": 9649
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.046287161367516e-05,
+      "loss": 0.0761,
+      "step": 9650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.044489089676004e-05,
+      "loss": 0.0827,
+      "step": 9651
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.04269108371875e-05,
+      "loss": 0.3125,
+      "step": 9652
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.040893143556196e-05,
+      "loss": 0.0815,
+      "step": 9653
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.039095269248776e-05,
+      "loss": 0.0557,
+      "step": 9654
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.037297460856931e-05,
+      "loss": 0.0811,
+      "step": 9655
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.03549971844109e-05,
+      "loss": 0.0855,
+      "step": 9656
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.03370204206168e-05,
+      "loss": 0.0725,
+      "step": 9657
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.031904431779138e-05,
+      "loss": 0.0756,
+      "step": 9658
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.030106887653883e-05,
+      "loss": 0.0545,
+      "step": 9659
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.028309409746344e-05,
+      "loss": 0.0974,
+      "step": 9660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.026511998116944e-05,
+      "loss": 0.0867,
+      "step": 9661
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.024714652826097e-05,
+      "loss": 0.1405,
+      "step": 9662
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.022917373934227e-05,
+      "loss": 0.1139,
+      "step": 9663
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.021120161501743e-05,
+      "loss": 0.0586,
+      "step": 9664
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.019323015589064e-05,
+      "loss": 0.093,
+      "step": 9665
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.017525936256597e-05,
+      "loss": 0.0883,
+      "step": 9666
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.015728923564754e-05,
+      "loss": 0.084,
+      "step": 9667
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.01393197757394e-05,
+      "loss": 0.0885,
+      "step": 9668
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.012135098344556e-05,
+      "loss": 0.0672,
+      "step": 9669
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.010338285937006e-05,
+      "loss": 0.1166,
+      "step": 9670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.00854154041169e-05,
+      "loss": 0.0787,
+      "step": 9671
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.006744861829005e-05,
+      "loss": 0.126,
+      "step": 9672
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.004948250249345e-05,
+      "loss": 0.0665,
+      "step": 9673
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.003151705733105e-05,
+      "loss": 0.0708,
+      "step": 9674
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.001355228340677e-05,
+      "loss": 0.0856,
+      "step": 9675
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.999558818132444e-05,
+      "loss": 0.1009,
+      "step": 9676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.997762475168797e-05,
+      "loss": 0.0797,
+      "step": 9677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.995966199510118e-05,
+      "loss": 0.1151,
+      "step": 9678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.99416999121679e-05,
+      "loss": 0.1349,
+      "step": 9679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.992373850349188e-05,
+      "loss": 0.062,
+      "step": 9680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.990577776967698e-05,
+      "loss": 0.0902,
+      "step": 9681
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.988781771132686e-05,
+      "loss": 0.0781,
+      "step": 9682
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.986985832904525e-05,
+      "loss": 0.0595,
+      "step": 9683
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.98518996234359e-05,
+      "loss": 0.0687,
+      "step": 9684
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.983394159510245e-05,
+      "loss": 0.0799,
+      "step": 9685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.981598424464857e-05,
+      "loss": 0.0612,
+      "step": 9686
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.979802757267792e-05,
+      "loss": 0.0868,
+      "step": 9687
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.978007157979405e-05,
+      "loss": 0.2045,
+      "step": 9688
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.976211626660062e-05,
+      "loss": 0.0676,
+      "step": 9689
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.974416163370115e-05,
+      "loss": 0.0549,
+      "step": 9690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.97262076816992e-05,
+      "loss": 0.0574,
+      "step": 9691
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.970825441119829e-05,
+      "loss": 0.0507,
+      "step": 9692
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.969030182280192e-05,
+      "loss": 0.0805,
+      "step": 9693
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.967234991711356e-05,
+      "loss": 0.0577,
+      "step": 9694
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.965439869473664e-05,
+      "loss": 0.0517,
+      "step": 9695
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.963644815627462e-05,
+      "loss": 0.0705,
+      "step": 9696
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.961849830233087e-05,
+      "loss": 0.1483,
+      "step": 9697
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.96005491335088e-05,
+      "loss": 0.0651,
+      "step": 9698
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.958260065041177e-05,
+      "loss": 0.1045,
+      "step": 9699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.95646528536431e-05,
+      "loss": 0.1064,
+      "step": 9700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.95467057438061e-05,
+      "loss": 0.0524,
+      "step": 9701
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.952875932150407e-05,
+      "loss": 0.0803,
+      "step": 9702
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.951081358734031e-05,
+      "loss": 0.0587,
+      "step": 9703
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.9492868541918e-05,
+      "loss": 0.0909,
+      "step": 9704
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.947492418584041e-05,
+      "loss": 0.1104,
+      "step": 9705
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.94569805197107e-05,
+      "loss": 0.081,
+      "step": 9706
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.94390375441321e-05,
+      "loss": 0.0723,
+      "step": 9707
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.942109525970773e-05,
+      "loss": 0.0686,
+      "step": 9708
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.940315366704067e-05,
+      "loss": 0.069,
+      "step": 9709
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.938521276673408e-05,
+      "loss": 0.1089,
+      "step": 9710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.936727255939102e-05,
+      "loss": 0.0879,
+      "step": 9711
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.934933304561457e-05,
+      "loss": 0.0534,
+      "step": 9712
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.933139422600773e-05,
+      "loss": 0.0778,
+      "step": 9713
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.931345610117353e-05,
+      "loss": 0.0996,
+      "step": 9714
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.929551867171499e-05,
+      "loss": 0.0574,
+      "step": 9715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.927758193823501e-05,
+      "loss": 0.091,
+      "step": 9716
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.925964590133659e-05,
+      "loss": 0.0563,
+      "step": 9717
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.92417105616226e-05,
+      "loss": 0.1656,
+      "step": 9718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.922377591969599e-05,
+      "loss": 0.0941,
+      "step": 9719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.92058419761596e-05,
+      "loss": 0.0966,
+      "step": 9720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.918790873161623e-05,
+      "loss": 0.0563,
+      "step": 9721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.91699761866688e-05,
+      "loss": 0.0612,
+      "step": 9722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.915204434192e-05,
+      "loss": 0.1059,
+      "step": 9723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.91341131979727e-05,
+      "loss": 0.0577,
+      "step": 9724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.91161827554296e-05,
+      "loss": 0.1686,
+      "step": 9725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.909825301489348e-05,
+      "loss": 0.0668,
+      "step": 9726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.908032397696698e-05,
+      "loss": 0.0933,
+      "step": 9727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.906239564225282e-05,
+      "loss": 0.0559,
+      "step": 9728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.904446801135368e-05,
+      "loss": 0.0809,
+      "step": 9729
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.902654108487214e-05,
+      "loss": 0.2587,
+      "step": 9730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.900861486341086e-05,
+      "loss": 0.0827,
+      "step": 9731
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.89906893475724e-05,
+      "loss": 0.0556,
+      "step": 9732
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.897276453795939e-05,
+      "loss": 0.1159,
+      "step": 9733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.895484043517427e-05,
+      "loss": 0.2271,
+      "step": 9734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.893691703981957e-05,
+      "loss": 0.1383,
+      "step": 9735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.891899435249786e-05,
+      "loss": 0.0911,
+      "step": 9736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.890107237381153e-05,
+      "loss": 0.0724,
+      "step": 9737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.888315110436308e-05,
+      "loss": 0.0859,
+      "step": 9738
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.88652305447549e-05,
+      "loss": 0.064,
+      "step": 9739
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.88473106955894e-05,
+      "loss": 0.0899,
+      "step": 9740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.882939155746893e-05,
+      "loss": 0.0866,
+      "step": 9741
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.881147313099586e-05,
+      "loss": 0.0896,
+      "step": 9742
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.879355541677254e-05,
+      "loss": 0.0763,
+      "step": 9743
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.877563841540122e-05,
+      "loss": 0.0667,
+      "step": 9744
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.875772212748423e-05,
+      "loss": 0.0545,
+      "step": 9745
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.873980655362382e-05,
+      "loss": 0.0721,
+      "step": 9746
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.872189169442215e-05,
+      "loss": 0.0888,
+      "step": 9747
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.87039775504815e-05,
+      "loss": 0.064,
+      "step": 9748
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.8686064122404e-05,
+      "loss": 0.0695,
+      "step": 9749
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.866815141079186e-05,
+      "loss": 0.1248,
+      "step": 9750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.865023941624716e-05,
+      "loss": 0.0649,
+      "step": 9751
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.863232813937205e-05,
+      "loss": 0.1082,
+      "step": 9752
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.861441758076859e-05,
+      "loss": 0.0763,
+      "step": 9753
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.859650774103889e-05,
+      "loss": 0.0813,
+      "step": 9754
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.857859862078492e-05,
+      "loss": 0.1224,
+      "step": 9755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.85606902206087e-05,
+      "loss": 0.0744,
+      "step": 9756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.854278254111228e-05,
+      "loss": 0.0567,
+      "step": 9757
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.852487558289756e-05,
+      "loss": 0.1454,
+      "step": 9758
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.850696934656657e-05,
+      "loss": 0.0903,
+      "step": 9759
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.84890638327211e-05,
+      "loss": 0.132,
+      "step": 9760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.847115904196311e-05,
+      "loss": 0.0634,
+      "step": 9761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.845325497489449e-05,
+      "loss": 0.0618,
+      "step": 9762
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.843535163211701e-05,
+      "loss": 0.081,
+      "step": 9763
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.841744901423257e-05,
+      "loss": 0.0765,
+      "step": 9764
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.83995471218429e-05,
+      "loss": 0.0759,
+      "step": 9765
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.838164595554983e-05,
+      "loss": 0.0831,
+      "step": 9766
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.836374551595502e-05,
+      "loss": 0.1059,
+      "step": 9767
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.834584580366027e-05,
+      "loss": 0.1839,
+      "step": 9768
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.832794681926726e-05,
+      "loss": 0.0983,
+      "step": 9769
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.831004856337764e-05,
+      "loss": 0.0684,
+      "step": 9770
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.829215103659309e-05,
+      "loss": 0.0864,
+      "step": 9771
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.827425423951523e-05,
+      "loss": 0.1064,
+      "step": 9772
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.82563581727456e-05,
+      "loss": 0.0551,
+      "step": 9773
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.823846283688583e-05,
+      "loss": 0.1284,
+      "step": 9774
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.822056823253745e-05,
+      "loss": 0.0889,
+      "step": 9775
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.8202674360302e-05,
+      "loss": 0.0807,
+      "step": 9776
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.818478122078096e-05,
+      "loss": 0.1213,
+      "step": 9777
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.816688881457582e-05,
+      "loss": 0.0928,
+      "step": 9778
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.814899714228802e-05,
+      "loss": 0.0792,
+      "step": 9779
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.813110620451902e-05,
+      "loss": 0.0634,
+      "step": 9780
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.811321600187016e-05,
+      "loss": 0.1727,
+      "step": 9781
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.809532653494285e-05,
+      "loss": 0.063,
+      "step": 9782
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.807743780433847e-05,
+      "loss": 0.0761,
+      "step": 9783
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.805954981065835e-05,
+      "loss": 0.0668,
+      "step": 9784
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.804166255450373e-05,
+      "loss": 0.0966,
+      "step": 9785
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.802377603647592e-05,
+      "loss": 0.1415,
+      "step": 9786
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.800589025717617e-05,
+      "loss": 0.1498,
+      "step": 9787
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.798800521720572e-05,
+      "loss": 0.1149,
+      "step": 9788
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.797012091716577e-05,
+      "loss": 0.0642,
+      "step": 9789
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.795223735765751e-05,
+      "loss": 0.104,
+      "step": 9790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.793435453928205e-05,
+      "loss": 0.0707,
+      "step": 9791
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.791647246264058e-05,
+      "loss": 0.0737,
+      "step": 9792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.789859112833414e-05,
+      "loss": 0.059,
+      "step": 9793
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.788071053696387e-05,
+      "loss": 0.0651,
+      "step": 9794
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.786283068913076e-05,
+      "loss": 0.0725,
+      "step": 9795
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.784495158543589e-05,
+      "loss": 0.2714,
+      "step": 9796
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.782707322648028e-05,
+      "loss": 0.1108,
+      "step": 9797
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.780919561286485e-05,
+      "loss": 0.0748,
+      "step": 9798
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.779131874519059e-05,
+      "loss": 0.0644,
+      "step": 9799
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.77734426240584e-05,
+      "loss": 0.1089,
+      "step": 9800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.775556725006921e-05,
+      "loss": 0.0734,
+      "step": 9801
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.773769262382391e-05,
+      "loss": 0.0543,
+      "step": 9802
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.77198187459233e-05,
+      "loss": 0.0596,
+      "step": 9803
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.770194561696828e-05,
+      "loss": 0.0661,
+      "step": 9804
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.768407323755957e-05,
+      "loss": 0.0645,
+      "step": 9805
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.766620160829803e-05,
+      "loss": 0.1026,
+      "step": 9806
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.764833072978435e-05,
+      "loss": 0.064,
+      "step": 9807
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.763046060261932e-05,
+      "loss": 0.0882,
+      "step": 9808
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.761259122740357e-05,
+      "loss": 0.0817,
+      "step": 9809
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.759472260473786e-05,
+      "loss": 0.0924,
+      "step": 9810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.757685473522276e-05,
+      "loss": 0.0973,
+      "step": 9811
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.755898761945893e-05,
+      "loss": 0.0662,
+      "step": 9812
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.754112125804696e-05,
+      "loss": 0.0621,
+      "step": 9813
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.752325565158746e-05,
+      "loss": 0.0758,
+      "step": 9814
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.750539080068094e-05,
+      "loss": 0.0717,
+      "step": 9815
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.748752670592796e-05,
+      "loss": 0.0552,
+      "step": 9816
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.746966336792897e-05,
+      "loss": 0.083,
+      "step": 9817
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.74518007872845e-05,
+      "loss": 0.0737,
+      "step": 9818
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.743393896459496e-05,
+      "loss": 0.2487,
+      "step": 9819
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.741607790046081e-05,
+      "loss": 0.0619,
+      "step": 9820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.73982175954824e-05,
+      "loss": 0.0787,
+      "step": 9821
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.738035805026012e-05,
+      "loss": 0.0648,
+      "step": 9822
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.736249926539439e-05,
+      "loss": 0.0493,
+      "step": 9823
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.734464124148539e-05,
+      "loss": 0.0826,
+      "step": 9824
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.732678397913351e-05,
+      "loss": 0.1098,
+      "step": 9825
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.7308927478939e-05,
+      "loss": 0.0628,
+      "step": 9826
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.729107174150209e-05,
+      "loss": 0.0938,
+      "step": 9827
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.727321676742303e-05,
+      "loss": 0.2675,
+      "step": 9828
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.725536255730198e-05,
+      "loss": 0.0592,
+      "step": 9829
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.723750911173911e-05,
+      "loss": 0.0656,
+      "step": 9830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.721965643133458e-05,
+      "loss": 0.0779,
+      "step": 9831
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.72018045166885e-05,
+      "loss": 0.0619,
+      "step": 9832
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.718395336840096e-05,
+      "loss": 0.0855,
+      "step": 9833
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.716610298707204e-05,
+      "loss": 0.0714,
+      "step": 9834
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.714825337330171e-05,
+      "loss": 0.1062,
+      "step": 9835
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.713040452769009e-05,
+      "loss": 0.0785,
+      "step": 9836
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.71125564508371e-05,
+      "loss": 0.0563,
+      "step": 9837
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.709470914334268e-05,
+      "loss": 0.0835,
+      "step": 9838
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.70768626058068e-05,
+      "loss": 0.0516,
+      "step": 9839
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.705901683882937e-05,
+      "loss": 0.2259,
+      "step": 9840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.704117184301026e-05,
+      "loss": 0.0551,
+      "step": 9841
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.702332761894934e-05,
+      "loss": 0.0888,
+      "step": 9842
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.700548416724644e-05,
+      "loss": 0.0705,
+      "step": 9843
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.698764148850137e-05,
+      "loss": 0.2254,
+      "step": 9844
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.696979958331388e-05,
+      "loss": 0.1389,
+      "step": 9845
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.695195845228377e-05,
+      "loss": 0.071,
+      "step": 9846
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.69341180960107e-05,
+      "loss": 0.0886,
+      "step": 9847
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.691627851509446e-05,
+      "loss": 0.0808,
+      "step": 9848
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.68984397101347e-05,
+      "loss": 0.083,
+      "step": 9849
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.688060168173099e-05,
+      "loss": 0.0667,
+      "step": 9850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.686276443048302e-05,
+      "loss": 0.0841,
+      "step": 9851
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.68449279569904e-05,
+      "loss": 0.1021,
+      "step": 9852
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.682709226185265e-05,
+      "loss": 0.1171,
+      "step": 9853
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.680925734566937e-05,
+      "loss": 0.0728,
+      "step": 9854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.679142320904004e-05,
+      "loss": 0.0954,
+      "step": 9855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.677358985256418e-05,
+      "loss": 0.0508,
+      "step": 9856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.675575727684122e-05,
+      "loss": 0.1672,
+      "step": 9857
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.673792548247063e-05,
+      "loss": 0.0824,
+      "step": 9858
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.67200944700518e-05,
+      "loss": 0.0737,
+      "step": 9859
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.670226424018415e-05,
+      "loss": 0.0474,
+      "step": 9860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.6684434793467e-05,
+      "loss": 0.0728,
+      "step": 9861
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.666660613049977e-05,
+      "loss": 0.1366,
+      "step": 9862
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.664877825188166e-05,
+      "loss": 0.0801,
+      "step": 9863
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.663095115821198e-05,
+      "loss": 0.0562,
+      "step": 9864
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.661312485009e-05,
+      "loss": 0.0823,
+      "step": 9865
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.659529932811498e-05,
+      "loss": 0.0635,
+      "step": 9866
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.657747459288608e-05,
+      "loss": 0.0546,
+      "step": 9867
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.65596506450025e-05,
+      "loss": 0.0919,
+      "step": 9868
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.654182748506336e-05,
+      "loss": 0.0717,
+      "step": 9869
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.652400511366785e-05,
+      "loss": 0.0895,
+      "step": 9870
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.650618353141496e-05,
+      "loss": 0.079,
+      "step": 9871
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.648836273890387e-05,
+      "loss": 0.0894,
+      "step": 9872
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.647054273673354e-05,
+      "loss": 0.0727,
+      "step": 9873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.645272352550305e-05,
+      "loss": 0.0517,
+      "step": 9874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.643490510581138e-05,
+      "loss": 0.0769,
+      "step": 9875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.641708747825744e-05,
+      "loss": 0.1216,
+      "step": 9876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.639927064344022e-05,
+      "loss": 0.0514,
+      "step": 9877
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.63814546019586e-05,
+      "loss": 0.0817,
+      "step": 9878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.636363935441148e-05,
+      "loss": 0.09,
+      "step": 9879
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.634582490139772e-05,
+      "loss": 0.0654,
+      "step": 9880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.632801124351615e-05,
+      "loss": 0.0764,
+      "step": 9881
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.631019838136557e-05,
+      "loss": 0.0827,
+      "step": 9882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.629238631554475e-05,
+      "loss": 0.0671,
+      "step": 9883
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.627457504665246e-05,
+      "loss": 0.0882,
+      "step": 9884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.62567645752874e-05,
+      "loss": 0.0517,
+      "step": 9885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.62389549020483e-05,
+      "loss": 0.0639,
+      "step": 9886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.622114602753379e-05,
+      "loss": 0.1061,
+      "step": 9887
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.620333795234256e-05,
+      "loss": 0.0978,
+      "step": 9888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.618553067707318e-05,
+      "loss": 0.0559,
+      "step": 9889
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.616772420232424e-05,
+      "loss": 0.1321,
+      "step": 9890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.614991852869434e-05,
+      "loss": 0.0557,
+      "step": 9891
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.613211365678198e-05,
+      "loss": 0.0696,
+      "step": 9892
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.611430958718568e-05,
+      "loss": 0.0785,
+      "step": 9893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.609650632050394e-05,
+      "loss": 0.0629,
+      "step": 9894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.60787038573352e-05,
+      "loss": 0.1017,
+      "step": 9895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.606090219827788e-05,
+      "loss": 0.0574,
+      "step": 9896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.604310134393039e-05,
+      "loss": 0.1067,
+      "step": 9897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.602530129489111e-05,
+      "loss": 0.0925,
+      "step": 9898
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.600750205175837e-05,
+      "loss": 0.0931,
+      "step": 9899
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.598970361513051e-05,
+      "loss": 0.0599,
+      "step": 9900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.597190598560583e-05,
+      "loss": 0.0568,
+      "step": 9901
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.595410916378254e-05,
+      "loss": 0.0625,
+      "step": 9902
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.593631315025894e-05,
+      "loss": 0.1517,
+      "step": 9903
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.591851794563317e-05,
+      "loss": 0.0843,
+      "step": 9904
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.590072355050348e-05,
+      "loss": 0.0706,
+      "step": 9905
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.5882929965468e-05,
+      "loss": 0.0726,
+      "step": 9906
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.586513719112485e-05,
+      "loss": 0.0611,
+      "step": 9907
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.584734522807218e-05,
+      "loss": 0.0954,
+      "step": 9908
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.582955407690801e-05,
+      "loss": 0.1406,
+      "step": 9909
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.581176373823041e-05,
+      "loss": 0.0869,
+      "step": 9910
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.579397421263738e-05,
+      "loss": 0.163,
+      "step": 9911
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.577618550072698e-05,
+      "loss": 0.107,
+      "step": 9912
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.575839760309711e-05,
+      "loss": 0.0828,
+      "step": 9913
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.57406105203457e-05,
+      "loss": 0.0582,
+      "step": 9914
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.57228242530707e-05,
+      "loss": 0.2905,
+      "step": 9915
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.570503880186996e-05,
+      "loss": 0.0542,
+      "step": 9916
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.568725416734136e-05,
+      "loss": 0.1408,
+      "step": 9917
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.56694703500827e-05,
+      "loss": 0.1817,
+      "step": 9918
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.565168735069181e-05,
+      "loss": 0.1025,
+      "step": 9919
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.563390516976648e-05,
+      "loss": 0.0714,
+      "step": 9920
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.561612380790441e-05,
+      "loss": 0.0905,
+      "step": 9921
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.559834326570337e-05,
+      "loss": 0.06,
+      "step": 9922
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.558056354376098e-05,
+      "loss": 0.0648,
+      "step": 9923
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.556278464267496e-05,
+      "loss": 0.0747,
+      "step": 9924
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.554500656304292e-05,
+      "loss": 0.1307,
+      "step": 9925
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.552722930546253e-05,
+      "loss": 0.0666,
+      "step": 9926
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.550945287053128e-05,
+      "loss": 0.0558,
+      "step": 9927
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.549167725884674e-05,
+      "loss": 0.0717,
+      "step": 9928
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.547390247100647e-05,
+      "loss": 0.0531,
+      "step": 9929
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.545612850760793e-05,
+      "loss": 0.158,
+      "step": 9930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.543835536924865e-05,
+      "loss": 0.0776,
+      "step": 9931
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.542058305652601e-05,
+      "loss": 0.1855,
+      "step": 9932
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.540281157003743e-05,
+      "loss": 0.0628,
+      "step": 9933
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.538504091038036e-05,
+      "loss": 0.061,
+      "step": 9934
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.536727107815207e-05,
+      "loss": 0.0835,
+      "step": 9935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.534950207394996e-05,
+      "loss": 0.0552,
+      "step": 9936
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.533173389837129e-05,
+      "loss": 0.0585,
+      "step": 9937
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.531396655201337e-05,
+      "loss": 0.0977,
+      "step": 9938
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.529620003547345e-05,
+      "loss": 0.0808,
+      "step": 9939
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.527843434934868e-05,
+      "loss": 0.0534,
+      "step": 9940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.526066949423633e-05,
+      "loss": 0.0844,
+      "step": 9941
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.524290547073349e-05,
+      "loss": 0.0842,
+      "step": 9942
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.522514227943739e-05,
+      "loss": 0.0841,
+      "step": 9943
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.520737992094505e-05,
+      "loss": 0.0737,
+      "step": 9944
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.51896183958536e-05,
+      "loss": 0.0593,
+      "step": 9945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.517185770476006e-05,
+      "loss": 0.4017,
+      "step": 9946
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.515409784826147e-05,
+      "loss": 0.0788,
+      "step": 9947
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.513633882695485e-05,
+      "loss": 0.0807,
+      "step": 9948
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.511858064143711e-05,
+      "loss": 0.0976,
+      "step": 9949
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.510082329230524e-05,
+      "loss": 0.1641,
+      "step": 9950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.508306678015612e-05,
+      "loss": 0.0674,
+      "step": 9951
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.50653111055867e-05,
+      "loss": 0.0616,
+      "step": 9952
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.504755626919374e-05,
+      "loss": 0.0897,
+      "step": 9953
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.50298022715741e-05,
+      "loss": 0.0586,
+      "step": 9954
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.50120491133246e-05,
+      "loss": 0.0739,
+      "step": 9955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.499429679504197e-05,
+      "loss": 0.0895,
+      "step": 9956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.4976545317323e-05,
+      "loss": 0.0558,
+      "step": 9957
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.495879468076437e-05,
+      "loss": 0.0767,
+      "step": 9958
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.494104488596278e-05,
+      "loss": 0.0606,
+      "step": 9959
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.49232959335149e-05,
+      "loss": 0.0846,
+      "step": 9960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.490554782401733e-05,
+      "loss": 0.0941,
+      "step": 9961
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.488780055806667e-05,
+      "loss": 0.0551,
+      "step": 9962
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.487005413625951e-05,
+      "loss": 0.0744,
+      "step": 9963
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.485230855919242e-05,
+      "loss": 0.1372,
+      "step": 9964
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.48345638274619e-05,
+      "loss": 0.1361,
+      "step": 9965
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.481681994166438e-05,
+      "loss": 0.057,
+      "step": 9966
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.479907690239637e-05,
+      "loss": 0.0676,
+      "step": 9967
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.478133471025428e-05,
+      "loss": 0.0557,
+      "step": 9968
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.476359336583454e-05,
+      "loss": 0.0776,
+      "step": 9969
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.474585286973349e-05,
+      "loss": 0.1296,
+      "step": 9970
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.472811322254749e-05,
+      "loss": 0.0796,
+      "step": 9971
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.471037442487286e-05,
+      "loss": 0.1006,
+      "step": 9972
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.469263647730586e-05,
+      "loss": 0.0545,
+      "step": 9973
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.467489938044281e-05,
+      "loss": 0.0556,
+      "step": 9974
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.465716313487989e-05,
+      "loss": 0.1724,
+      "step": 9975
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.463942774121331e-05,
+      "loss": 0.0839,
+      "step": 9976
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.462169320003924e-05,
+      "loss": 0.1461,
+      "step": 9977
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.460395951195388e-05,
+      "loss": 0.0744,
+      "step": 9978
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.458622667755327e-05,
+      "loss": 0.0643,
+      "step": 9979
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.45684946974335e-05,
+      "loss": 0.0638,
+      "step": 9980
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.455076357219069e-05,
+      "loss": 0.1294,
+      "step": 9981
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.45330333024208e-05,
+      "loss": 0.0896,
+      "step": 9982
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.451530388871988e-05,
+      "loss": 0.0615,
+      "step": 9983
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.449757533168387e-05,
+      "loss": 0.1019,
+      "step": 9984
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.447984763190874e-05,
+      "loss": 0.0651,
+      "step": 9985
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.446212078999039e-05,
+      "loss": 0.0627,
+      "step": 9986
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.444439480652469e-05,
+      "loss": 0.0896,
+      "step": 9987
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.442666968210755e-05,
+      "loss": 0.1072,
+      "step": 9988
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.440894541733475e-05,
+      "loss": 0.1049,
+      "step": 9989
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.43912220128021e-05,
+      "loss": 0.0897,
+      "step": 9990
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.437349946910541e-05,
+      "loss": 0.0717,
+      "step": 9991
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.435577778684033e-05,
+      "loss": 0.0554,
+      "step": 9992
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.433805696660266e-05,
+      "loss": 0.1456,
+      "step": 9993
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.432033700898801e-05,
+      "loss": 0.0921,
+      "step": 9994
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.430261791459211e-05,
+      "loss": 0.0821,
+      "step": 9995
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.428489968401053e-05,
+      "loss": 0.0903,
+      "step": 9996
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.42671823178389e-05,
+      "loss": 0.0732,
+      "step": 9997
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.424946581667276e-05,
+      "loss": 0.0708,
+      "step": 9998
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.423175018110766e-05,
+      "loss": 0.1092,
+      "step": 9999
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.42140354117391e-05,
+      "loss": 0.0945,
+      "step": 10000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.419632150916257e-05,
+      "loss": 0.3091,
+      "step": 10001
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.417860847397354e-05,
+      "loss": 0.0518,
+      "step": 10002
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.416089630676738e-05,
+      "loss": 0.0724,
+      "step": 10003
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.414318500813956e-05,
+      "loss": 0.0521,
+      "step": 10004
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.412547457868538e-05,
+      "loss": 0.0654,
+      "step": 10005
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.410776501900016e-05,
+      "loss": 0.0804,
+      "step": 10006
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.409005632967926e-05,
+      "loss": 0.0864,
+      "step": 10007
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.40723485113179e-05,
+      "loss": 0.0846,
+      "step": 10008
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.405464156451139e-05,
+      "loss": 0.0734,
+      "step": 10009
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.403693548985489e-05,
+      "loss": 0.0461,
+      "step": 10010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.401923028794363e-05,
+      "loss": 0.0581,
+      "step": 10011
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.400152595937273e-05,
+      "loss": 0.0564,
+      "step": 10012
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.398382250473732e-05,
+      "loss": 0.0718,
+      "step": 10013
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.396611992463256e-05,
+      "loss": 0.0575,
+      "step": 10014
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.394841821965345e-05,
+      "loss": 0.0917,
+      "step": 10015
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.393071739039507e-05,
+      "loss": 0.0654,
+      "step": 10016
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.391301743745245e-05,
+      "loss": 0.1115,
+      "step": 10017
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.38953183614205e-05,
+      "loss": 0.072,
+      "step": 10018
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.387762016289422e-05,
+      "loss": 0.0614,
+      "step": 10019
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.385992284246852e-05,
+      "loss": 0.0894,
+      "step": 10020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.38422264007383e-05,
+      "loss": 0.0589,
+      "step": 10021
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.382453083829842e-05,
+      "loss": 0.0563,
+      "step": 10022
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.380683615574373e-05,
+      "loss": 0.0513,
+      "step": 10023
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.378914235366901e-05,
+      "loss": 0.0896,
+      "step": 10024
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.377144943266905e-05,
+      "loss": 0.0685,
+      "step": 10025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.375375739333858e-05,
+      "loss": 0.0906,
+      "step": 10026
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.373606623627233e-05,
+      "loss": 0.0762,
+      "step": 10027
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.3718375962065e-05,
+      "loss": 0.0747,
+      "step": 10028
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.37006865713112e-05,
+      "loss": 0.1298,
+      "step": 10029
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.368299806460564e-05,
+      "loss": 0.0643,
+      "step": 10030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.366531044254284e-05,
+      "loss": 0.0656,
+      "step": 10031
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.364762370571738e-05,
+      "loss": 0.0562,
+      "step": 10032
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.362993785472382e-05,
+      "loss": 0.0728,
+      "step": 10033
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.361225289015664e-05,
+      "loss": 0.08,
+      "step": 10034
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.359456881261035e-05,
+      "loss": 0.2809,
+      "step": 10035
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.357688562267935e-05,
+      "loss": 0.0768,
+      "step": 10036
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.355920332095811e-05,
+      "loss": 0.0617,
+      "step": 10037
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.3541521908041e-05,
+      "loss": 0.0657,
+      "step": 10038
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.352384138452239e-05,
+      "loss": 0.0878,
+      "step": 10039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.350616175099656e-05,
+      "loss": 0.0752,
+      "step": 10040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.348848300805786e-05,
+      "loss": 0.0768,
+      "step": 10041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.347080515630059e-05,
+      "loss": 0.0549,
+      "step": 10042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.34531281963189e-05,
+      "loss": 0.075,
+      "step": 10043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.343545212870704e-05,
+      "loss": 0.1127,
+      "step": 10044
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.34177769540592e-05,
+      "loss": 0.0972,
+      "step": 10045
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.34001026729695e-05,
+      "loss": 0.0817,
+      "step": 10046
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.338242928603209e-05,
+      "loss": 0.0693,
+      "step": 10047
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.336475679384104e-05,
+      "loss": 0.0956,
+      "step": 10048
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.334708519699043e-05,
+      "loss": 0.1355,
+      "step": 10049
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.332941449607424e-05,
+      "loss": 0.1083,
+      "step": 10050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.331174469168653e-05,
+      "loss": 0.0768,
+      "step": 10051
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.329407578442123e-05,
+      "loss": 0.0716,
+      "step": 10052
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.327640777487231e-05,
+      "loss": 0.0688,
+      "step": 10053
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.325874066363362e-05,
+      "loss": 0.0786,
+      "step": 10054
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.324107445129912e-05,
+      "loss": 0.1308,
+      "step": 10055
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.322340913846257e-05,
+      "loss": 0.0989,
+      "step": 10056
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.320574472571785e-05,
+      "loss": 0.1065,
+      "step": 10057
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.31880812136587e-05,
+      "loss": 0.0602,
+      "step": 10058
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.317041860287893e-05,
+      "loss": 0.08,
+      "step": 10059
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.315275689397221e-05,
+      "loss": 0.0542,
+      "step": 10060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.313509608753231e-05,
+      "loss": 0.1327,
+      "step": 10061
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.311743618415281e-05,
+      "loss": 0.0745,
+      "step": 10062
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.30997771844274e-05,
+      "loss": 0.2027,
+      "step": 10063
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.308211908894967e-05,
+      "loss": 0.119,
+      "step": 10064
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.30644618983132e-05,
+      "loss": 0.0912,
+      "step": 10065
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.304680561311152e-05,
+      "loss": 0.0812,
+      "step": 10066
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.302915023393815e-05,
+      "loss": 0.0572,
+      "step": 10067
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.301149576138664e-05,
+      "loss": 0.0798,
+      "step": 10068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.299384219605032e-05,
+      "loss": 0.0754,
+      "step": 10069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.297618953852268e-05,
+      "loss": 0.0647,
+      "step": 10070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.295853778939712e-05,
+      "loss": 0.0663,
+      "step": 10071
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.294088694926698e-05,
+      "loss": 0.1662,
+      "step": 10072
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.292323701872559e-05,
+      "loss": 0.0807,
+      "step": 10073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.290558799836625e-05,
+      "loss": 0.0796,
+      "step": 10074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.288793988878226e-05,
+      "loss": 0.09,
+      "step": 10075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.287029269056681e-05,
+      "loss": 0.07,
+      "step": 10076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.285264640431316e-05,
+      "loss": 0.0677,
+      "step": 10077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.283500103061445e-05,
+      "loss": 0.0888,
+      "step": 10078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.281735657006384e-05,
+      "loss": 0.0733,
+      "step": 10079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.279971302325443e-05,
+      "loss": 0.0735,
+      "step": 10080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.278207039077938e-05,
+      "loss": 0.0702,
+      "step": 10081
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.276442867323166e-05,
+      "loss": 0.0829,
+      "step": 10082
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.274678787120428e-05,
+      "loss": 0.0844,
+      "step": 10083
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.27291479852903e-05,
+      "loss": 0.0953,
+      "step": 10084
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.271150901608266e-05,
+      "loss": 0.0935,
+      "step": 10085
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.269387096417427e-05,
+      "loss": 0.0736,
+      "step": 10086
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.267623383015807e-05,
+      "loss": 0.0658,
+      "step": 10087
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.26585976146269e-05,
+      "loss": 0.0943,
+      "step": 10088
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.264096231817362e-05,
+      "loss": 0.0635,
+      "step": 10089
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.2623327941391e-05,
+      "loss": 0.0644,
+      "step": 10090
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.260569448487187e-05,
+      "loss": 0.0563,
+      "step": 10091
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.258806194920894e-05,
+      "loss": 0.0776,
+      "step": 10092
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.257043033499495e-05,
+      "loss": 0.0622,
+      "step": 10093
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.255279964282259e-05,
+      "loss": 0.0632,
+      "step": 10094
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.253516987328446e-05,
+      "loss": 0.1384,
+      "step": 10095
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.25175410269732e-05,
+      "loss": 0.0492,
+      "step": 10096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.249991310448146e-05,
+      "loss": 0.0767,
+      "step": 10097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.248228610640171e-05,
+      "loss": 0.0764,
+      "step": 10098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.246466003332657e-05,
+      "loss": 0.0574,
+      "step": 10099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.244703488584848e-05,
+      "loss": 0.1434,
+      "step": 10100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.242941066455994e-05,
+      "loss": 0.0805,
+      "step": 10101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.241178737005335e-05,
+      "loss": 0.0613,
+      "step": 10102
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.239416500292114e-05,
+      "loss": 0.0666,
+      "step": 10103
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.237654356375568e-05,
+      "loss": 0.0671,
+      "step": 10104
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.235892305314931e-05,
+      "loss": 0.0867,
+      "step": 10105
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.234130347169434e-05,
+      "loss": 0.0894,
+      "step": 10106
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.232368481998309e-05,
+      "loss": 0.0753,
+      "step": 10107
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.230606709860774e-05,
+      "loss": 0.0957,
+      "step": 10108
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.228845030816052e-05,
+      "loss": 0.0986,
+      "step": 10109
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.227083444923364e-05,
+      "loss": 0.0829,
+      "step": 10110
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.225321952241926e-05,
+      "loss": 0.0679,
+      "step": 10111
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.223560552830948e-05,
+      "loss": 0.0706,
+      "step": 10112
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.221799246749643e-05,
+      "loss": 0.0503,
+      "step": 10113
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.220038034057213e-05,
+      "loss": 0.0532,
+      "step": 10114
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.218276914812863e-05,
+      "loss": 0.0523,
+      "step": 10115
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.216515889075792e-05,
+      "loss": 0.0742,
+      "step": 10116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.214754956905198e-05,
+      "loss": 0.1701,
+      "step": 10117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.212994118360272e-05,
+      "loss": 0.1237,
+      "step": 10118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.211233373500208e-05,
+      "loss": 0.1246,
+      "step": 10119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.209472722384193e-05,
+      "loss": 0.0838,
+      "step": 10120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.207712165071405e-05,
+      "loss": 0.1919,
+      "step": 10121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.205951701621032e-05,
+      "loss": 0.0554,
+      "step": 10122
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.204191332092246e-05,
+      "loss": 0.0621,
+      "step": 10123
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.202431056544225e-05,
+      "loss": 0.0613,
+      "step": 10124
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.200670875036142e-05,
+      "loss": 0.1462,
+      "step": 10125
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.19891078762716e-05,
+      "loss": 0.0643,
+      "step": 10126
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.19715079437645e-05,
+      "loss": 0.0751,
+      "step": 10127
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.195390895343171e-05,
+      "loss": 0.1053,
+      "step": 10128
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.193631090586484e-05,
+      "loss": 0.0662,
+      "step": 10129
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.191871380165538e-05,
+      "loss": 0.0802,
+      "step": 10130
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.190111764139495e-05,
+      "loss": 0.0851,
+      "step": 10131
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.188352242567495e-05,
+      "loss": 0.1561,
+      "step": 10132
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.186592815508694e-05,
+      "loss": 0.0796,
+      "step": 10133
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.184833483022229e-05,
+      "loss": 0.0653,
+      "step": 10134
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.183074245167236e-05,
+      "loss": 0.0514,
+      "step": 10135
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.181315102002858e-05,
+      "loss": 0.07,
+      "step": 10136
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.179556053588225e-05,
+      "loss": 0.0866,
+      "step": 10137
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.177797099982467e-05,
+      "loss": 0.0553,
+      "step": 10138
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.176038241244716e-05,
+      "loss": 0.0561,
+      "step": 10139
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.17427947743409e-05,
+      "loss": 0.0634,
+      "step": 10140
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.172520808609712e-05,
+      "loss": 0.0889,
+      "step": 10141
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.170762234830699e-05,
+      "loss": 0.0849,
+      "step": 10142
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.169003756156168e-05,
+      "loss": 0.0883,
+      "step": 10143
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.167245372645224e-05,
+      "loss": 0.1356,
+      "step": 10144
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.165487084356982e-05,
+      "loss": 0.0813,
+      "step": 10145
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.163728891350546e-05,
+      "loss": 0.0827,
+      "step": 10146
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.161970793685007e-05,
+      "loss": 0.1002,
+      "step": 10147
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.160212791419474e-05,
+      "loss": 0.0595,
+      "step": 10148
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.158454884613037e-05,
+      "loss": 0.0802,
+      "step": 10149
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.156697073324791e-05,
+      "loss": 0.0835,
+      "step": 10150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.154939357613824e-05,
+      "loss": 0.0517,
+      "step": 10151
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.153181737539219e-05,
+      "loss": 0.1011,
+      "step": 10152
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.151424213160061e-05,
+      "loss": 0.0746,
+      "step": 10153
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.149666784535425e-05,
+      "loss": 0.0652,
+      "step": 10154
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.147909451724392e-05,
+      "loss": 0.0776,
+      "step": 10155
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.146152214786032e-05,
+      "loss": 0.0466,
+      "step": 10156
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.144395073779416e-05,
+      "loss": 0.0575,
+      "step": 10157
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.142638028763605e-05,
+      "loss": 0.0714,
+      "step": 10158
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.140881079797668e-05,
+      "loss": 0.0553,
+      "step": 10159
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.139124226940662e-05,
+      "loss": 0.1493,
+      "step": 10160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.137367470251641e-05,
+      "loss": 0.1836,
+      "step": 10161
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.135610809789663e-05,
+      "loss": 0.0734,
+      "step": 10162
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.133854245613771e-05,
+      "loss": 0.0573,
+      "step": 10163
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.132097777783018e-05,
+      "loss": 0.0853,
+      "step": 10164
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.130341406356446e-05,
+      "loss": 0.0683,
+      "step": 10165
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.128585131393095e-05,
+      "loss": 0.1354,
+      "step": 10166
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.126828952952002e-05,
+      "loss": 0.0706,
+      "step": 10167
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.125072871092199e-05,
+      "loss": 0.0596,
+      "step": 10168
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.12331688587272e-05,
+      "loss": 0.0874,
+      "step": 10169
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.121560997352587e-05,
+      "loss": 0.0618,
+      "step": 10170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.119805205590831e-05,
+      "loss": 0.0985,
+      "step": 10171
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.118049510646469e-05,
+      "loss": 0.0615,
+      "step": 10172
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.116293912578513e-05,
+      "loss": 0.0816,
+      "step": 10173
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.114538411445988e-05,
+      "loss": 0.0737,
+      "step": 10174
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.112783007307896e-05,
+      "loss": 0.113,
+      "step": 10175
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.11102770022325e-05,
+      "loss": 0.0691,
+      "step": 10176
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.109272490251049e-05,
+      "loss": 0.065,
+      "step": 10177
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.107517377450299e-05,
+      "loss": 0.1076,
+      "step": 10178
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.10576236188e-05,
+      "loss": 0.0918,
+      "step": 10179
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.104007443599139e-05,
+      "loss": 0.0902,
+      "step": 10180
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.102252622666714e-05,
+      "loss": 0.0619,
+      "step": 10181
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.10049789914171e-05,
+      "loss": 0.074,
+      "step": 10182
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.098743273083112e-05,
+      "loss": 0.0662,
+      "step": 10183
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.096988744549908e-05,
+      "loss": 0.057,
+      "step": 10184
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.095234313601063e-05,
+      "loss": 0.0748,
+      "step": 10185
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.093479980295563e-05,
+      "loss": 0.066,
+      "step": 10186
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.091725744692374e-05,
+      "loss": 0.0797,
+      "step": 10187
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.089971606850468e-05,
+      "loss": 0.0807,
+      "step": 10188
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.088217566828805e-05,
+      "loss": 0.064,
+      "step": 10189
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.086463624686355e-05,
+      "loss": 0.0614,
+      "step": 10190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.084709780482069e-05,
+      "loss": 0.0609,
+      "step": 10191
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.082956034274903e-05,
+      "loss": 0.0566,
+      "step": 10192
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.081202386123817e-05,
+      "loss": 0.1076,
+      "step": 10193
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.079448836087748e-05,
+      "loss": 0.0536,
+      "step": 10194
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.077695384225652e-05,
+      "loss": 0.0752,
+      "step": 10195
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.075942030596462e-05,
+      "loss": 0.0636,
+      "step": 10196
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.074188775259128e-05,
+      "loss": 0.0645,
+      "step": 10197
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.072435618272575e-05,
+      "loss": 0.0526,
+      "step": 10198
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.070682559695736e-05,
+      "loss": 0.135,
+      "step": 10199
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.068929599587545e-05,
+      "loss": 0.0931,
+      "step": 10200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.067176738006923e-05,
+      "loss": 0.1509,
+      "step": 10201
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.065423975012797e-05,
+      "loss": 0.081,
+      "step": 10202
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.06367131066408e-05,
+      "loss": 0.0552,
+      "step": 10203
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.061918745019691e-05,
+      "loss": 0.0741,
+      "step": 10204
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.060166278138545e-05,
+      "loss": 0.0594,
+      "step": 10205
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.058413910079544e-05,
+      "loss": 0.0533,
+      "step": 10206
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.056661640901601e-05,
+      "loss": 0.06,
+      "step": 10207
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.054909470663613e-05,
+      "loss": 0.094,
+      "step": 10208
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.053157399424483e-05,
+      "loss": 0.0601,
+      "step": 10209
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.051405427243105e-05,
+      "loss": 0.1513,
+      "step": 10210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.049653554178369e-05,
+      "loss": 0.0587,
+      "step": 10211
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.047901780289168e-05,
+      "loss": 0.0923,
+      "step": 10212
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.046150105634381e-05,
+      "loss": 0.0616,
+      "step": 10213
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.0443985302729e-05,
+      "loss": 0.0549,
+      "step": 10214
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.042647054263596e-05,
+      "loss": 0.0519,
+      "step": 10215
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.040895677665349e-05,
+      "loss": 0.0558,
+      "step": 10216
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.039144400537028e-05,
+      "loss": 0.1146,
+      "step": 10217
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.037393222937503e-05,
+      "loss": 0.0726,
+      "step": 10218
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.035642144925642e-05,
+      "loss": 0.0945,
+      "step": 10219
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.033891166560307e-05,
+      "loss": 0.0803,
+      "step": 10220
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.032140287900355e-05,
+      "loss": 0.0652,
+      "step": 10221
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.03038950900464e-05,
+      "loss": 0.0655,
+      "step": 10222
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.028638829932021e-05,
+      "loss": 0.05,
+      "step": 10223
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.026888250741342e-05,
+      "loss": 0.0798,
+      "step": 10224
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.025137771491446e-05,
+      "loss": 0.0598,
+      "step": 10225
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.023387392241179e-05,
+      "loss": 0.0637,
+      "step": 10226
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.021637113049376e-05,
+      "loss": 0.0672,
+      "step": 10227
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.019886933974878e-05,
+      "loss": 0.0784,
+      "step": 10228
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.018136855076514e-05,
+      "loss": 0.076,
+      "step": 10229
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.016386876413112e-05,
+      "loss": 0.062,
+      "step": 10230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.014636998043498e-05,
+      "loss": 0.0622,
+      "step": 10231
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.012887220026494e-05,
+      "loss": 0.058,
+      "step": 10232
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.01113754242092e-05,
+      "loss": 0.0712,
+      "step": 10233
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.009387965285588e-05,
+      "loss": 0.0517,
+      "step": 10234
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.007638488679313e-05,
+      "loss": 0.0545,
+      "step": 10235
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.005889112660907e-05,
+      "loss": 0.077,
+      "step": 10236
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.004139837289163e-05,
+      "loss": 0.0629,
+      "step": 10237
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.002390662622894e-05,
+      "loss": 0.0877,
+      "step": 10238
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.000641588720891e-05,
+      "loss": 0.0667,
+      "step": 10239
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.998892615641955e-05,
+      "loss": 0.0577,
+      "step": 10240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.997143743444873e-05,
+      "loss": 0.0616,
+      "step": 10241
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.995394972188436e-05,
+      "loss": 0.0606,
+      "step": 10242
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.993646301931424e-05,
+      "loss": 0.0871,
+      "step": 10243
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.991897732732626e-05,
+      "loss": 0.0626,
+      "step": 10244
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.990149264650814e-05,
+      "loss": 0.0712,
+      "step": 10245
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.988400897744762e-05,
+      "loss": 0.0521,
+      "step": 10246
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.986652632073248e-05,
+      "loss": 0.0719,
+      "step": 10247
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.984904467695033e-05,
+      "loss": 0.0564,
+      "step": 10248
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.983156404668886e-05,
+      "loss": 0.0572,
+      "step": 10249
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.981408443053563e-05,
+      "loss": 0.0653,
+      "step": 10250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.979660582907824e-05,
+      "loss": 0.0609,
+      "step": 10251
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.977912824290424e-05,
+      "loss": 0.063,
+      "step": 10252
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.976165167260111e-05,
+      "loss": 0.0652,
+      "step": 10253
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.974417611875634e-05,
+      "loss": 0.0514,
+      "step": 10254
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.972670158195738e-05,
+      "loss": 0.0694,
+      "step": 10255
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.970922806279163e-05,
+      "loss": 0.0588,
+      "step": 10256
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.969175556184641e-05,
+      "loss": 0.0854,
+      "step": 10257
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.967428407970914e-05,
+      "loss": 0.0575,
+      "step": 10258
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.965681361696708e-05,
+      "loss": 0.0551,
+      "step": 10259
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.963934417420747e-05,
+      "loss": 0.0633,
+      "step": 10260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.96218757520176e-05,
+      "loss": 0.0674,
+      "step": 10261
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.960440835098468e-05,
+      "loss": 0.0637,
+      "step": 10262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.958694197169578e-05,
+      "loss": 0.0556,
+      "step": 10263
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.95694766147381e-05,
+      "loss": 0.0558,
+      "step": 10264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.95520122806987e-05,
+      "loss": 0.0833,
+      "step": 10265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.95345489701647e-05,
+      "loss": 0.0658,
+      "step": 10266
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.951708668372303e-05,
+      "loss": 0.0647,
+      "step": 10267
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.94996254219608e-05,
+      "loss": 0.0616,
+      "step": 10268
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.948216518546487e-05,
+      "loss": 0.0873,
+      "step": 10269
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.946470597482225e-05,
+      "loss": 0.0712,
+      "step": 10270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.944724779061973e-05,
+      "loss": 0.0746,
+      "step": 10271
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.942979063344424e-05,
+      "loss": 0.0628,
+      "step": 10272
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.941233450388259e-05,
+      "loss": 0.0633,
+      "step": 10273
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.939487940252154e-05,
+      "loss": 0.08,
+      "step": 10274
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.937742532994788e-05,
+      "loss": 0.0903,
+      "step": 10275
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.93599722867483e-05,
+      "loss": 0.0628,
+      "step": 10276
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.934252027350944e-05,
+      "loss": 0.0617,
+      "step": 10277
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.932506929081803e-05,
+      "loss": 0.0691,
+      "step": 10278
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.930761933926061e-05,
+      "loss": 0.0688,
+      "step": 10279
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.929017041942383e-05,
+      "loss": 0.059,
+      "step": 10280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.927272253189416e-05,
+      "loss": 0.2024,
+      "step": 10281
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.925527567725818e-05,
+      "loss": 0.0684,
+      "step": 10282
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.92378298561023e-05,
+      "loss": 0.0493,
+      "step": 10283
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.922038506901299e-05,
+      "loss": 0.0638,
+      "step": 10284
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.920294131657664e-05,
+      "loss": 0.0717,
+      "step": 10285
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.918549859937963e-05,
+      "loss": 0.0627,
+      "step": 10286
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.916805691800832e-05,
+      "loss": 0.0551,
+      "step": 10287
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.915061627304899e-05,
+      "loss": 0.0756,
+      "step": 10288
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.913317666508789e-05,
+      "loss": 0.0514,
+      "step": 10289
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.911573809471126e-05,
+      "loss": 0.0716,
+      "step": 10290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 0.063,
+      "step": 10291
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.908086406905613e-05,
+      "loss": 0.0672,
+      "step": 10292
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.906342861494991e-05,
+      "loss": 0.0782,
+      "step": 10293
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.904599420077277e-05,
+      "loss": 0.0661,
+      "step": 10294
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.902856082711071e-05,
+      "loss": 0.0495,
+      "step": 10295
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.901112849454978e-05,
+      "loss": 0.0567,
+      "step": 10296
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.899369720367594e-05,
+      "loss": 0.0602,
+      "step": 10297
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.897626695507518e-05,
+      "loss": 0.0995,
+      "step": 10298
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.895883774933336e-05,
+      "loss": 0.0606,
+      "step": 10299
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.894140958703638e-05,
+      "loss": 0.0699,
+      "step": 10300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.892398246877016e-05,
+      "loss": 0.0623,
+      "step": 10301
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.890655639512043e-05,
+      "loss": 0.1268,
+      "step": 10302
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.888913136667295e-05,
+      "loss": 0.0999,
+      "step": 10303
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.887170738401352e-05,
+      "loss": 0.0583,
+      "step": 10304
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.885428444772778e-05,
+      "loss": 0.0938,
+      "step": 10305
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.883686255840147e-05,
+      "loss": 0.0736,
+      "step": 10306
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.881944171662018e-05,
+      "loss": 0.1277,
+      "step": 10307
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.880202192296954e-05,
+      "loss": 0.0874,
+      "step": 10308
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.878460317803505e-05,
+      "loss": 0.0878,
+      "step": 10309
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.876718548240233e-05,
+      "loss": 0.0614,
+      "step": 10310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.874976883665682e-05,
+      "loss": 0.1487,
+      "step": 10311
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.873235324138396e-05,
+      "loss": 0.06,
+      "step": 10312
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.871493869716928e-05,
+      "loss": 0.0703,
+      "step": 10313
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.869752520459803e-05,
+      "loss": 0.0588,
+      "step": 10314
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.868011276425559e-05,
+      "loss": 0.0878,
+      "step": 10315
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.866270137672735e-05,
+      "loss": 0.0574,
+      "step": 10316
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.864529104259854e-05,
+      "loss": 0.0784,
+      "step": 10317
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.862788176245441e-05,
+      "loss": 0.0788,
+      "step": 10318
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.861047353688017e-05,
+      "loss": 0.0549,
+      "step": 10319
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.859306636646102e-05,
+      "loss": 0.048,
+      "step": 10320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.857566025178205e-05,
+      "loss": 0.0624,
+      "step": 10321
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.855825519342844e-05,
+      "loss": 0.0721,
+      "step": 10322
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.854085119198519e-05,
+      "loss": 0.0493,
+      "step": 10323
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.852344824803736e-05,
+      "loss": 0.0862,
+      "step": 10324
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.850604636216995e-05,
+      "loss": 0.0701,
+      "step": 10325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.848864553496794e-05,
+      "loss": 0.0678,
+      "step": 10326
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.847124576701622e-05,
+      "loss": 0.063,
+      "step": 10327
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.845384705889968e-05,
+      "loss": 0.0752,
+      "step": 10328
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.843644941120318e-05,
+      "loss": 0.0606,
+      "step": 10329
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.841905282451158e-05,
+      "loss": 0.0676,
+      "step": 10330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.840165729940961e-05,
+      "loss": 0.0547,
+      "step": 10331
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.838426283648207e-05,
+      "loss": 0.0796,
+      "step": 10332
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.836686943631362e-05,
+      "loss": 0.0717,
+      "step": 10333
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.834947709948898e-05,
+      "loss": 0.073,
+      "step": 10334
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.833208582659276e-05,
+      "loss": 0.0753,
+      "step": 10335
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.831469561820961e-05,
+      "loss": 0.0619,
+      "step": 10336
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.829730647492404e-05,
+      "loss": 0.0607,
+      "step": 10337
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.827991839732064e-05,
+      "loss": 0.0737,
+      "step": 10338
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.82625313859839e-05,
+      "loss": 0.1261,
+      "step": 10339
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.824514544149821e-05,
+      "loss": 0.0545,
+      "step": 10340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.822776056444806e-05,
+      "loss": 0.0749,
+      "step": 10341
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.821037675541786e-05,
+      "loss": 0.0658,
+      "step": 10342
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.819299401499191e-05,
+      "loss": 0.0625,
+      "step": 10343
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.817561234375458e-05,
+      "loss": 0.0774,
+      "step": 10344
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.815823174229009e-05,
+      "loss": 0.0712,
+      "step": 10345
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.814085221118276e-05,
+      "loss": 0.0811,
+      "step": 10346
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.812347375101673e-05,
+      "loss": 0.0673,
+      "step": 10347
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.810609636237624e-05,
+      "loss": 0.0451,
+      "step": 10348
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.808872004584538e-05,
+      "loss": 0.0754,
+      "step": 10349
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.807134480200828e-05,
+      "loss": 0.0855,
+      "step": 10350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.805397063144899e-05,
+      "loss": 0.0655,
+      "step": 10351
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.803659753475159e-05,
+      "loss": 0.0936,
+      "step": 10352
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.80192255125e-05,
+      "loss": 0.0685,
+      "step": 10353
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.80018545652782e-05,
+      "loss": 0.0517,
+      "step": 10354
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.798448469367012e-05,
+      "loss": 0.0594,
+      "step": 10355
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.796711589825967e-05,
+      "loss": 0.0592,
+      "step": 10356
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.794974817963067e-05,
+      "loss": 0.101,
+      "step": 10357
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.793238153836694e-05,
+      "loss": 0.053,
+      "step": 10358
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.791501597505226e-05,
+      "loss": 0.0569,
+      "step": 10359
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.789765149027039e-05,
+      "loss": 0.0596,
+      "step": 10360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.788028808460498e-05,
+      "loss": 0.0491,
+      "step": 10361
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.786292575863977e-05,
+      "loss": 0.1311,
+      "step": 10362
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.784556451295834e-05,
+      "loss": 0.0634,
+      "step": 10363
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.782820434814432e-05,
+      "loss": 0.0671,
+      "step": 10364
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.781084526478128e-05,
+      "loss": 0.056,
+      "step": 10365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.779348726345267e-05,
+      "loss": 0.0752,
+      "step": 10366
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.777613034474205e-05,
+      "loss": 0.0589,
+      "step": 10367
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.775877450923283e-05,
+      "loss": 0.0638,
+      "step": 10368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.774141975750843e-05,
+      "loss": 0.0603,
+      "step": 10369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.772406609015225e-05,
+      "loss": 0.1272,
+      "step": 10370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.770671350774761e-05,
+      "loss": 0.1018,
+      "step": 10371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.768936201087784e-05,
+      "loss": 0.0571,
+      "step": 10372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.767201160012617e-05,
+      "loss": 0.0797,
+      "step": 10373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.765466227607589e-05,
+      "loss": 0.0791,
+      "step": 10374
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.763731403931013e-05,
+      "loss": 0.0882,
+      "step": 10375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.761996689041209e-05,
+      "loss": 0.0576,
+      "step": 10376
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.760262082996486e-05,
+      "loss": 0.1323,
+      "step": 10377
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.758527585855162e-05,
+      "loss": 0.0676,
+      "step": 10378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.756793197675529e-05,
+      "loss": 0.06,
+      "step": 10379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.755058918515892e-05,
+      "loss": 0.0912,
+      "step": 10380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.753324748434554e-05,
+      "loss": 0.0548,
+      "step": 10381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.751590687489802e-05,
+      "loss": 0.0689,
+      "step": 10382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.749856735739928e-05,
+      "loss": 0.0987,
+      "step": 10383
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.748122893243226e-05,
+      "loss": 0.0703,
+      "step": 10384
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.746389160057969e-05,
+      "loss": 0.0668,
+      "step": 10385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.744655536242441e-05,
+      "loss": 0.0783,
+      "step": 10386
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.742922021854916e-05,
+      "loss": 0.0561,
+      "step": 10387
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.741188616953669e-05,
+      "loss": 0.0679,
+      "step": 10388
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.739455321596964e-05,
+      "loss": 0.3946,
+      "step": 10389
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.737722135843069e-05,
+      "loss": 0.0685,
+      "step": 10390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.735989059750245e-05,
+      "loss": 0.0698,
+      "step": 10391
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.734256093376743e-05,
+      "loss": 0.0645,
+      "step": 10392
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.732523236780823e-05,
+      "loss": 0.0627,
+      "step": 10393
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.730790490020731e-05,
+      "loss": 0.0648,
+      "step": 10394
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.729057853154715e-05,
+      "loss": 0.0916,
+      "step": 10395
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.727325326241018e-05,
+      "loss": 0.0778,
+      "step": 10396
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.725592909337876e-05,
+      "loss": 0.1024,
+      "step": 10397
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.723860602503527e-05,
+      "loss": 0.0532,
+      "step": 10398
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.722128405796199e-05,
+      "loss": 0.0538,
+      "step": 10399
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.720396319274125e-05,
+      "loss": 0.0613,
+      "step": 10400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.718664342995522e-05,
+      "loss": 0.0554,
+      "step": 10401
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.716932477018616e-05,
+      "loss": 0.0752,
+      "step": 10402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.715200721401619e-05,
+      "loss": 0.057,
+      "step": 10403
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.713469076202749e-05,
+      "loss": 0.0585,
+      "step": 10404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.711737541480212e-05,
+      "loss": 0.0521,
+      "step": 10405
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.710006117292209e-05,
+      "loss": 0.0616,
+      "step": 10406
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.708274803696948e-05,
+      "loss": 0.0569,
+      "step": 10407
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.706543600752624e-05,
+      "loss": 0.0941,
+      "step": 10408
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.70481250851743e-05,
+      "loss": 0.0502,
+      "step": 10409
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.703081527049562e-05,
+      "loss": 0.0643,
+      "step": 10410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.7013506564072e-05,
+      "loss": 0.0635,
+      "step": 10411
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.699619896648533e-05,
+      "loss": 0.0907,
+      "step": 10412
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.697889247831732e-05,
+      "loss": 0.0843,
+      "step": 10413
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.696158710014984e-05,
+      "loss": 0.0786,
+      "step": 10414
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.69442828325645e-05,
+      "loss": 0.0576,
+      "step": 10415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.692697967614305e-05,
+      "loss": 0.053,
+      "step": 10416
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.690967763146714e-05,
+      "loss": 0.0758,
+      "step": 10417
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.689237669911829e-05,
+      "loss": 0.1176,
+      "step": 10418
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.687507687967817e-05,
+      "loss": 0.1011,
+      "step": 10419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.685777817372824e-05,
+      "loss": 0.1026,
+      "step": 10420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.684048058185004e-05,
+      "loss": 0.0836,
+      "step": 10421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.682318410462498e-05,
+      "loss": 0.0518,
+      "step": 10422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.680588874263451e-05,
+      "loss": 0.1904,
+      "step": 10423
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.678859449646002e-05,
+      "loss": 0.0657,
+      "step": 10424
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.677130136668283e-05,
+      "loss": 0.0564,
+      "step": 10425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.675400935388428e-05,
+      "loss": 0.0597,
+      "step": 10426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.673671845864558e-05,
+      "loss": 0.051,
+      "step": 10427
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.671942868154803e-05,
+      "loss": 0.067,
+      "step": 10428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.670214002317278e-05,
+      "loss": 0.0639,
+      "step": 10429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.668485248410104e-05,
+      "loss": 0.0572,
+      "step": 10430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.666756606491385e-05,
+      "loss": 0.0773,
+      "step": 10431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.665028076619231e-05,
+      "loss": 0.1277,
+      "step": 10432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.66329965885175e-05,
+      "loss": 0.0531,
+      "step": 10433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.66157135324704e-05,
+      "loss": 0.063,
+      "step": 10434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.6598431598632e-05,
+      "loss": 0.0535,
+      "step": 10435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.658115078758318e-05,
+      "loss": 0.0554,
+      "step": 10436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.656387109990488e-05,
+      "loss": 0.0653,
+      "step": 10437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.654659253617794e-05,
+      "loss": 0.0493,
+      "step": 10438
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.652931509698319e-05,
+      "loss": 0.0707,
+      "step": 10439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.651203878290139e-05,
+      "loss": 0.0479,
+      "step": 10440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.649476359451328e-05,
+      "loss": 0.0563,
+      "step": 10441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.647748953239961e-05,
+      "loss": 0.0653,
+      "step": 10442
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.646021659714098e-05,
+      "loss": 0.0573,
+      "step": 10443
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.644294478931803e-05,
+      "loss": 0.048,
+      "step": 10444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.642567410951138e-05,
+      "loss": 0.076,
+      "step": 10445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.640840455830155e-05,
+      "loss": 0.0791,
+      "step": 10446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.639113613626908e-05,
+      "loss": 0.0565,
+      "step": 10447
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.637386884399443e-05,
+      "loss": 0.1115,
+      "step": 10448
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.635660268205803e-05,
+      "loss": 0.0526,
+      "step": 10449
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.633933765104032e-05,
+      "loss": 0.0571,
+      "step": 10450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.632207375152162e-05,
+      "loss": 0.0698,
+      "step": 10451
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.630481098408228e-05,
+      "loss": 0.0678,
+      "step": 10452
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.628754934930256e-05,
+      "loss": 0.0905,
+      "step": 10453
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.627028884776274e-05,
+      "loss": 0.0684,
+      "step": 10454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.625302948004303e-05,
+      "loss": 0.1521,
+      "step": 10455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.623577124672354e-05,
+      "loss": 0.0866,
+      "step": 10456
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.621851414838448e-05,
+      "loss": 0.0648,
+      "step": 10457
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.620125818560587e-05,
+      "loss": 0.0737,
+      "step": 10458
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.618400335896786e-05,
+      "loss": 0.063,
+      "step": 10459
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.616674966905039e-05,
+      "loss": 0.0671,
+      "step": 10460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.614949711643347e-05,
+      "loss": 0.06,
+      "step": 10461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.613224570169702e-05,
+      "loss": 0.0875,
+      "step": 10462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.611499542542099e-05,
+      "loss": 0.0908,
+      "step": 10463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.609774628818524e-05,
+      "loss": 0.0645,
+      "step": 10464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.608049829056954e-05,
+      "loss": 0.0477,
+      "step": 10465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.606325143315375e-05,
+      "loss": 0.0819,
+      "step": 10466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.604600571651756e-05,
+      "loss": 0.0569,
+      "step": 10467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.602876114124076e-05,
+      "loss": 0.0659,
+      "step": 10468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.601151770790294e-05,
+      "loss": 0.0554,
+      "step": 10469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.599427541708378e-05,
+      "loss": 0.1126,
+      "step": 10470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.597703426936288e-05,
+      "loss": 0.0597,
+      "step": 10471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.595979426531976e-05,
+      "loss": 0.0503,
+      "step": 10472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.5942555405534e-05,
+      "loss": 0.0469,
+      "step": 10473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.592531769058502e-05,
+      "loss": 0.0662,
+      "step": 10474
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.590808112105232e-05,
+      "loss": 0.0591,
+      "step": 10475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.589084569751527e-05,
+      "loss": 0.063,
+      "step": 10476
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.587361142055322e-05,
+      "loss": 0.1495,
+      "step": 10477
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.585637829074559e-05,
+      "loss": 0.0556,
+      "step": 10478
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.583914630867154e-05,
+      "loss": 0.0635,
+      "step": 10479
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.582191547491042e-05,
+      "loss": 0.1022,
+      "step": 10480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.580468579004144e-05,
+      "loss": 0.0685,
+      "step": 10481
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.578745725464368e-05,
+      "loss": 0.076,
+      "step": 10482
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.577022986929637e-05,
+      "loss": 0.0618,
+      "step": 10483
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.575300363457855e-05,
+      "loss": 0.144,
+      "step": 10484
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.573577855106932e-05,
+      "loss": 0.0521,
+      "step": 10485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.571855461934768e-05,
+      "loss": 0.0941,
+      "step": 10486
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.57013318399926e-05,
+      "loss": 0.0767,
+      "step": 10487
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.568411021358303e-05,
+      "loss": 0.094,
+      "step": 10488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.566688974069788e-05,
+      "loss": 0.1025,
+      "step": 10489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.564967042191602e-05,
+      "loss": 0.0649,
+      "step": 10490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.563245225781622e-05,
+      "loss": 0.0485,
+      "step": 10491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.561523524897736e-05,
+      "loss": 0.0533,
+      "step": 10492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.559801939597811e-05,
+      "loss": 0.0734,
+      "step": 10493
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.558080469939725e-05,
+      "loss": 0.0578,
+      "step": 10494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.556359115981338e-05,
+      "loss": 0.105,
+      "step": 10495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.554637877780515e-05,
+      "loss": 0.0525,
+      "step": 10496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.552916755395117e-05,
+      "loss": 0.0586,
+      "step": 10497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.551195748882997e-05,
+      "loss": 0.1387,
+      "step": 10498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.54947485830201e-05,
+      "loss": 0.0617,
+      "step": 10499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.547754083709998e-05,
+      "loss": 0.0502,
+      "step": 10500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.54603342516481e-05,
+      "loss": 0.0994,
+      "step": 10501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.544312882724284e-05,
+      "loss": 0.0683,
+      "step": 10502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.542592456446252e-05,
+      "loss": 0.0603,
+      "step": 10503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.540872146388554e-05,
+      "loss": 0.0708,
+      "step": 10504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.53915195260901e-05,
+      "loss": 0.0913,
+      "step": 10505
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.53743187516545e-05,
+      "loss": 0.0486,
+      "step": 10506
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.535711914115695e-05,
+      "loss": 0.0547,
+      "step": 10507
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.533992069517551e-05,
+      "loss": 0.0686,
+      "step": 10508
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.53227234142884e-05,
+      "loss": 0.0573,
+      "step": 10509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.530552729907367e-05,
+      "loss": 0.0903,
+      "step": 10510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.528833235010939e-05,
+      "loss": 0.0704,
+      "step": 10511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.527113856797353e-05,
+      "loss": 0.0569,
+      "step": 10512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.525394595324409e-05,
+      "loss": 0.0546,
+      "step": 10513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.523675450649897e-05,
+      "loss": 0.0505,
+      "step": 10514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.521956422831609e-05,
+      "loss": 0.045,
+      "step": 10515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.520237511927325e-05,
+      "loss": 0.0679,
+      "step": 10516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.51851871799483e-05,
+      "loss": 0.0602,
+      "step": 10517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.516800041091903e-05,
+      "loss": 0.0748,
+      "step": 10518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.51508148127631e-05,
+      "loss": 0.0562,
+      "step": 10519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.513363038605831e-05,
+      "loss": 0.0617,
+      "step": 10520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.51164471313822e-05,
+      "loss": 0.1284,
+      "step": 10521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.509926504931243e-05,
+      "loss": 0.0514,
+      "step": 10522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.508208414042658e-05,
+      "loss": 0.0798,
+      "step": 10523
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.506490440530217e-05,
+      "loss": 0.0629,
+      "step": 10524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.504772584451672e-05,
+      "loss": 0.0694,
+      "step": 10525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.503054845864764e-05,
+      "loss": 0.0527,
+      "step": 10526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.501337224827239e-05,
+      "loss": 0.0526,
+      "step": 10527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.499619721396833e-05,
+      "loss": 0.0574,
+      "step": 10528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.497902335631279e-05,
+      "loss": 0.0532,
+      "step": 10529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.496185067588308e-05,
+      "loss": 0.0729,
+      "step": 10530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.494467917325642e-05,
+      "loss": 0.052,
+      "step": 10531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.492750884901008e-05,
+      "loss": 0.0637,
+      "step": 10532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.491033970372126e-05,
+      "loss": 0.061,
+      "step": 10533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.4893171737967e-05,
+      "loss": 0.0708,
+      "step": 10534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.487600495232449e-05,
+      "loss": 0.0607,
+      "step": 10535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.485883934737071e-05,
+      "loss": 0.0749,
+      "step": 10536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.484167492368276e-05,
+      "loss": 0.1031,
+      "step": 10537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.482451168183755e-05,
+      "loss": 0.0586,
+      "step": 10538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.480734962241208e-05,
+      "loss": 0.0662,
+      "step": 10539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.479018874598322e-05,
+      "loss": 0.0579,
+      "step": 10540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.477302905312784e-05,
+      "loss": 0.0749,
+      "step": 10541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.475587054442273e-05,
+      "loss": 0.1115,
+      "step": 10542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.473871322044473e-05,
+      "loss": 0.0579,
+      "step": 10543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.472155708177052e-05,
+      "loss": 0.0602,
+      "step": 10544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.470440212897683e-05,
+      "loss": 0.0813,
+      "step": 10545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.468724836264036e-05,
+      "loss": 0.0544,
+      "step": 10546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.467009578333768e-05,
+      "loss": 0.0646,
+      "step": 10547
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.465294439164536e-05,
+      "loss": 0.0966,
+      "step": 10548
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.463579418813998e-05,
+      "loss": 0.1024,
+      "step": 10549
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.461864517339803e-05,
+      "loss": 0.0712,
+      "step": 10550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.460149734799598e-05,
+      "loss": 0.0584,
+      "step": 10551
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.458435071251022e-05,
+      "loss": 0.0785,
+      "step": 10552
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.456720526751718e-05,
+      "loss": 0.0782,
+      "step": 10553
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.455006101359315e-05,
+      "loss": 0.0518,
+      "step": 10554
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.453291795131449e-05,
+      "loss": 0.0503,
+      "step": 10555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.45157760812574e-05,
+      "loss": 0.0686,
+      "step": 10556
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.449863540399814e-05,
+      "loss": 0.05,
+      "step": 10557
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.44814959201129e-05,
+      "loss": 0.0591,
+      "step": 10558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.446435763017784e-05,
+      "loss": 0.1516,
+      "step": 10559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.444722053476897e-05,
+      "loss": 0.0661,
+      "step": 10560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.443008463446243e-05,
+      "loss": 0.075,
+      "step": 10561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.441294992983422e-05,
+      "loss": 0.0931,
+      "step": 10562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.439581642146033e-05,
+      "loss": 0.0734,
+      "step": 10563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.437868410991665e-05,
+      "loss": 0.0702,
+      "step": 10564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.436155299577918e-05,
+      "loss": 0.0583,
+      "step": 10565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.434442307962367e-05,
+      "loss": 0.0867,
+      "step": 10566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.432729436202604e-05,
+      "loss": 0.0543,
+      "step": 10567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.4310166843562e-05,
+      "loss": 0.0521,
+      "step": 10568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.429304052480731e-05,
+      "loss": 0.0645,
+      "step": 10569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.427591540633766e-05,
+      "loss": 0.0895,
+      "step": 10570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.425879148872877e-05,
+      "loss": 0.0634,
+      "step": 10571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.42416687725562e-05,
+      "loss": 0.0625,
+      "step": 10572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.422454725839549e-05,
+      "loss": 0.0605,
+      "step": 10573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.420742694682222e-05,
+      "loss": 0.0933,
+      "step": 10574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.419030783841193e-05,
+      "loss": 0.0617,
+      "step": 10575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.417318993374002e-05,
+      "loss": 0.0504,
+      "step": 10576
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.415607323338192e-05,
+      "loss": 0.1276,
+      "step": 10577
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.413895773791303e-05,
+      "loss": 0.0665,
+      "step": 10578
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.412184344790866e-05,
+      "loss": 0.0865,
+      "step": 10579
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.410473036394408e-05,
+      "loss": 0.069,
+      "step": 10580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.408761848659462e-05,
+      "loss": 0.1286,
+      "step": 10581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.407050781643542e-05,
+      "loss": 0.0515,
+      "step": 10582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.405339835404168e-05,
+      "loss": 0.0747,
+      "step": 10583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.403629009998857e-05,
+      "loss": 0.0719,
+      "step": 10584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.40191830548511e-05,
+      "loss": 0.0616,
+      "step": 10585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.400207721920437e-05,
+      "loss": 0.06,
+      "step": 10586
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.398497259362341e-05,
+      "loss": 0.0522,
+      "step": 10587
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.396786917868315e-05,
+      "loss": 0.0553,
+      "step": 10588
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.395076697495854e-05,
+      "loss": 0.0836,
+      "step": 10589
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.393366598302446e-05,
+      "loss": 0.1723,
+      "step": 10590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.391656620345579e-05,
+      "loss": 0.1239,
+      "step": 10591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.389946763682726e-05,
+      "loss": 0.0618,
+      "step": 10592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.388237028371373e-05,
+      "loss": 0.0821,
+      "step": 10593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.386527414468984e-05,
+      "loss": 0.0538,
+      "step": 10594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.384817922033035e-05,
+      "loss": 0.0571,
+      "step": 10595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.383108551120985e-05,
+      "loss": 0.0581,
+      "step": 10596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.381399301790301e-05,
+      "loss": 0.0502,
+      "step": 10597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.379690174098432e-05,
+      "loss": 0.0566,
+      "step": 10598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.377981168102828e-05,
+      "loss": 0.0536,
+      "step": 10599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.376272283860945e-05,
+      "loss": 0.0494,
+      "step": 10600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.374563521430224e-05,
+      "loss": 0.0541,
+      "step": 10601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.372854880868102e-05,
+      "loss": 0.0538,
+      "step": 10602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.37114636223202e-05,
+      "loss": 0.0645,
+      "step": 10603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.369437965579403e-05,
+      "loss": 0.0705,
+      "step": 10604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.367729690967685e-05,
+      "loss": 0.0598,
+      "step": 10605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.366021538454285e-05,
+      "loss": 0.0697,
+      "step": 10606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.364313508096627e-05,
+      "loss": 0.0492,
+      "step": 10607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.36260559995212e-05,
+      "loss": 0.0728,
+      "step": 10608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.360897814078182e-05,
+      "loss": 0.0969,
+      "step": 10609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.359190150532217e-05,
+      "loss": 0.0773,
+      "step": 10610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.357482609371623e-05,
+      "loss": 0.0501,
+      "step": 10611
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.355775190653806e-05,
+      "loss": 0.071,
+      "step": 10612
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.354067894436155e-05,
+      "loss": 0.0571,
+      "step": 10613
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.352360720776064e-05,
+      "loss": 0.0623,
+      "step": 10614
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.350653669730922e-05,
+      "loss": 0.0663,
+      "step": 10615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.348946741358104e-05,
+      "loss": 0.0527,
+      "step": 10616
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.347239935714995e-05,
+      "loss": 0.1341,
+      "step": 10617
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.345533252858963e-05,
+      "loss": 0.0729,
+      "step": 10618
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.343826692847386e-05,
+      "loss": 0.0806,
+      "step": 10619
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.342120255737623e-05,
+      "loss": 0.0782,
+      "step": 10620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.340413941587037e-05,
+      "loss": 0.0581,
+      "step": 10621
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.338707750452986e-05,
+      "loss": 0.0611,
+      "step": 10622
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.337001682392826e-05,
+      "loss": 0.1095,
+      "step": 10623
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.335295737463904e-05,
+      "loss": 0.0878,
+      "step": 10624
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.333589915723562e-05,
+      "loss": 0.0565,
+      "step": 10625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.331884217229146e-05,
+      "loss": 0.0634,
+      "step": 10626
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.330178642037989e-05,
+      "loss": 0.0491,
+      "step": 10627
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.328473190207427e-05,
+      "loss": 0.0667,
+      "step": 10628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.326767861794788e-05,
+      "loss": 0.0766,
+      "step": 10629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.325062656857394e-05,
+      "loss": 0.0572,
+      "step": 10630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.323357575452569e-05,
+      "loss": 0.0689,
+      "step": 10631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.321652617637623e-05,
+      "loss": 0.0514,
+      "step": 10632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.319947783469875e-05,
+      "loss": 0.0615,
+      "step": 10633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.318243073006628e-05,
+      "loss": 0.0921,
+      "step": 10634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.31653848630519e-05,
+      "loss": 0.0655,
+      "step": 10635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.314834023422858e-05,
+      "loss": 0.1332,
+      "step": 10636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.313129684416926e-05,
+      "loss": 0.1606,
+      "step": 10637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.311425469344685e-05,
+      "loss": 0.0942,
+      "step": 10638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.309721378263422e-05,
+      "loss": 0.1162,
+      "step": 10639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.308017411230423e-05,
+      "loss": 0.1024,
+      "step": 10640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.306313568302963e-05,
+      "loss": 0.0652,
+      "step": 10641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.304609849538319e-05,
+      "loss": 0.0573,
+      "step": 10642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.302906254993761e-05,
+      "loss": 0.0513,
+      "step": 10643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.301202784726555e-05,
+      "loss": 0.0742,
+      "step": 10644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.299499438793961e-05,
+      "loss": 0.0561,
+      "step": 10645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.297796217253238e-05,
+      "loss": 0.0576,
+      "step": 10646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.296093120161642e-05,
+      "loss": 0.0618,
+      "step": 10647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.294390147576419e-05,
+      "loss": 0.1055,
+      "step": 10648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.292687299554818e-05,
+      "loss": 0.064,
+      "step": 10649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.290984576154075e-05,
+      "loss": 0.2045,
+      "step": 10650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.289281977431429e-05,
+      "loss": 0.0599,
+      "step": 10651
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.287579503444114e-05,
+      "loss": 0.0549,
+      "step": 10652
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.285877154249355e-05,
+      "loss": 0.0657,
+      "step": 10653
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.284174929904381e-05,
+      "loss": 0.05,
+      "step": 10654
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.282472830466409e-05,
+      "loss": 0.0613,
+      "step": 10655
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.280770855992655e-05,
+      "loss": 0.0748,
+      "step": 10656
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.279069006540332e-05,
+      "loss": 0.0751,
+      "step": 10657
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.277367282166647e-05,
+      "loss": 0.0554,
+      "step": 10658
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.275665682928803e-05,
+      "loss": 0.0534,
+      "step": 10659
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.273964208883998e-05,
+      "loss": 0.0816,
+      "step": 10660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.272262860089428e-05,
+      "loss": 0.0585,
+      "step": 10661
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.270561636602288e-05,
+      "loss": 0.0501,
+      "step": 10662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.268860538479756e-05,
+      "loss": 0.066,
+      "step": 10663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.267159565779017e-05,
+      "loss": 0.061,
+      "step": 10664
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.265458718557252e-05,
+      "loss": 0.0582,
+      "step": 10665
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.263757996871633e-05,
+      "loss": 0.0567,
+      "step": 10666
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.262057400779326e-05,
+      "loss": 0.1253,
+      "step": 10667
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.260356930337501e-05,
+      "loss": 0.0672,
+      "step": 10668
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.258656585603321e-05,
+      "loss": 0.049,
+      "step": 10669
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.256956366633937e-05,
+      "loss": 0.0495,
+      "step": 10670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.255256273486505e-05,
+      "loss": 0.163,
+      "step": 10671
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.253556306218173e-05,
+      "loss": 0.0604,
+      "step": 10672
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.251856464886086e-05,
+      "loss": 0.0709,
+      "step": 10673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.25015674954738e-05,
+      "loss": 0.0519,
+      "step": 10674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.2484571602592e-05,
+      "loss": 0.0804,
+      "step": 10675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.246757697078667e-05,
+      "loss": 0.0797,
+      "step": 10676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.245058360062911e-05,
+      "loss": 0.0753,
+      "step": 10677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.243359149269059e-05,
+      "loss": 0.0811,
+      "step": 10678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.241660064754225e-05,
+      "loss": 0.0729,
+      "step": 10679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.239961106575527e-05,
+      "loss": 0.0491,
+      "step": 10680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.238262274790072e-05,
+      "loss": 0.0641,
+      "step": 10681
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.23656356945497e-05,
+      "loss": 0.0716,
+      "step": 10682
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.23486499062732e-05,
+      "loss": 0.0512,
+      "step": 10683
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.23316653836422e-05,
+      "loss": 0.0488,
+      "step": 10684
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.231468212722764e-05,
+      "loss": 0.0775,
+      "step": 10685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.22977001376004e-05,
+      "loss": 0.0618,
+      "step": 10686
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.228071941533135e-05,
+      "loss": 0.0564,
+      "step": 10687
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.22637399609913e-05,
+      "loss": 0.0528,
+      "step": 10688
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.224676177515095e-05,
+      "loss": 0.0901,
+      "step": 10689
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.222978485838106e-05,
+      "loss": 0.056,
+      "step": 10690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.221280921125232e-05,
+      "loss": 0.0585,
+      "step": 10691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.219583483433535e-05,
+      "loss": 0.0537,
+      "step": 10692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.217886172820073e-05,
+      "loss": 0.078,
+      "step": 10693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.216188989341901e-05,
+      "loss": 0.0546,
+      "step": 10694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.214491933056074e-05,
+      "loss": 0.0649,
+      "step": 10695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.212795004019633e-05,
+      "loss": 0.053,
+      "step": 10696
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.211098202289624e-05,
+      "loss": 0.068,
+      "step": 10697
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.20940152792308e-05,
+      "loss": 0.0665,
+      "step": 10698
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.20770498097704e-05,
+      "loss": 0.073,
+      "step": 10699
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.206008561508529e-05,
+      "loss": 0.0551,
+      "step": 10700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.204312269574578e-05,
+      "loss": 0.0599,
+      "step": 10701
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.202616105232201e-05,
+      "loss": 0.0589,
+      "step": 10702
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.200920068538415e-05,
+      "loss": 0.0618,
+      "step": 10703
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.199224159550234e-05,
+      "loss": 0.0495,
+      "step": 10704
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.197528378324665e-05,
+      "loss": 0.0615,
+      "step": 10705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.195832724918713e-05,
+      "loss": 0.0723,
+      "step": 10706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.194137199389373e-05,
+      "loss": 0.0552,
+      "step": 10707
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.192441801793645e-05,
+      "loss": 0.0644,
+      "step": 10708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.19074653218852e-05,
+      "loss": 0.0788,
+      "step": 10709
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.189051390630979e-05,
+      "loss": 0.0664,
+      "step": 10710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.187356377178008e-05,
+      "loss": 0.058,
+      "step": 10711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.185661491886583e-05,
+      "loss": 0.0686,
+      "step": 10712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.18396673481368e-05,
+      "loss": 0.0542,
+      "step": 10713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.182272106016265e-05,
+      "loss": 0.0706,
+      "step": 10714
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.1805776055513e-05,
+      "loss": 0.0771,
+      "step": 10715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.178883233475754e-05,
+      "loss": 0.0615,
+      "step": 10716
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.177188989846575e-05,
+      "loss": 0.0545,
+      "step": 10717
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.17549487472072e-05,
+      "loss": 0.052,
+      "step": 10718
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.173800888155131e-05,
+      "loss": 0.0534,
+      "step": 10719
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.172107030206758e-05,
+      "loss": 0.0599,
+      "step": 10720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.170413300932533e-05,
+      "loss": 0.0689,
+      "step": 10721
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.168719700389396e-05,
+      "loss": 0.0618,
+      "step": 10722
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.167026228634276e-05,
+      "loss": 0.0506,
+      "step": 10723
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.165332885724096e-05,
+      "loss": 0.0919,
+      "step": 10724
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.163639671715781e-05,
+      "loss": 0.0945,
+      "step": 10725
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.16194658666625e-05,
+      "loss": 0.0697,
+      "step": 10726
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.160253630632407e-05,
+      "loss": 0.0496,
+      "step": 10727
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.158560803671168e-05,
+      "loss": 0.1189,
+      "step": 10728
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.156868105839432e-05,
+      "loss": 0.0579,
+      "step": 10729
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.155175537194107e-05,
+      "loss": 0.0598,
+      "step": 10730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.15348309779208e-05,
+      "loss": 0.0603,
+      "step": 10731
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.151790787690248e-05,
+      "loss": 0.0887,
+      "step": 10732
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.150098606945492e-05,
+      "loss": 0.0542,
+      "step": 10733
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.148406555614701e-05,
+      "loss": 0.1032,
+      "step": 10734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.146714633754748e-05,
+      "loss": 0.0591,
+      "step": 10735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.145022841422508e-05,
+      "loss": 0.0685,
+      "step": 10736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.143331178674854e-05,
+      "loss": 0.2515,
+      "step": 10737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.141639645568646e-05,
+      "loss": 0.0569,
+      "step": 10738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.139948242160749e-05,
+      "loss": 0.0756,
+      "step": 10739
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.138256968508017e-05,
+      "loss": 0.0656,
+      "step": 10740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.1365658246673e-05,
+      "loss": 0.0629,
+      "step": 10741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.134874810695447e-05,
+      "loss": 0.0569,
+      "step": 10742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.133183926649304e-05,
+      "loss": 0.0613,
+      "step": 10743
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.131493172585707e-05,
+      "loss": 0.0511,
+      "step": 10744
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.12980254856149e-05,
+      "loss": 0.0672,
+      "step": 10745
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.128112054633488e-05,
+      "loss": 0.0605,
+      "step": 10746
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.126421690858518e-05,
+      "loss": 0.0616,
+      "step": 10747
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.124731457293409e-05,
+      "loss": 0.0576,
+      "step": 10748
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.123041353994976e-05,
+      "loss": 0.0665,
+      "step": 10749
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.121351381020029e-05,
+      "loss": 0.0525,
+      "step": 10750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.119661538425381e-05,
+      "loss": 0.1294,
+      "step": 10751
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.117971826267834e-05,
+      "loss": 0.0555,
+      "step": 10752
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.116282244604183e-05,
+      "loss": 0.0772,
+      "step": 10753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.114592793491231e-05,
+      "loss": 0.1204,
+      "step": 10754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.11290347298576e-05,
+      "loss": 0.0595,
+      "step": 10755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.111214283144564e-05,
+      "loss": 0.0588,
+      "step": 10756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.109525224024418e-05,
+      "loss": 0.0572,
+      "step": 10757
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.107836295682105e-05,
+      "loss": 0.0732,
+      "step": 10758
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.106147498174398e-05,
+      "loss": 0.1374,
+      "step": 10759
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.104458831558063e-05,
+      "loss": 0.0582,
+      "step": 10760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.102770295889863e-05,
+      "loss": 0.0921,
+      "step": 10761
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.10108189122656e-05,
+      "loss": 0.0604,
+      "step": 10762
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.099393617624912e-05,
+      "loss": 0.1222,
+      "step": 10763
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.097705475141666e-05,
+      "loss": 0.053,
+      "step": 10764
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0960174638335754e-05,
+      "loss": 0.1144,
+      "step": 10765
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.094329583757375e-05,
+      "loss": 0.0533,
+      "step": 10766
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0926418349698036e-05,
+      "loss": 0.0572,
+      "step": 10767
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.090954217527598e-05,
+      "loss": 0.0753,
+      "step": 10768
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.089266731487484e-05,
+      "loss": 0.0545,
+      "step": 10769
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0875793769061896e-05,
+      "loss": 0.0681,
+      "step": 10770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0858921538404314e-05,
+      "loss": 0.0695,
+      "step": 10771
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.084205062346928e-05,
+      "loss": 0.0802,
+      "step": 10772
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.082518102482388e-05,
+      "loss": 0.0607,
+      "step": 10773
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0808312743035236e-05,
+      "loss": 0.0662,
+      "step": 10774
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0791445778670306e-05,
+      "loss": 0.047,
+      "step": 10775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.077458013229611e-05,
+      "loss": 0.0585,
+      "step": 10776
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0757715804479595e-05,
+      "loss": 0.0592,
+      "step": 10777
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.074085279578767e-05,
+      "loss": 0.0546,
+      "step": 10778
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.07239911067871e-05,
+      "loss": 0.0654,
+      "step": 10779
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0707130738044746e-05,
+      "loss": 0.0523,
+      "step": 10780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0690271690127356e-05,
+      "loss": 0.0673,
+      "step": 10781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0673413963601665e-05,
+      "loss": 0.0625,
+      "step": 10782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.065655755903431e-05,
+      "loss": 0.0527,
+      "step": 10783
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0639702476991954e-05,
+      "loss": 0.0625,
+      "step": 10784
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0622848718041135e-05,
+      "loss": 0.0606,
+      "step": 10785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.060599628274843e-05,
+      "loss": 0.0492,
+      "step": 10786
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.0589145171680306e-05,
+      "loss": 0.0615,
+      "step": 10787
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.057229538540323e-05,
+      "loss": 0.0556,
+      "step": 10788
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0555446924483586e-05,
+      "loss": 0.0712,
+      "step": 10789
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0538599789487725e-05,
+      "loss": 0.048,
+      "step": 10790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.052175398098203e-05,
+      "loss": 0.0566,
+      "step": 10791
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.050490949953269e-05,
+      "loss": 0.0835,
+      "step": 10792
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.048806634570596e-05,
+      "loss": 0.0483,
+      "step": 10793
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.047122452006803e-05,
+      "loss": 0.069,
+      "step": 10794
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0454384023185e-05,
+      "loss": 0.0932,
+      "step": 10795
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0437544855623005e-05,
+      "loss": 0.0617,
+      "step": 10796
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.042070701794806e-05,
+      "loss": 0.0893,
+      "step": 10797
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.040387051072621e-05,
+      "loss": 0.1456,
+      "step": 10798
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0387035334523356e-05,
+      "loss": 0.063,
+      "step": 10799
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0370201489905455e-05,
+      "loss": 0.0659,
+      "step": 10800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.035336897743834e-05,
+      "loss": 0.0674,
+      "step": 10801
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.033653779768784e-05,
+      "loss": 0.0596,
+      "step": 10802
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0319707951219786e-05,
+      "loss": 0.0545,
+      "step": 10803
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.030287943859987e-05,
+      "loss": 0.0603,
+      "step": 10804
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0286052260393745e-05,
+      "loss": 0.0711,
+      "step": 10805
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0269226417167115e-05,
+      "loss": 0.0647,
+      "step": 10806
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.025240190948552e-05,
+      "loss": 0.0641,
+      "step": 10807
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0235578737914575e-05,
+      "loss": 0.0722,
+      "step": 10808
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0218756903019745e-05,
+      "loss": 0.0775,
+      "step": 10809
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.020193640536652e-05,
+      "loss": 0.0557,
+      "step": 10810
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.018511724552029e-05,
+      "loss": 0.0783,
+      "step": 10811
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.016829942404647e-05,
+      "loss": 0.0885,
+      "step": 10812
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.015148294151034e-05,
+      "loss": 0.0573,
+      "step": 10813
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0134667798477226e-05,
+      "loss": 0.0772,
+      "step": 10814
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0117853995512355e-05,
+      "loss": 0.0713,
+      "step": 10815
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.01010415331809e-05,
+      "loss": 0.0641,
+      "step": 10816
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.008423041204809e-05,
+      "loss": 0.054,
+      "step": 10817
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0067420632678895e-05,
+      "loss": 0.1091,
+      "step": 10818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.005061219563846e-05,
+      "loss": 0.0623,
+      "step": 10819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.0033805101491794e-05,
+      "loss": 0.0815,
+      "step": 10820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.001699935080383e-05,
+      "loss": 0.0438,
+      "step": 10821
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.000019494413954e-05,
+      "loss": 0.074,
+      "step": 10822
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9983391882063764e-05,
+      "loss": 0.0773,
+      "step": 10823
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.996659016514137e-05,
+      "loss": 0.0574,
+      "step": 10824
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.994978979393708e-05,
+      "loss": 0.0657,
+      "step": 10825
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.993299076901573e-05,
+      "loss": 0.056,
+      "step": 10826
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.991619309094194e-05,
+      "loss": 0.0565,
+      "step": 10827
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.98993967602804e-05,
+      "loss": 0.0758,
+      "step": 10828
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9882601777595684e-05,
+      "loss": 0.0724,
+      "step": 10829
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.986580814345243e-05,
+      "loss": 0.069,
+      "step": 10830
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.984901585841505e-05,
+      "loss": 0.0605,
+      "step": 10831
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.983222492304809e-05,
+      "loss": 0.0654,
+      "step": 10832
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.981543533791591e-05,
+      "loss": 0.0584,
+      "step": 10833
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.979864710358297e-05,
+      "loss": 0.0619,
+      "step": 10834
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.978186022061354e-05,
+      "loss": 0.0626,
+      "step": 10835
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.976507468957195e-05,
+      "loss": 0.0536,
+      "step": 10836
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.9748290511022395e-05,
+      "loss": 0.0884,
+      "step": 10837
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.973150768552914e-05,
+      "loss": 0.0545,
+      "step": 10838
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.971472621365628e-05,
+      "loss": 0.063,
+      "step": 10839
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.969794609596795e-05,
+      "loss": 0.0632,
+      "step": 10840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.96811673330282e-05,
+      "loss": 0.0812,
+      "step": 10841
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.966438992540109e-05,
+      "loss": 0.0545,
+      "step": 10842
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.964761387365052e-05,
+      "loss": 0.0677,
+      "step": 10843
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.963083917834045e-05,
+      "loss": 0.0615,
+      "step": 10844
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.961406584003473e-05,
+      "loss": 0.063,
+      "step": 10845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.959729385929725e-05,
+      "loss": 0.0689,
+      "step": 10846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9580523236691756e-05,
+      "loss": 0.0602,
+      "step": 10847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.956375397278201e-05,
+      "loss": 0.072,
+      "step": 10848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.954698606813169e-05,
+      "loss": 0.06,
+      "step": 10849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9530219523304476e-05,
+      "loss": 0.0548,
+      "step": 10850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9513454338863926e-05,
+      "loss": 0.0542,
+      "step": 10851
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9496690515373655e-05,
+      "loss": 0.0663,
+      "step": 10852
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9479928053397136e-05,
+      "loss": 0.0541,
+      "step": 10853
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.946316695349787e-05,
+      "loss": 0.1118,
+      "step": 10854
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9446407216239285e-05,
+      "loss": 0.0553,
+      "step": 10855
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9429648842184696e-05,
+      "loss": 0.0571,
+      "step": 10856
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9412891831897475e-05,
+      "loss": 0.0499,
+      "step": 10857
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.93961361859409e-05,
+      "loss": 0.0468,
+      "step": 10858
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.937938190487819e-05,
+      "loss": 0.0695,
+      "step": 10859
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9362628989272605e-05,
+      "loss": 0.0549,
+      "step": 10860
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.934587743968723e-05,
+      "loss": 0.0548,
+      "step": 10861
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9329127256685205e-05,
+      "loss": 0.0617,
+      "step": 10862
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.931237844082953e-05,
+      "loss": 0.0668,
+      "step": 10863
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.929563099268327e-05,
+      "loss": 0.0727,
+      "step": 10864
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9278884912809354e-05,
+      "loss": 0.06,
+      "step": 10865
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.926214020177074e-05,
+      "loss": 0.0517,
+      "step": 10866
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.924539686013024e-05,
+      "loss": 0.0779,
+      "step": 10867
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.922865488845076e-05,
+      "loss": 0.0664,
+      "step": 10868
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.921191428729502e-05,
+      "loss": 0.0515,
+      "step": 10869
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.919517505722574e-05,
+      "loss": 0.0652,
+      "step": 10870
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9178437198805636e-05,
+      "loss": 0.1005,
+      "step": 10871
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9161700712597346e-05,
+      "loss": 0.0835,
+      "step": 10872
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.914496559916345e-05,
+      "loss": 0.0542,
+      "step": 10873
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9128231859066526e-05,
+      "loss": 0.0544,
+      "step": 10874
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.911149949286906e-05,
+      "loss": 0.0648,
+      "step": 10875
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.909476850113351e-05,
+      "loss": 0.0838,
+      "step": 10876
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.907803888442229e-05,
+      "loss": 0.0881,
+      "step": 10877
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.906131064329776e-05,
+      "loss": 0.0496,
+      "step": 10878
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.9044583778322225e-05,
+      "loss": 0.0543,
+      "step": 10879
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.902785829005799e-05,
+      "loss": 0.0626,
+      "step": 10880
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.901113417906727e-05,
+      "loss": 0.0509,
+      "step": 10881
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.899441144591221e-05,
+      "loss": 0.061,
+      "step": 10882
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.897769009115498e-05,
+      "loss": 0.0992,
+      "step": 10883
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.896097011535764e-05,
+      "loss": 0.0598,
+      "step": 10884
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.894425151908223e-05,
+      "loss": 0.0599,
+      "step": 10885
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.892753430289078e-05,
+      "loss": 0.0583,
+      "step": 10886
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8910818467345185e-05,
+      "loss": 0.0541,
+      "step": 10887
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8894104013007404e-05,
+      "loss": 0.054,
+      "step": 10888
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.887739094043923e-05,
+      "loss": 0.0601,
+      "step": 10889
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8860679250202534e-05,
+      "loss": 0.1233,
+      "step": 10890
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.884396894285902e-05,
+      "loss": 0.0629,
+      "step": 10891
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.882726001897045e-05,
+      "loss": 0.0492,
+      "step": 10892
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8810552479098455e-05,
+      "loss": 0.0588,
+      "step": 10893
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.87938463238047e-05,
+      "loss": 0.0477,
+      "step": 10894
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.877714155365073e-05,
+      "loss": 0.051,
+      "step": 10895
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.876043816919804e-05,
+      "loss": 0.049,
+      "step": 10896
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8743736171008176e-05,
+      "loss": 0.0576,
+      "step": 10897
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8727035559642526e-05,
+      "loss": 0.0539,
+      "step": 10898
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.87103363356625e-05,
+      "loss": 0.0893,
+      "step": 10899
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8693638499629466e-05,
+      "loss": 0.0503,
+      "step": 10900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.8676942052104665e-05,
+      "loss": 0.0746,
+      "step": 10901
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.86602469936494e-05,
+      "loss": 0.0654,
+      "step": 10902
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.864355332482483e-05,
+      "loss": 0.0669,
+      "step": 10903
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.862686104619215e-05,
+      "loss": 0.0671,
+      "step": 10904
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.861017015831242e-05,
+      "loss": 0.0662,
+      "step": 10905
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8593480661746746e-05,
+      "loss": 0.0536,
+      "step": 10906
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.857679255705615e-05,
+      "loss": 0.0506,
+      "step": 10907
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.856010584480154e-05,
+      "loss": 0.0527,
+      "step": 10908
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8543420525543903e-05,
+      "loss": 0.0567,
+      "step": 10909
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.852673659984407e-05,
+      "loss": 0.0649,
+      "step": 10910
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.851005406826289e-05,
+      "loss": 0.0916,
+      "step": 10911
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.849337293136112e-05,
+      "loss": 0.0564,
+      "step": 10912
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.847669318969953e-05,
+      "loss": 0.0834,
+      "step": 10913
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8460014843838784e-05,
+      "loss": 0.0606,
+      "step": 10914
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.844333789433953e-05,
+      "loss": 0.0546,
+      "step": 10915
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.842666234176238e-05,
+      "loss": 0.079,
+      "step": 10916
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.840998818666785e-05,
+      "loss": 0.0458,
+      "step": 10917
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8393315429616456e-05,
+      "loss": 0.0475,
+      "step": 10918
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.837664407116865e-05,
+      "loss": 0.0636,
+      "step": 10919
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.835997411188487e-05,
+      "loss": 0.0926,
+      "step": 10920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.834330555232539e-05,
+      "loss": 0.0648,
+      "step": 10921
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.832663839305057e-05,
+      "loss": 0.0554,
+      "step": 10922
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.830997263462067e-05,
+      "loss": 0.0659,
+      "step": 10923
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.829330827759596e-05,
+      "loss": 0.0549,
+      "step": 10924
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.827664532253652e-05,
+      "loss": 0.0551,
+      "step": 10925
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8259983770002505e-05,
+      "loss": 0.0548,
+      "step": 10926
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.824332362055401e-05,
+      "loss": 0.0705,
+      "step": 10927
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.822666487475106e-05,
+      "loss": 0.0656,
+      "step": 10928
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8210007533153666e-05,
+      "loss": 0.0601,
+      "step": 10929
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.81933515963217e-05,
+      "loss": 0.0558,
+      "step": 10930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8176697064815076e-05,
+      "loss": 0.1003,
+      "step": 10931
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.816004393919363e-05,
+      "loss": 0.0638,
+      "step": 10932
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.814339222001721e-05,
+      "loss": 0.0639,
+      "step": 10933
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.812674190784551e-05,
+      "loss": 0.0474,
+      "step": 10934
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.811009300323818e-05,
+      "loss": 0.1038,
+      "step": 10935
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.809344550675493e-05,
+      "loss": 0.0622,
+      "step": 10936
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.8076799418955354e-05,
+      "loss": 0.0764,
+      "step": 10937
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.806015474039905e-05,
+      "loss": 0.0688,
+      "step": 10938
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.804351147164544e-05,
+      "loss": 0.0542,
+      "step": 10939
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.802686961325404e-05,
+      "loss": 0.0648,
+      "step": 10940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.801022916578426e-05,
+      "loss": 0.05,
+      "step": 10941
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.799359012979546e-05,
+      "loss": 0.0557,
+      "step": 10942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.797695250584699e-05,
+      "loss": 0.0587,
+      "step": 10943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.796031629449806e-05,
+      "loss": 0.0696,
+      "step": 10944
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.794368149630793e-05,
+      "loss": 0.0564,
+      "step": 10945
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.792704811183583e-05,
+      "loss": 0.0582,
+      "step": 10946
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7910416141640766e-05,
+      "loss": 0.0722,
+      "step": 10947
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7893785586281935e-05,
+      "loss": 0.0846,
+      "step": 10948
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7877156446318284e-05,
+      "loss": 0.0831,
+      "step": 10949
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7860528722308846e-05,
+      "loss": 0.0677,
+      "step": 10950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.784390241481255e-05,
+      "loss": 0.0465,
+      "step": 10951
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7827277524388324e-05,
+      "loss": 0.1702,
+      "step": 10952
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.781065405159494e-05,
+      "loss": 0.0511,
+      "step": 10953
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.779403199699123e-05,
+      "loss": 0.0479,
+      "step": 10954
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7777411361135936e-05,
+      "loss": 0.0562,
+      "step": 10955
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.776079214458778e-05,
+      "loss": 0.0581,
+      "step": 10956
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.774417434790543e-05,
+      "loss": 0.0552,
+      "step": 10957
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7727557971647427e-05,
+      "loss": 0.0507,
+      "step": 10958
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.771094301637241e-05,
+      "loss": 0.0621,
+      "step": 10959
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.76943294826388e-05,
+      "loss": 0.0756,
+      "step": 10960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.767771737100509e-05,
+      "loss": 0.0637,
+      "step": 10961
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7661106682029756e-05,
+      "loss": 0.0524,
+      "step": 10962
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.764449741627107e-05,
+      "loss": 0.0611,
+      "step": 10963
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.76278895742874e-05,
+      "loss": 0.0526,
+      "step": 10964
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.761128315663701e-05,
+      "loss": 0.0463,
+      "step": 10965
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.759467816387811e-05,
+      "loss": 0.0643,
+      "step": 10966
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.757807459656894e-05,
+      "loss": 0.0501,
+      "step": 10967
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.756147245526754e-05,
+      "loss": 0.0566,
+      "step": 10968
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7544871740532025e-05,
+      "loss": 0.0596,
+      "step": 10969
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.752827245292041e-05,
+      "loss": 0.0656,
+      "step": 10970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.751167459299074e-05,
+      "loss": 0.0602,
+      "step": 10971
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.749507816130091e-05,
+      "loss": 0.1294,
+      "step": 10972
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.747848315840875e-05,
+      "loss": 0.0788,
+      "step": 10973
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.746188958487215e-05,
+      "loss": 0.0536,
+      "step": 10974
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7445297441248914e-05,
+      "loss": 0.06,
+      "step": 10975
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.742870672809679e-05,
+      "loss": 0.06,
+      "step": 10976
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.741211744597342e-05,
+      "loss": 0.1129,
+      "step": 10977
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.739552959543649e-05,
+      "loss": 0.0782,
+      "step": 10978
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.73789431770436e-05,
+      "loss": 0.0607,
+      "step": 10979
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7362358191352275e-05,
+      "loss": 0.0533,
+      "step": 10980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.734577463892008e-05,
+      "loss": 0.0766,
+      "step": 10981
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.73291925203044e-05,
+      "loss": 0.0641,
+      "step": 10982
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7312611836062645e-05,
+      "loss": 0.0583,
+      "step": 10983
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.729603258675225e-05,
+      "loss": 0.1371,
+      "step": 10984
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.727945477293042e-05,
+      "loss": 0.0697,
+      "step": 10985
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7262878395154494e-05,
+      "loss": 0.0505,
+      "step": 10986
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.724630345398163e-05,
+      "loss": 0.0653,
+      "step": 10987
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.722972994996901e-05,
+      "loss": 0.1123,
+      "step": 10988
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.721315788367376e-05,
+      "loss": 0.0587,
+      "step": 10989
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.719658725565299e-05,
+      "loss": 0.0532,
+      "step": 10990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.718001806646363e-05,
+      "loss": 0.0705,
+      "step": 10991
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7163450316662704e-05,
+      "loss": 0.0541,
+      "step": 10992
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.714688400680711e-05,
+      "loss": 0.0738,
+      "step": 10993
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7130319137453745e-05,
+      "loss": 0.0479,
+      "step": 10994
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.711375570915949e-05,
+      "loss": 0.0558,
+      "step": 10995
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.709719372248099e-05,
+      "loss": 0.0518,
+      "step": 10996
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.708063317797512e-05,
+      "loss": 0.0527,
+      "step": 10997
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7064074076198423e-05,
+      "loss": 0.0588,
+      "step": 10998
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7047516417707606e-05,
+      "loss": 0.051,
+      "step": 10999
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.703096020305927e-05,
+      "loss": 0.0502,
+      "step": 11000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.701440543280989e-05,
+      "loss": 0.0612,
+      "step": 11001
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.699785210751599e-05,
+      "loss": 0.0555,
+      "step": 11002
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.6981300227734e-05,
+      "loss": 0.0621,
+      "step": 11003
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.6964749794020354e-05,
+      "loss": 0.0883,
+      "step": 11004
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.694820080693131e-05,
+      "loss": 0.0636,
+      "step": 11005
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.6931653267023187e-05,
+      "loss": 0.0543,
+      "step": 11006
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.6915107174852265e-05,
+      "loss": 0.0609,
+      "step": 11007
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.68985625309747e-05,
+      "loss": 0.0594,
+      "step": 11008
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.68820193359467e-05,
+      "loss": 0.0538,
+      "step": 11009
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.686547759032432e-05,
+      "loss": 0.0465,
+      "step": 11010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.6848937294663564e-05,
+      "loss": 0.0543,
+      "step": 11011
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.683239844952046e-05,
+      "loss": 0.0568,
+      "step": 11012
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.6815861055451004e-05,
+      "loss": 0.0564,
+      "step": 11013
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.679932511301108e-05,
+      "loss": 0.066,
+      "step": 11014
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.67827906227565e-05,
+      "loss": 0.0499,
+      "step": 11015
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.67662575852431e-05,
+      "loss": 0.0695,
+      "step": 11016
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6749726001026625e-05,
+      "loss": 0.0658,
+      "step": 11017
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6733195870662836e-05,
+      "loss": 0.0667,
+      "step": 11018
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6716667194707316e-05,
+      "loss": 0.0567,
+      "step": 11019
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.670013997371569e-05,
+      "loss": 0.0562,
+      "step": 11020
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6683614208243544e-05,
+      "loss": 0.0576,
+      "step": 11021
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.666708989884637e-05,
+      "loss": 0.0519,
+      "step": 11022
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6650567046079696e-05,
+      "loss": 0.0566,
+      "step": 11023
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.663404565049886e-05,
+      "loss": 0.0522,
+      "step": 11024
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.661752571265921e-05,
+      "loss": 0.0578,
+      "step": 11025
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.66010072331161e-05,
+      "loss": 0.0482,
+      "step": 11026
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6584490212424804e-05,
+      "loss": 0.08,
+      "step": 11027
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6567974651140566e-05,
+      "loss": 0.0555,
+      "step": 11028
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.655146054981847e-05,
+      "loss": 0.0483,
+      "step": 11029
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.653494790901369e-05,
+      "loss": 0.0595,
+      "step": 11030
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.651843672928129e-05,
+      "loss": 0.0543,
+      "step": 11031
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.650192701117634e-05,
+      "loss": 0.089,
+      "step": 11032
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6485418755253704e-05,
+      "loss": 0.0579,
+      "step": 11033
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6468911962068384e-05,
+      "loss": 0.0559,
+      "step": 11034
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.645240663217521e-05,
+      "loss": 0.0753,
+      "step": 11035
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.643590276612909e-05,
+      "loss": 0.049,
+      "step": 11036
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.64194003644847e-05,
+      "loss": 0.0584,
+      "step": 11037
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.640289942779685e-05,
+      "loss": 0.1179,
+      "step": 11038
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.638639995662013e-05,
+      "loss": 0.0505,
+      "step": 11039
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.636990195150921e-05,
+      "loss": 0.064,
+      "step": 11040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.635340541301867e-05,
+      "loss": 0.0473,
+      "step": 11041
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.633691034170307e-05,
+      "loss": 0.0625,
+      "step": 11042
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.632041673811683e-05,
+      "loss": 0.0532,
+      "step": 11043
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6303924602814416e-05,
+      "loss": 0.0505,
+      "step": 11044
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.628743393635019e-05,
+      "loss": 0.0568,
+      "step": 11045
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.627094473927854e-05,
+      "loss": 0.0744,
+      "step": 11046
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.625445701215367e-05,
+      "loss": 0.0506,
+      "step": 11047
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.623797075552983e-05,
+      "loss": 0.0723,
+      "step": 11048
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.622148596996128e-05,
+      "loss": 0.0665,
+      "step": 11049
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.620500265600206e-05,
+      "loss": 0.0542,
+      "step": 11050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.618852081420627e-05,
+      "loss": 0.0613,
+      "step": 11051
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.617204044512802e-05,
+      "loss": 0.058,
+      "step": 11052
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6155561549321176e-05,
+      "loss": 0.0591,
+      "step": 11053
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.613908412733977e-05,
+      "loss": 0.0548,
+      "step": 11054
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.612260817973763e-05,
+      "loss": 0.0498,
+      "step": 11055
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.610613370706867e-05,
+      "loss": 0.0547,
+      "step": 11056
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6089660709886593e-05,
+      "loss": 0.0533,
+      "step": 11057
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6073189188745165e-05,
+      "loss": 0.0675,
+      "step": 11058
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6056719144198076e-05,
+      "loss": 0.0575,
+      "step": 11059
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.604025057679903e-05,
+      "loss": 0.0549,
+      "step": 11060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.6023783487101486e-05,
+      "loss": 0.0997,
+      "step": 11061
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.600731787565912e-05,
+      "loss": 0.0813,
+      "step": 11062
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.599085374302528e-05,
+      "loss": 0.0578,
+      "step": 11063
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.597439108975351e-05,
+      "loss": 0.0627,
+      "step": 11064
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.595792991639715e-05,
+      "loss": 0.0961,
+      "step": 11065
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.59414702235096e-05,
+      "loss": 0.0495,
+      "step": 11066
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.592501201164407e-05,
+      "loss": 0.0801,
+      "step": 11067
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.5908555281353844e-05,
+      "loss": 0.0771,
+      "step": 11068
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.589210003319212e-05,
+      "loss": 0.0621,
+      "step": 11069
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.5875646267712044e-05,
+      "loss": 0.0501,
+      "step": 11070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.5859193985466675e-05,
+      "loss": 0.065,
+      "step": 11071
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.584274318700906e-05,
+      "loss": 0.0644,
+      "step": 11072
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.58262938728922e-05,
+      "loss": 0.0736,
+      "step": 11073
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.580984604366909e-05,
+      "loss": 0.0596,
+      "step": 11074
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.579339969989257e-05,
+      "loss": 0.0571,
+      "step": 11075
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5776954842115425e-05,
+      "loss": 0.0561,
+      "step": 11076
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.576051147089052e-05,
+      "loss": 0.0702,
+      "step": 11077
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.574406958677058e-05,
+      "loss": 0.0581,
+      "step": 11078
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.57276291903083e-05,
+      "loss": 0.0505,
+      "step": 11079
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.571119028205635e-05,
+      "loss": 0.055,
+      "step": 11080
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.569475286256724e-05,
+      "loss": 0.0558,
+      "step": 11081
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5678316932393596e-05,
+      "loss": 0.0751,
+      "step": 11082
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.566188249208786e-05,
+      "loss": 0.0523,
+      "step": 11083
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.564544954220252e-05,
+      "loss": 0.0514,
+      "step": 11084
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.562901808328992e-05,
+      "loss": 0.0644,
+      "step": 11085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.561258811590241e-05,
+      "loss": 0.0757,
+      "step": 11086
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.559615964059229e-05,
+      "loss": 0.0507,
+      "step": 11087
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5579732657911856e-05,
+      "loss": 0.0735,
+      "step": 11088
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5563307168413226e-05,
+      "loss": 0.0702,
+      "step": 11089
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.554688317264853e-05,
+      "loss": 0.053,
+      "step": 11090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.553046067116988e-05,
+      "loss": 0.0517,
+      "step": 11091
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5514039664529335e-05,
+      "loss": 0.0649,
+      "step": 11092
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5497620153278876e-05,
+      "loss": 0.054,
+      "step": 11093
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.548120213797048e-05,
+      "loss": 0.0578,
+      "step": 11094
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5464785619155955e-05,
+      "loss": 0.0508,
+      "step": 11095
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.544837059738719e-05,
+      "loss": 0.0633,
+      "step": 11096
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5431957073215966e-05,
+      "loss": 0.0731,
+      "step": 11097
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.541554504719406e-05,
+      "loss": 0.0633,
+      "step": 11098
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.539913451987309e-05,
+      "loss": 0.0495,
+      "step": 11099
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.538272549180472e-05,
+      "loss": 0.063,
+      "step": 11100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.536631796354059e-05,
+      "loss": 0.0637,
+      "step": 11101
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.534991193563216e-05,
+      "loss": 0.078,
+      "step": 11102
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.533350740863095e-05,
+      "loss": 0.0604,
+      "step": 11103
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5317104383088415e-05,
+      "loss": 0.068,
+      "step": 11104
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.53007028595559e-05,
+      "loss": 0.0593,
+      "step": 11105
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.528430283858474e-05,
+      "loss": 0.0735,
+      "step": 11106
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.526790432072626e-05,
+      "loss": 0.0537,
+      "step": 11107
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5251507306531713e-05,
+      "loss": 0.0492,
+      "step": 11108
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.523511179655221e-05,
+      "loss": 0.0532,
+      "step": 11109
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5218717791338905e-05,
+      "loss": 0.0628,
+      "step": 11110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.520232529144289e-05,
+      "loss": 0.0494,
+      "step": 11111
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.518593429741525e-05,
+      "loss": 0.0545,
+      "step": 11112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5169544809806914e-05,
+      "loss": 0.073,
+      "step": 11113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5153156829168784e-05,
+      "loss": 0.0572,
+      "step": 11114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.513677035605176e-05,
+      "loss": 0.0601,
+      "step": 11115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.512038539100669e-05,
+      "loss": 0.0611,
+      "step": 11116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.510400193458435e-05,
+      "loss": 0.0442,
+      "step": 11117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5087619987335494e-05,
+      "loss": 0.0616,
+      "step": 11118
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.507123954981073e-05,
+      "loss": 0.0521,
+      "step": 11119
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.505486062256072e-05,
+      "loss": 0.0609,
+      "step": 11120
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.503848320613605e-05,
+      "loss": 0.0691,
+      "step": 11121
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.5022107301087275e-05,
+      "loss": 0.0561,
+      "step": 11122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.500573290796479e-05,
+      "loss": 0.0562,
+      "step": 11123
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.498936002731906e-05,
+      "loss": 0.0507,
+      "step": 11124
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.497298865970046e-05,
+      "loss": 0.0531,
+      "step": 11125
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.495661880565934e-05,
+      "loss": 0.0567,
+      "step": 11126
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.494025046574595e-05,
+      "loss": 0.0689,
+      "step": 11127
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.4923883640510456e-05,
+      "loss": 0.0487,
+      "step": 11128
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.490751833050306e-05,
+      "loss": 0.0451,
+      "step": 11129
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.4891154536273894e-05,
+      "loss": 0.0615,
+      "step": 11130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.487479225837303e-05,
+      "loss": 0.0584,
+      "step": 11131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.485843149735051e-05,
+      "loss": 0.0499,
+      "step": 11132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4842072253756215e-05,
+      "loss": 0.0654,
+      "step": 11133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.482571452814012e-05,
+      "loss": 0.0503,
+      "step": 11134
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.480935832105206e-05,
+      "loss": 0.0581,
+      "step": 11135
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4793003633041915e-05,
+      "loss": 0.0519,
+      "step": 11136
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4776650464659354e-05,
+      "loss": 0.0466,
+      "step": 11137
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4760298816454115e-05,
+      "loss": 0.0485,
+      "step": 11138
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.474394868897592e-05,
+      "loss": 0.0543,
+      "step": 11139
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.472760008277427e-05,
+      "loss": 0.0512,
+      "step": 11140
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.471125299839882e-05,
+      "loss": 0.0588,
+      "step": 11141
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4694907436399e-05,
+      "loss": 0.0919,
+      "step": 11142
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.467856339732429e-05,
+      "loss": 0.0542,
+      "step": 11143
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.466222088172408e-05,
+      "loss": 0.058,
+      "step": 11144
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.464587989014776e-05,
+      "loss": 0.0526,
+      "step": 11145
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4629540423144645e-05,
+      "loss": 0.0564,
+      "step": 11146
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4613202481263913e-05,
+      "loss": 0.0519,
+      "step": 11147
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4596866065054786e-05,
+      "loss": 0.0555,
+      "step": 11148
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.458053117506643e-05,
+      "loss": 0.0705,
+      "step": 11149
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4564197811847965e-05,
+      "loss": 0.0533,
+      "step": 11150
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4547865975948364e-05,
+      "loss": 0.0542,
+      "step": 11151
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.453153566791671e-05,
+      "loss": 0.0447,
+      "step": 11152
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.451520688830185e-05,
+      "loss": 0.0647,
+      "step": 11153
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.44988796376527e-05,
+      "loss": 0.067,
+      "step": 11154
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4482553916518174e-05,
+      "loss": 0.0491,
+      "step": 11155
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.446622972544695e-05,
+      "loss": 0.0519,
+      "step": 11156
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.444990706498782e-05,
+      "loss": 0.0659,
+      "step": 11157
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.443358593568946e-05,
+      "loss": 0.068,
+      "step": 11158
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.441726633810049e-05,
+      "loss": 0.055,
+      "step": 11159
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.440094827276956e-05,
+      "loss": 0.0517,
+      "step": 11160
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4384631740245104e-05,
+      "loss": 0.0932,
+      "step": 11161
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4368316741075654e-05,
+      "loss": 0.0577,
+      "step": 11162
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.435200327580961e-05,
+      "loss": 0.0525,
+      "step": 11163
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.433569134499541e-05,
+      "loss": 0.0729,
+      "step": 11164
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.431938094918132e-05,
+      "loss": 0.0574,
+      "step": 11165
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4303072088915585e-05,
+      "loss": 0.0482,
+      "step": 11166
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.428676476474646e-05,
+      "loss": 0.0576,
+      "step": 11167
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4270458977222115e-05,
+      "loss": 0.051,
+      "step": 11168
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.42541547268907e-05,
+      "loss": 0.0538,
+      "step": 11169
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.423785201430023e-05,
+      "loss": 0.0503,
+      "step": 11170
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4221550839998724e-05,
+      "loss": 0.0476,
+      "step": 11171
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.420525120453416e-05,
+      "loss": 0.0636,
+      "step": 11172
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.418895310845444e-05,
+      "loss": 0.065,
+      "step": 11173
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4172656552307464e-05,
+      "loss": 0.0449,
+      "step": 11174
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4156361536640965e-05,
+      "loss": 0.0569,
+      "step": 11175
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4140068062002737e-05,
+      "loss": 0.053,
+      "step": 11176
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4123776128940484e-05,
+      "loss": 0.0606,
+      "step": 11177
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4107485738001886e-05,
+      "loss": 0.0486,
+      "step": 11178
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.409119688973452e-05,
+      "loss": 0.0605,
+      "step": 11179
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4074909584685875e-05,
+      "loss": 0.0637,
+      "step": 11180
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.40586238234035e-05,
+      "loss": 0.0603,
+      "step": 11181
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.4042339606434834e-05,
+      "loss": 0.0711,
+      "step": 11182
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.402605693432731e-05,
+      "loss": 0.08,
+      "step": 11183
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.400977580762818e-05,
+      "loss": 0.0597,
+      "step": 11184
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.399349622688479e-05,
+      "loss": 0.0517,
+      "step": 11185
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.397721819264436e-05,
+      "loss": 0.0584,
+      "step": 11186
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.396094170545409e-05,
+      "loss": 0.054,
+      "step": 11187
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.394466676586114e-05,
+      "loss": 0.0534,
+      "step": 11188
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3928393374412514e-05,
+      "loss": 0.0498,
+      "step": 11189
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.391212153165528e-05,
+      "loss": 0.0525,
+      "step": 11190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.389585123813644e-05,
+      "loss": 0.0481,
+      "step": 11191
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.387958249440286e-05,
+      "loss": 0.0573,
+      "step": 11192
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.386331530100148e-05,
+      "loss": 0.0455,
+      "step": 11193
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3847049658479045e-05,
+      "loss": 0.0489,
+      "step": 11194
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.383078556738235e-05,
+      "loss": 0.0591,
+      "step": 11195
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.381452302825814e-05,
+      "loss": 0.0927,
+      "step": 11196
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.379826204165307e-05,
+      "loss": 0.0541,
+      "step": 11197
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3782002608113725e-05,
+      "loss": 0.0554,
+      "step": 11198
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.376574472818665e-05,
+      "loss": 0.0656,
+      "step": 11199
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.37494884024184e-05,
+      "loss": 0.0538,
+      "step": 11200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.373323363135541e-05,
+      "loss": 0.0603,
+      "step": 11201
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.371698041554412e-05,
+      "loss": 0.0528,
+      "step": 11202
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.370072875553079e-05,
+      "loss": 0.0647,
+      "step": 11203
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3684478651861815e-05,
+      "loss": 0.0515,
+      "step": 11204
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.366823010508336e-05,
+      "loss": 0.0571,
+      "step": 11205
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3651983115741646e-05,
+      "loss": 0.0569,
+      "step": 11206
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.363573768438286e-05,
+      "loss": 0.0499,
+      "step": 11207
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3619493811553025e-05,
+      "loss": 0.0577,
+      "step": 11208
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.360325149779818e-05,
+      "loss": 0.0637,
+      "step": 11209
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3587010743664334e-05,
+      "loss": 0.0636,
+      "step": 11210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.357077154969742e-05,
+      "loss": 0.051,
+      "step": 11211
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.355453391644334e-05,
+      "loss": 0.0526,
+      "step": 11212
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.353829784444786e-05,
+      "loss": 0.0755,
+      "step": 11213
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.352206333425678e-05,
+      "loss": 0.0496,
+      "step": 11214
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3505830386415835e-05,
+      "loss": 0.0577,
+      "step": 11215
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.34895990014707e-05,
+      "loss": 0.0465,
+      "step": 11216
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3473369179966994e-05,
+      "loss": 0.0547,
+      "step": 11217
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.345714092245021e-05,
+      "loss": 0.0552,
+      "step": 11218
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.344091422946591e-05,
+      "loss": 0.0517,
+      "step": 11219
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.342468910155957e-05,
+      "loss": 0.0451,
+      "step": 11220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.34084655392766e-05,
+      "loss": 0.0678,
+      "step": 11221
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.33922435431623e-05,
+      "loss": 0.061,
+      "step": 11222
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3376023113761995e-05,
+      "loss": 0.0564,
+      "step": 11223
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.335980425162094e-05,
+      "loss": 0.0437,
+      "step": 11224
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.334358695728432e-05,
+      "loss": 0.0589,
+      "step": 11225
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.332737123129734e-05,
+      "loss": 0.0724,
+      "step": 11226
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3311157074205e-05,
+      "loss": 0.0682,
+      "step": 11227
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.329494448655236e-05,
+      "loss": 0.0705,
+      "step": 11228
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.327873346888441e-05,
+      "loss": 0.0675,
+      "step": 11229
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.326252402174614e-05,
+      "loss": 0.0524,
+      "step": 11230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3246316145682385e-05,
+      "loss": 0.0472,
+      "step": 11231
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.323010984123791e-05,
+      "loss": 0.0573,
+      "step": 11232
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.321390510895754e-05,
+      "loss": 0.0522,
+      "step": 11233
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3197701949386e-05,
+      "loss": 0.0672,
+      "step": 11234
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3181500363067993e-05,
+      "loss": 0.1318,
+      "step": 11235
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.316530035054805e-05,
+      "loss": 0.0576,
+      "step": 11236
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.314910191237079e-05,
+      "loss": 0.0531,
+      "step": 11237
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3132905049080695e-05,
+      "loss": 0.0718,
+      "step": 11238
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3116709761222225e-05,
+      "loss": 0.0562,
+      "step": 11239
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3100516049339853e-05,
+      "loss": 0.0846,
+      "step": 11240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.308432391397781e-05,
+      "loss": 0.0473,
+      "step": 11241
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.306813335568048e-05,
+      "loss": 0.0591,
+      "step": 11242
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.3051944374992056e-05,
+      "loss": 0.0637,
+      "step": 11243
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.303575697245674e-05,
+      "loss": 0.0563,
+      "step": 11244
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.301957114861871e-05,
+      "loss": 0.0688,
+      "step": 11245
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.300338690402199e-05,
+      "loss": 0.0654,
+      "step": 11246
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.298720423921063e-05,
+      "loss": 0.061,
+      "step": 11247
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2971023154728614e-05,
+      "loss": 0.0621,
+      "step": 11248
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2954843651119926e-05,
+      "loss": 0.0489,
+      "step": 11249
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.293866572892834e-05,
+      "loss": 0.0495,
+      "step": 11250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2922489388697704e-05,
+      "loss": 0.0979,
+      "step": 11251
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.29063146309718e-05,
+      "loss": 0.0596,
+      "step": 11252
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.289014145629434e-05,
+      "loss": 0.0511,
+      "step": 11253
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2873969865209025e-05,
+      "loss": 0.0685,
+      "step": 11254
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2857799858259425e-05,
+      "loss": 0.0554,
+      "step": 11255
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.284163143598904e-05,
+      "loss": 0.0558,
+      "step": 11256
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.28254645989414e-05,
+      "loss": 0.0541,
+      "step": 11257
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.280929934765997e-05,
+      "loss": 0.0541,
+      "step": 11258
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.279313568268819e-05,
+      "loss": 0.0481,
+      "step": 11259
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2776973604569304e-05,
+      "loss": 0.0601,
+      "step": 11260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.276081311384664e-05,
+      "loss": 0.0631,
+      "step": 11261
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.274465421106343e-05,
+      "loss": 0.0566,
+      "step": 11262
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.27284968967629e-05,
+      "loss": 0.0556,
+      "step": 11263
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.27123411714881e-05,
+      "loss": 0.0535,
+      "step": 11264
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.269618703578213e-05,
+      "loss": 0.053,
+      "step": 11265
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.268003449018801e-05,
+      "loss": 0.08,
+      "step": 11266
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2663883535248714e-05,
+      "loss": 0.0507,
+      "step": 11267
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.264773417150719e-05,
+      "loss": 0.0499,
+      "step": 11268
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.263158639950627e-05,
+      "loss": 0.0614,
+      "step": 11269
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2615440219788706e-05,
+      "loss": 0.0492,
+      "step": 11270
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2599295632897275e-05,
+      "loss": 0.0914,
+      "step": 11271
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.25831526393747e-05,
+      "loss": 0.0798,
+      "step": 11272
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2567011239763664e-05,
+      "loss": 0.0502,
+      "step": 11273
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.255087143460666e-05,
+      "loss": 0.0589,
+      "step": 11274
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.253473322444629e-05,
+      "loss": 0.0713,
+      "step": 11275
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.251859660982501e-05,
+      "loss": 0.0551,
+      "step": 11276
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2502461591285316e-05,
+      "loss": 0.0553,
+      "step": 11277
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.248632816936949e-05,
+      "loss": 0.0507,
+      "step": 11278
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2470196344619895e-05,
+      "loss": 0.0476,
+      "step": 11279
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2454066117578815e-05,
+      "loss": 0.107,
+      "step": 11280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.243793748878849e-05,
+      "loss": 0.0543,
+      "step": 11281
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2421810458791e-05,
+      "loss": 0.0568,
+      "step": 11282
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2405685028128546e-05,
+      "loss": 0.0687,
+      "step": 11283
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.238956119734311e-05,
+      "loss": 0.0592,
+      "step": 11284
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2373438966976695e-05,
+      "loss": 0.0604,
+      "step": 11285
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.235731833757129e-05,
+      "loss": 0.0585,
+      "step": 11286
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.23411993096688e-05,
+      "loss": 0.054,
+      "step": 11287
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2325081883811e-05,
+      "loss": 0.0569,
+      "step": 11288
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.23089660605397e-05,
+      "loss": 0.0546,
+      "step": 11289
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.229285184039665e-05,
+      "loss": 0.0931,
+      "step": 11290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.227673922392354e-05,
+      "loss": 0.0492,
+      "step": 11291
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.2260628211661955e-05,
+      "loss": 0.1047,
+      "step": 11292
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.224451880415346e-05,
+      "loss": 0.0581,
+      "step": 11293
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.222841100193965e-05,
+      "loss": 0.0527,
+      "step": 11294
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.221230480556189e-05,
+      "loss": 0.0537,
+      "step": 11295
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.219620021556162e-05,
+      "loss": 0.0688,
+      "step": 11296
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.218009723248023e-05,
+      "loss": 0.0522,
+      "step": 11297
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.216399585685896e-05,
+      "loss": 0.073,
+      "step": 11298
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.214789608923909e-05,
+      "loss": 0.0542,
+      "step": 11299
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.213179793016181e-05,
+      "loss": 0.0531,
+      "step": 11300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.211570138016828e-05,
+      "loss": 0.0573,
+      "step": 11301
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.209960643979952e-05,
+      "loss": 0.0536,
+      "step": 11302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.2083513109596616e-05,
+      "loss": 0.0522,
+      "step": 11303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.2067421390100515e-05,
+      "loss": 0.0489,
+      "step": 11304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.205133128185219e-05,
+      "loss": 0.063,
+      "step": 11305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.203524278539242e-05,
+      "loss": 0.0517,
+      "step": 11306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.201915590126212e-05,
+      "loss": 0.0544,
+      "step": 11307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.200307063000194e-05,
+      "loss": 0.0511,
+      "step": 11308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.198698697215265e-05,
+      "loss": 0.055,
+      "step": 11309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1970904928254874e-05,
+      "loss": 0.0555,
+      "step": 11310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.195482449884928e-05,
+      "loss": 0.0556,
+      "step": 11311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.19387456844763e-05,
+      "loss": 0.0578,
+      "step": 11312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1922668485676463e-05,
+      "loss": 0.0499,
+      "step": 11313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.190659290299023e-05,
+      "loss": 0.0539,
+      "step": 11314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1890518936957987e-05,
+      "loss": 0.0649,
+      "step": 11315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.187444658812e-05,
+      "loss": 0.0811,
+      "step": 11316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1858375857016575e-05,
+      "loss": 0.1141,
+      "step": 11317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.184230674418792e-05,
+      "loss": 0.0551,
+      "step": 11318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1826239250174246e-05,
+      "loss": 0.0501,
+      "step": 11319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.181017337551561e-05,
+      "loss": 0.0577,
+      "step": 11320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.179410912075202e-05,
+      "loss": 0.0497,
+      "step": 11321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.177804648642355e-05,
+      "loss": 0.0616,
+      "step": 11322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.17619854730701e-05,
+      "loss": 0.0543,
+      "step": 11323
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1745926081231564e-05,
+      "loss": 0.0524,
+      "step": 11324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.172986831144785e-05,
+      "loss": 0.0495,
+      "step": 11325
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.171381216425863e-05,
+      "loss": 0.0662,
+      "step": 11326
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.169775764020366e-05,
+      "loss": 0.0556,
+      "step": 11327
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1681704739822647e-05,
+      "loss": 0.0661,
+      "step": 11328
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.166565346365522e-05,
+      "loss": 0.052,
+      "step": 11329
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.164960381224087e-05,
+      "loss": 0.0599,
+      "step": 11330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1633555786119125e-05,
+      "loss": 0.051,
+      "step": 11331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1617509385829474e-05,
+      "loss": 0.055,
+      "step": 11332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.160146461191133e-05,
+      "loss": 0.0524,
+      "step": 11333
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1585421464903994e-05,
+      "loss": 0.0522,
+      "step": 11334
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.156937994534672e-05,
+      "loss": 0.0529,
+      "step": 11335
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.155334005377879e-05,
+      "loss": 0.0585,
+      "step": 11336
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.153730179073937e-05,
+      "loss": 0.0434,
+      "step": 11337
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.152126515676759e-05,
+      "loss": 0.0526,
+      "step": 11338
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1505230152402564e-05,
+      "loss": 0.0705,
+      "step": 11339
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.148919677818323e-05,
+      "loss": 0.0672,
+      "step": 11340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1473165034648565e-05,
+      "loss": 0.0793,
+      "step": 11341
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.14571349223375e-05,
+      "loss": 0.0648,
+      "step": 11342
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.14411064417889e-05,
+      "loss": 0.0548,
+      "step": 11343
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1425079593541505e-05,
+      "loss": 0.0552,
+      "step": 11344
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.140905437813409e-05,
+      "loss": 0.0711,
+      "step": 11345
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1393030796105365e-05,
+      "loss": 0.0612,
+      "step": 11346
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1377008847993904e-05,
+      "loss": 0.0539,
+      "step": 11347
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1360988534338306e-05,
+      "loss": 0.0525,
+      "step": 11348
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.134496985567714e-05,
+      "loss": 0.0537,
+      "step": 11349
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.132895281254877e-05,
+      "loss": 0.0538,
+      "step": 11350
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1312937405491676e-05,
+      "loss": 0.0549,
+      "step": 11351
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.129692363504421e-05,
+      "loss": 0.1347,
+      "step": 11352
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.12809115017447e-05,
+      "loss": 0.0586,
+      "step": 11353
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1264901006131326e-05,
+      "loss": 0.0591,
+      "step": 11354
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1248892148742314e-05,
+      "loss": 0.0498,
+      "step": 11355
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.123288493011578e-05,
+      "loss": 0.0864,
+      "step": 11356
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.121687935078987e-05,
+      "loss": 0.0486,
+      "step": 11357
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.120087541130252e-05,
+      "loss": 0.0533,
+      "step": 11358
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1184873112191775e-05,
+      "loss": 0.0623,
+      "step": 11359
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1168872453995484e-05,
+      "loss": 0.0513,
+      "step": 11360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1152873437251535e-05,
+      "loss": 0.0597,
+      "step": 11361
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.113687606249775e-05,
+      "loss": 0.0548,
+      "step": 11362
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1120880330271894e-05,
+      "loss": 0.0568,
+      "step": 11363
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.11048862411116e-05,
+      "loss": 0.0512,
+      "step": 11364
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.108889379555454e-05,
+      "loss": 0.0525,
+      "step": 11365
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.107290299413829e-05,
+      "loss": 0.0739,
+      "step": 11366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1056913837400446e-05,
+      "loss": 0.0538,
+      "step": 11367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.104092632587838e-05,
+      "loss": 0.0638,
+      "step": 11368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1024940460109547e-05,
+      "loss": 0.0506,
+      "step": 11369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.1008956240631325e-05,
+      "loss": 0.0489,
+      "step": 11370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0992973667981035e-05,
+      "loss": 0.0521,
+      "step": 11371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0976992742695925e-05,
+      "loss": 0.0504,
+      "step": 11372
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.096101346531314e-05,
+      "loss": 0.0587,
+      "step": 11373
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.094503583636985e-05,
+      "loss": 0.0589,
+      "step": 11374
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.092905985640315e-05,
+      "loss": 0.051,
+      "step": 11375
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0913085525950056e-05,
+      "loss": 0.0706,
+      "step": 11376
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.08971128455476e-05,
+      "loss": 0.0586,
+      "step": 11377
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.088114181573262e-05,
+      "loss": 0.0605,
+      "step": 11378
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.086517243704201e-05,
+      "loss": 0.0589,
+      "step": 11379
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.08492047100126e-05,
+      "loss": 0.0579,
+      "step": 11380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0833238635181154e-05,
+      "loss": 0.0599,
+      "step": 11381
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.081727421308431e-05,
+      "loss": 0.0508,
+      "step": 11382
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0801311444258746e-05,
+      "loss": 0.064,
+      "step": 11383
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.078535032924109e-05,
+      "loss": 0.0579,
+      "step": 11384
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.076939086856778e-05,
+      "loss": 0.0563,
+      "step": 11385
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.075343306277539e-05,
+      "loss": 0.0544,
+      "step": 11386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.073747691240025e-05,
+      "loss": 0.0693,
+      "step": 11387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.072152241797876e-05,
+      "loss": 0.0602,
+      "step": 11388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.070556958004724e-05,
+      "loss": 0.0579,
+      "step": 11389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0689618399141934e-05,
+      "loss": 0.0574,
+      "step": 11390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0673668875799075e-05,
+      "loss": 0.0488,
+      "step": 11391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.065772101055474e-05,
+      "loss": 0.0589,
+      "step": 11392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.064177480394503e-05,
+      "loss": 0.0528,
+      "step": 11393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0625830256506e-05,
+      "loss": 0.0529,
+      "step": 11394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.060988736877366e-05,
+      "loss": 0.0507,
+      "step": 11395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.059394614128383e-05,
+      "loss": 0.0518,
+      "step": 11396
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.057800657457248e-05,
+      "loss": 0.057,
+      "step": 11397
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.05620686691753e-05,
+      "loss": 0.0556,
+      "step": 11398
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0546132425628115e-05,
+      "loss": 0.0582,
+      "step": 11399
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0530197844466645e-05,
+      "loss": 0.0542,
+      "step": 11400
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.051426492622644e-05,
+      "loss": 0.0604,
+      "step": 11401
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.049833367144314e-05,
+      "loss": 0.0632,
+      "step": 11402
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.048240408065227e-05,
+      "loss": 0.0508,
+      "step": 11403
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.046647615438929e-05,
+      "loss": 0.058,
+      "step": 11404
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.045054989318967e-05,
+      "loss": 0.0507,
+      "step": 11405
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.043462529758867e-05,
+      "loss": 0.0725,
+      "step": 11406
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.041870236812165e-05,
+      "loss": 0.1066,
+      "step": 11407
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.040278110532386e-05,
+      "loss": 0.0555,
+      "step": 11408
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.038686150973051e-05,
+      "loss": 0.0585,
+      "step": 11409
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.037094358187673e-05,
+      "loss": 0.0478,
+      "step": 11410
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.035502732229751e-05,
+      "loss": 0.0519,
+      "step": 11411
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.033911273152796e-05,
+      "loss": 0.0632,
+      "step": 11412
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0323199810103026e-05,
+      "loss": 0.0492,
+      "step": 11413
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0307288558557655e-05,
+      "loss": 0.0561,
+      "step": 11414
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0291378977426626e-05,
+      "loss": 0.0474,
+      "step": 11415
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0275471067244784e-05,
+      "loss": 0.0914,
+      "step": 11416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.025956482854686e-05,
+      "loss": 0.0553,
+      "step": 11417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.024366026186755e-05,
+      "loss": 0.062,
+      "step": 11418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0227757367741525e-05,
+      "loss": 0.0529,
+      "step": 11419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.021185614670327e-05,
+      "loss": 0.0599,
+      "step": 11420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0195956599287355e-05,
+      "loss": 0.0527,
+      "step": 11421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0180058726028235e-05,
+      "loss": 0.0606,
+      "step": 11422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0164162527460345e-05,
+      "loss": 0.0529,
+      "step": 11423
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0148268004118005e-05,
+      "loss": 0.0721,
+      "step": 11424
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.013237515653546e-05,
+      "loss": 0.0556,
+      "step": 11425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0116483985247e-05,
+      "loss": 0.0605,
+      "step": 11426
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.010059449078679e-05,
+      "loss": 0.0587,
+      "step": 11427
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.008470667368901e-05,
+      "loss": 0.0485,
+      "step": 11428
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.006882053448763e-05,
+      "loss": 0.0528,
+      "step": 11429
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0052936073716696e-05,
+      "loss": 0.0565,
+      "step": 11430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.003705329191017e-05,
+      "loss": 0.0533,
+      "step": 11431
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0021172189601974e-05,
+      "loss": 0.0527,
+      "step": 11432
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.000529276732595e-05,
+      "loss": 0.0476,
+      "step": 11433
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9989415025615835e-05,
+      "loss": 0.0543,
+      "step": 11434
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9973538965005374e-05,
+      "loss": 0.0502,
+      "step": 11435
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9957664586028295e-05,
+      "loss": 0.0584,
+      "step": 11436
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.994179188921813e-05,
+      "loss": 0.0498,
+      "step": 11437
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.99259208751085e-05,
+      "loss": 0.0538,
+      "step": 11438
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.991005154423285e-05,
+      "loss": 0.0582,
+      "step": 11439
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.989418389712466e-05,
+      "loss": 0.0475,
+      "step": 11440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.987831793431731e-05,
+      "loss": 0.0473,
+      "step": 11441
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.986245365634418e-05,
+      "loss": 0.073,
+      "step": 11442
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.984659106373846e-05,
+      "loss": 0.0662,
+      "step": 11443
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.983073015703341e-05,
+      "loss": 0.0495,
+      "step": 11444
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9814870936762204e-05,
+      "loss": 0.0474,
+      "step": 11445
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.979901340345794e-05,
+      "loss": 0.0518,
+      "step": 11446
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9783157557653706e-05,
+      "loss": 0.0555,
+      "step": 11447
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.976730339988241e-05,
+      "loss": 0.049,
+      "step": 11448
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.975145093067708e-05,
+      "loss": 0.0592,
+      "step": 11449
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.973560015057049e-05,
+      "loss": 0.053,
+      "step": 11450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.971975106009554e-05,
+      "loss": 0.0524,
+      "step": 11451
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9703903659785e-05,
+      "loss": 0.0633,
+      "step": 11452
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.968805795017152e-05,
+      "loss": 0.0527,
+      "step": 11453
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.967221393178778e-05,
+      "loss": 0.0767,
+      "step": 11454
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9656371605166384e-05,
+      "loss": 0.0628,
+      "step": 11455
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.964053097083987e-05,
+      "loss": 0.0559,
+      "step": 11456
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9624692029340744e-05,
+      "loss": 0.0808,
+      "step": 11457
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.960885478120136e-05,
+      "loss": 0.0547,
+      "step": 11458
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9593019226954126e-05,
+      "loss": 0.0495,
+      "step": 11459
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.957718536713135e-05,
+      "loss": 0.0531,
+      "step": 11460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9561353202265336e-05,
+      "loss": 0.0513,
+      "step": 11461
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.954552273288822e-05,
+      "loss": 0.0628,
+      "step": 11462
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.952969395953212e-05,
+      "loss": 0.0471,
+      "step": 11463
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9513866882729146e-05,
+      "loss": 0.0661,
+      "step": 11464
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.949804150301133e-05,
+      "loss": 0.0562,
+      "step": 11465
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.948221782091067e-05,
+      "loss": 0.0637,
+      "step": 11466
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.946639583695901e-05,
+      "loss": 0.0503,
+      "step": 11467
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.945057555168824e-05,
+      "loss": 0.0563,
+      "step": 11468
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.943475696563016e-05,
+      "loss": 0.058,
+      "step": 11469
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.94189400793165e-05,
+      "loss": 0.0574,
+      "step": 11470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.940312489327898e-05,
+      "loss": 0.0516,
+      "step": 11471
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9387311408049175e-05,
+      "loss": 0.0623,
+      "step": 11472
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.9371499624158654e-05,
+      "loss": 0.0584,
+      "step": 11473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.935568954213896e-05,
+      "loss": 0.055,
+      "step": 11474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.933988116252156e-05,
+      "loss": 0.0547,
+      "step": 11475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.932407448583784e-05,
+      "loss": 0.0561,
+      "step": 11476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.930826951261907e-05,
+      "loss": 0.0483,
+      "step": 11477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9292466243396594e-05,
+      "loss": 0.0615,
+      "step": 11478
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9276664678701626e-05,
+      "loss": 0.05,
+      "step": 11479
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.926086481906538e-05,
+      "loss": 0.0584,
+      "step": 11480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.924506666501887e-05,
+      "loss": 0.0531,
+      "step": 11481
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9229270217093214e-05,
+      "loss": 0.0691,
+      "step": 11482
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.921347547581939e-05,
+      "loss": 0.0579,
+      "step": 11483
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9197682441728355e-05,
+      "loss": 0.0486,
+      "step": 11484
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9181891115351e-05,
+      "loss": 0.0498,
+      "step": 11485
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9166101497218084e-05,
+      "loss": 0.0542,
+      "step": 11486
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9150313587860433e-05,
+      "loss": 0.051,
+      "step": 11487
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.913452738780876e-05,
+      "loss": 0.0598,
+      "step": 11488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.911874289759366e-05,
+      "loss": 0.0505,
+      "step": 11489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.91029601177458e-05,
+      "loss": 0.0523,
+      "step": 11490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.9087179048795626e-05,
+      "loss": 0.0584,
+      "step": 11491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.907139969127368e-05,
+      "loss": 0.0644,
+      "step": 11492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.905562204571037e-05,
+      "loss": 0.0527,
+      "step": 11493
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.903984611263609e-05,
+      "loss": 0.0722,
+      "step": 11494
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.902407189258108e-05,
+      "loss": 0.0562,
+      "step": 11495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.900829938607562e-05,
+      "loss": 0.0712,
+      "step": 11496
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.899252859364991e-05,
+      "loss": 0.0518,
+      "step": 11497
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.897675951583408e-05,
+      "loss": 0.0542,
+      "step": 11498
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.896099215315823e-05,
+      "loss": 0.0558,
+      "step": 11499
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.894522650615236e-05,
+      "loss": 0.0546,
+      "step": 11500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.892946257534639e-05,
+      "loss": 0.0625,
+      "step": 11501
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.891370036127025e-05,
+      "loss": 0.0581,
+      "step": 11502
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.889793986445379e-05,
+      "loss": 0.0557,
+      "step": 11503
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8882181085426836e-05,
+      "loss": 0.0524,
+      "step": 11504
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.886642402471904e-05,
+      "loss": 0.053,
+      "step": 11505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.88506686828601e-05,
+      "loss": 0.0549,
+      "step": 11506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.883491506037965e-05,
+      "loss": 0.0489,
+      "step": 11507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.881916315780728e-05,
+      "loss": 0.0543,
+      "step": 11508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8803412975672404e-05,
+      "loss": 0.0472,
+      "step": 11509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8787664514504504e-05,
+      "loss": 0.0533,
+      "step": 11510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.877191777483296e-05,
+      "loss": 0.0493,
+      "step": 11511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.87561727571871e-05,
+      "loss": 0.0493,
+      "step": 11512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.874042946209624e-05,
+      "loss": 0.0589,
+      "step": 11513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8724687890089525e-05,
+      "loss": 0.0482,
+      "step": 11514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.870894804169609e-05,
+      "loss": 0.0461,
+      "step": 11515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.869320991744506e-05,
+      "loss": 0.0567,
+      "step": 11516
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8677473517865457e-05,
+      "loss": 0.0512,
+      "step": 11517
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.866173884348632e-05,
+      "loss": 0.0547,
+      "step": 11518
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.864600589483648e-05,
+      "loss": 0.0648,
+      "step": 11519
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.863027467244482e-05,
+      "loss": 0.0484,
+      "step": 11520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8614545176840174e-05,
+      "loss": 0.0522,
+      "step": 11521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.859881740855131e-05,
+      "loss": 0.0499,
+      "step": 11522
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.858309136810684e-05,
+      "loss": 0.0569,
+      "step": 11523
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.856736705603542e-05,
+      "loss": 0.0525,
+      "step": 11524
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8551644472865634e-05,
+      "loss": 0.0534,
+      "step": 11525
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8535923619126034e-05,
+      "loss": 0.0519,
+      "step": 11526
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.852020449534499e-05,
+      "loss": 0.0745,
+      "step": 11527
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.850448710205099e-05,
+      "loss": 0.0726,
+      "step": 11528
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.848877143977227e-05,
+      "loss": 0.0668,
+      "step": 11529
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.8473057509037166e-05,
+      "loss": 0.0548,
+      "step": 11530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.84573453103739e-05,
+      "loss": 0.0524,
+      "step": 11531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8441634844310665e-05,
+      "loss": 0.0454,
+      "step": 11532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8425926111375506e-05,
+      "loss": 0.0545,
+      "step": 11533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.841021911209649e-05,
+      "loss": 0.0545,
+      "step": 11534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.839451384700162e-05,
+      "loss": 0.0561,
+      "step": 11535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.837881031661885e-05,
+      "loss": 0.0513,
+      "step": 11536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.836310852147598e-05,
+      "loss": 0.059,
+      "step": 11537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8347408462100884e-05,
+      "loss": 0.0512,
+      "step": 11538
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8331710139021326e-05,
+      "loss": 0.0456,
+      "step": 11539
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.831601355276495e-05,
+      "loss": 0.0508,
+      "step": 11540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8300318703859414e-05,
+      "loss": 0.0575,
+      "step": 11541
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8284625592832346e-05,
+      "loss": 0.0876,
+      "step": 11542
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.826893422021118e-05,
+      "loss": 0.0579,
+      "step": 11543
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.825324458652344e-05,
+      "loss": 0.0607,
+      "step": 11544
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8237556692296515e-05,
+      "loss": 0.0547,
+      "step": 11545
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.82218705380578e-05,
+      "loss": 0.0553,
+      "step": 11546
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.82061861243345e-05,
+      "loss": 0.0493,
+      "step": 11547
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8190503451653865e-05,
+      "loss": 0.05,
+      "step": 11548
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.817482252054311e-05,
+      "loss": 0.064,
+      "step": 11549
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.815914333152934e-05,
+      "loss": 0.0497,
+      "step": 11550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.814346588513956e-05,
+      "loss": 0.0476,
+      "step": 11551
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8127790181900837e-05,
+      "loss": 0.0836,
+      "step": 11552
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.811211622234003e-05,
+      "loss": 0.0577,
+      "step": 11553
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8096444006984054e-05,
+      "loss": 0.0495,
+      "step": 11554
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.808077353635972e-05,
+      "loss": 0.0586,
+      "step": 11555
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.8065104810993856e-05,
+      "loss": 0.0501,
+      "step": 11556
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.804943783141306e-05,
+      "loss": 0.0551,
+      "step": 11557
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.803377259814401e-05,
+      "loss": 0.0506,
+      "step": 11558
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.801810911171332e-05,
+      "loss": 0.0743,
+      "step": 11559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.800244737264753e-05,
+      "loss": 0.0568,
+      "step": 11560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.798678738147303e-05,
+      "loss": 0.0605,
+      "step": 11561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.797112913871629e-05,
+      "loss": 0.0622,
+      "step": 11562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7955472644903634e-05,
+      "loss": 0.0563,
+      "step": 11563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.793981790056141e-05,
+      "loss": 0.0672,
+      "step": 11564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7924164906215796e-05,
+      "loss": 0.0521,
+      "step": 11565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.790851366239295e-05,
+      "loss": 0.0565,
+      "step": 11566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7892864169619e-05,
+      "loss": 0.057,
+      "step": 11567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7877216428420015e-05,
+      "loss": 0.0553,
+      "step": 11568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.786157043932199e-05,
+      "loss": 0.0555,
+      "step": 11569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.78459262028509e-05,
+      "loss": 0.0449,
+      "step": 11570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.783028371953254e-05,
+      "loss": 0.0524,
+      "step": 11571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.781464298989279e-05,
+      "loss": 0.0522,
+      "step": 11572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7799004014457394e-05,
+      "loss": 0.0517,
+      "step": 11573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7783366793752096e-05,
+      "loss": 0.0514,
+      "step": 11574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7767731328302454e-05,
+      "loss": 0.1028,
+      "step": 11575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.775209761863411e-05,
+      "loss": 0.0478,
+      "step": 11576
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.773646566527257e-05,
+      "loss": 0.0508,
+      "step": 11577
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.772083546874335e-05,
+      "loss": 0.0497,
+      "step": 11578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.770520702957182e-05,
+      "loss": 0.0515,
+      "step": 11579
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.768958034828329e-05,
+      "loss": 0.0629,
+      "step": 11580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7673955425403075e-05,
+      "loss": 0.0464,
+      "step": 11581
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.7658332261456416e-05,
+      "loss": 0.0449,
+      "step": 11582
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.764271085696848e-05,
+      "loss": 0.0517,
+      "step": 11583
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.762709121246442e-05,
+      "loss": 0.0635,
+      "step": 11584
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.76114733284692e-05,
+      "loss": 0.0529,
+      "step": 11585
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.759585720550788e-05,
+      "loss": 0.0601,
+      "step": 11586
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.758024284410536e-05,
+      "loss": 0.0621,
+      "step": 11587
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.756463024478657e-05,
+      "loss": 0.0481,
+      "step": 11588
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.754901940807624e-05,
+      "loss": 0.0564,
+      "step": 11589
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.753341033449918e-05,
+      "loss": 0.0566,
+      "step": 11590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.751780302458012e-05,
+      "loss": 0.059,
+      "step": 11591
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.750219747884362e-05,
+      "loss": 0.0538,
+      "step": 11592
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.74865936978143e-05,
+      "loss": 0.0758,
+      "step": 11593
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7470991682016694e-05,
+      "loss": 0.055,
+      "step": 11594
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.745539143197522e-05,
+      "loss": 0.0583,
+      "step": 11595
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.743979294821429e-05,
+      "loss": 0.0508,
+      "step": 11596
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.742419623125826e-05,
+      "loss": 0.0541,
+      "step": 11597
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.740860128163145e-05,
+      "loss": 0.0548,
+      "step": 11598
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7393008099857995e-05,
+      "loss": 0.0562,
+      "step": 11599
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7377416686462106e-05,
+      "loss": 0.0533,
+      "step": 11600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.736182704196788e-05,
+      "loss": 0.0523,
+      "step": 11601
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.734623916689941e-05,
+      "loss": 0.0495,
+      "step": 11602
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7330653061780586e-05,
+      "loss": 0.0497,
+      "step": 11603
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.731506872713542e-05,
+      "loss": 0.0568,
+      "step": 11604
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.729948616348772e-05,
+      "loss": 0.0633,
+      "step": 11605
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.728390537136129e-05,
+      "loss": 0.0444,
+      "step": 11606
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.726832635127991e-05,
+      "loss": 0.0591,
+      "step": 11607
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.725274910376728e-05,
+      "loss": 0.071,
+      "step": 11608
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.723717362934697e-05,
+      "loss": 0.057,
+      "step": 11609
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7221599928542595e-05,
+      "loss": 0.0637,
+      "step": 11610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.720602800187762e-05,
+      "loss": 0.0516,
+      "step": 11611
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.719045784987558e-05,
+      "loss": 0.0636,
+      "step": 11612
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.717488947305976e-05,
+      "loss": 0.0597,
+      "step": 11613
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7159322871953525e-05,
+      "loss": 0.0502,
+      "step": 11614
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7143758047080164e-05,
+      "loss": 0.0535,
+      "step": 11615
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7128194998962907e-05,
+      "loss": 0.0619,
+      "step": 11616
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.711263372812488e-05,
+      "loss": 0.0542,
+      "step": 11617
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.7097074235089125e-05,
+      "loss": 0.0521,
+      "step": 11618
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.708151652037871e-05,
+      "loss": 0.0531,
+      "step": 11619
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.70659605845166e-05,
+      "loss": 0.0576,
+      "step": 11620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.705040642802573e-05,
+      "loss": 0.0474,
+      "step": 11621
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.703485405142897e-05,
+      "loss": 0.0591,
+      "step": 11622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.701930345524903e-05,
+      "loss": 0.0572,
+      "step": 11623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.70037546400087e-05,
+      "loss": 0.0511,
+      "step": 11624
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.698820760623064e-05,
+      "loss": 0.0582,
+      "step": 11625
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.697266235443749e-05,
+      "loss": 0.0498,
+      "step": 11626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.695711888515174e-05,
+      "loss": 0.0503,
+      "step": 11627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.694157719889593e-05,
+      "loss": 0.0501,
+      "step": 11628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.692603729619245e-05,
+      "loss": 0.0506,
+      "step": 11629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.691049917756375e-05,
+      "loss": 0.0732,
+      "step": 11630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.68949628435321e-05,
+      "loss": 0.0586,
+      "step": 11631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.687942829461969e-05,
+      "loss": 0.0487,
+      "step": 11632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6863895531348765e-05,
+      "loss": 0.0494,
+      "step": 11633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6848364554241466e-05,
+      "loss": 0.0517,
+      "step": 11634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.683283536381985e-05,
+      "loss": 0.0622,
+      "step": 11635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.681730796060597e-05,
+      "loss": 0.0536,
+      "step": 11636
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.680178234512171e-05,
+      "loss": 0.0507,
+      "step": 11637
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.678625851788899e-05,
+      "loss": 0.053,
+      "step": 11638
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6770736479429655e-05,
+      "loss": 0.0603,
+      "step": 11639
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.67552162302655e-05,
+      "loss": 0.0515,
+      "step": 11640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6739697770918154e-05,
+      "loss": 0.0471,
+      "step": 11641
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6724181101909357e-05,
+      "loss": 0.0561,
+      "step": 11642
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.6708666223760635e-05,
+      "loss": 0.048,
+      "step": 11643
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.669315313699354e-05,
+      "loss": 0.0488,
+      "step": 11644
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.667764184212958e-05,
+      "loss": 0.0501,
+      "step": 11645
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.666213233969009e-05,
+      "loss": 0.0511,
+      "step": 11646
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6646624630196464e-05,
+      "loss": 0.059,
+      "step": 11647
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.663111871417e-05,
+      "loss": 0.0508,
+      "step": 11648
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.661561459213191e-05,
+      "loss": 0.0757,
+      "step": 11649
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6600112264603403e-05,
+      "loss": 0.0604,
+      "step": 11650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.658461173210552e-05,
+      "loss": 0.0499,
+      "step": 11651
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.656911299515934e-05,
+      "loss": 0.0545,
+      "step": 11652
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6553616054285865e-05,
+      "loss": 0.0496,
+      "step": 11653
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.653812091000605e-05,
+      "loss": 0.0496,
+      "step": 11654
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6522627562840716e-05,
+      "loss": 0.0541,
+      "step": 11655
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6507136013310634e-05,
+      "loss": 0.056,
+      "step": 11656
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.649164626193662e-05,
+      "loss": 0.0508,
+      "step": 11657
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.647615830923933e-05,
+      "loss": 0.0465,
+      "step": 11658
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.646067215573942e-05,
+      "loss": 0.0503,
+      "step": 11659
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.64451878019574e-05,
+      "loss": 0.0487,
+      "step": 11660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.642970524841381e-05,
+      "loss": 0.0515,
+      "step": 11661
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.641422449562909e-05,
+      "loss": 0.0505,
+      "step": 11662
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6398745544123614e-05,
+      "loss": 0.0511,
+      "step": 11663
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.6383268394417754e-05,
+      "loss": 0.054,
+      "step": 11664
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.63677930470317e-05,
+      "loss": 0.0562,
+      "step": 11665
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6352319502485686e-05,
+      "loss": 0.0518,
+      "step": 11666
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.633684776129985e-05,
+      "loss": 0.0471,
+      "step": 11667
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6321377823994326e-05,
+      "loss": 0.044,
+      "step": 11668
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.630590969108908e-05,
+      "loss": 0.0437,
+      "step": 11669
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.629044336310403e-05,
+      "loss": 0.0481,
+      "step": 11670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.627497884055912e-05,
+      "loss": 0.0444,
+      "step": 11671
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.62595161239742e-05,
+      "loss": 0.044,
+      "step": 11672
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.624405521386907e-05,
+      "loss": 0.0443,
+      "step": 11673
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.622859611076337e-05,
+      "loss": 0.0467,
+      "step": 11674
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.62131388151768e-05,
+      "loss": 0.046,
+      "step": 11675
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.619768332762896e-05,
+      "loss": 0.0471,
+      "step": 11676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6182229648639365e-05,
+      "loss": 0.0493,
+      "step": 11677
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.616677777872756e-05,
+      "loss": 0.0621,
+      "step": 11678
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.615132771841284e-05,
+      "loss": 0.0449,
+      "step": 11679
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.613587946821463e-05,
+      "loss": 0.0507,
+      "step": 11680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.6120433028652235e-05,
+      "loss": 0.0502,
+      "step": 11681
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.610498840024482e-05,
+      "loss": 0.0441,
+      "step": 11682
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.608954558351164e-05,
+      "loss": 0.0467,
+      "step": 11683
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.607410457897171e-05,
+      "loss": 0.0471,
+      "step": 11684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.605866538714412e-05,
+      "loss": 0.0456,
+      "step": 11685
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.604322800854787e-05,
+      "loss": 0.0472,
+      "step": 11686
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.60277924437019e-05,
+      "loss": 0.045,
+      "step": 11687
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.601235869312501e-05,
+      "loss": 0.0473,
+      "step": 11688
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.599692675733606e-05,
+      "loss": 0.0514,
+      "step": 11689
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.598149663685376e-05,
+      "loss": 0.0485,
+      "step": 11690
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.596606833219681e-05,
+      "loss": 0.0516,
+      "step": 11691
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.595064184388387e-05,
+      "loss": 0.0458,
+      "step": 11692
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.593521717243341e-05,
+      "loss": 0.0492,
+      "step": 11693
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.591979431836402e-05,
+      "loss": 0.0484,
+      "step": 11694
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5904373282194054e-05,
+      "loss": 0.0487,
+      "step": 11695
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.588895406444192e-05,
+      "loss": 0.0442,
+      "step": 11696
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5873536665625994e-05,
+      "loss": 0.0448,
+      "step": 11697
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.585812108626443e-05,
+      "loss": 0.0465,
+      "step": 11698
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.584270732687547e-05,
+      "loss": 0.0493,
+      "step": 11699
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5827295387977235e-05,
+      "loss": 0.0486,
+      "step": 11700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5811885270087805e-05,
+      "loss": 0.0514,
+      "step": 11701
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.579647697372522e-05,
+      "loss": 0.0455,
+      "step": 11702
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.578107049940737e-05,
+      "loss": 0.0443,
+      "step": 11703
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.576566584765216e-05,
+      "loss": 0.0458,
+      "step": 11704
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.575026301897742e-05,
+      "loss": 0.0442,
+      "step": 11705
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.573486201390096e-05,
+      "loss": 0.0487,
+      "step": 11706
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.571946283294044e-05,
+      "loss": 0.06,
+      "step": 11707
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5704065476613465e-05,
+      "loss": 0.0436,
+      "step": 11708
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.568866994543765e-05,
+      "loss": 0.0437,
+      "step": 11709
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.567327623993051e-05,
+      "loss": 0.0458,
+      "step": 11710
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.565788436060955e-05,
+      "loss": 0.0469,
+      "step": 11711
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.564249430799208e-05,
+      "loss": 0.0446,
+      "step": 11712
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.562710608259548e-05,
+      "loss": 0.053,
+      "step": 11713
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5611719684937026e-05,
+      "loss": 0.0443,
+      "step": 11714
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.559633511553393e-05,
+      "loss": 0.0444,
+      "step": 11715
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.558095237490336e-05,
+      "loss": 0.0437,
+      "step": 11716
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5565571463562365e-05,
+      "loss": 0.0485,
+      "step": 11717
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.555019238202798e-05,
+      "loss": 0.048,
+      "step": 11718
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5534815130817175e-05,
+      "loss": 0.045,
+      "step": 11719
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.551943971044691e-05,
+      "loss": 0.0449,
+      "step": 11720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5504066121433974e-05,
+      "loss": 0.0588,
+      "step": 11721
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5488694364295115e-05,
+      "loss": 0.0478,
+      "step": 11722
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.54733244395471e-05,
+      "loss": 0.0513,
+      "step": 11723
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.545795634770657e-05,
+      "loss": 0.0444,
+      "step": 11724
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.544259008929016e-05,
+      "loss": 0.0454,
+      "step": 11725
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.542722566481434e-05,
+      "loss": 0.0544,
+      "step": 11726
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.541186307479562e-05,
+      "loss": 0.0447,
+      "step": 11727
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5396502319750415e-05,
+      "loss": 0.0481,
+      "step": 11728
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.538114340019507e-05,
+      "loss": 0.0538,
+      "step": 11729
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.53657863166459e-05,
+      "loss": 0.0447,
+      "step": 11730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.535043106961907e-05,
+      "loss": 0.0451,
+      "step": 11731
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.533507765963077e-05,
+      "loss": 0.046,
+      "step": 11732
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.531972608719717e-05,
+      "loss": 0.0494,
+      "step": 11733
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.53043763528342e-05,
+      "loss": 0.048,
+      "step": 11734
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.528902845705794e-05,
+      "loss": 0.0467,
+      "step": 11735
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.527368240038421e-05,
+      "loss": 0.0437,
+      "step": 11736
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.525833818332892e-05,
+      "loss": 0.0436,
+      "step": 11737
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.524299580640785e-05,
+      "loss": 0.0452,
+      "step": 11738
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5227655270136795e-05,
+      "loss": 0.046,
+      "step": 11739
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.521231657503132e-05,
+      "loss": 0.045,
+      "step": 11740
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5196979721607104e-05,
+      "loss": 0.0528,
+      "step": 11741
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.518164471037965e-05,
+      "loss": 0.0442,
+      "step": 11742
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.516631154186449e-05,
+      "loss": 0.0483,
+      "step": 11743
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5150980216577055e-05,
+      "loss": 0.045,
+      "step": 11744
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.513565073503263e-05,
+      "loss": 0.0508,
+      "step": 11745
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.51203230977466e-05,
+      "loss": 0.0456,
+      "step": 11746
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.510499730523412e-05,
+      "loss": 0.0444,
+      "step": 11747
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.508967335801041e-05,
+      "loss": 0.0423,
+      "step": 11748
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.507435125659061e-05,
+      "loss": 0.0466,
+      "step": 11749
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.5059031001489695e-05,
+      "loss": 0.0444,
+      "step": 11750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.50437125932227e-05,
+      "loss": 0.0619,
+      "step": 11751
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.5028396032304556e-05,
+      "loss": 0.0466,
+      "step": 11752
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.501308131925014e-05,
+      "loss": 0.0435,
+      "step": 11753
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.499776845457421e-05,
+      "loss": 0.0463,
+      "step": 11754
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4982457438791514e-05,
+      "loss": 0.0453,
+      "step": 11755
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4967148272416756e-05,
+      "loss": 0.0455,
+      "step": 11756
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4951840955964544e-05,
+      "loss": 0.0437,
+      "step": 11757
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.493653548994945e-05,
+      "loss": 0.0439,
+      "step": 11758
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.492123187488595e-05,
+      "loss": 0.0443,
+      "step": 11759
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.490593011128844e-05,
+      "loss": 0.0434,
+      "step": 11760
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.48906301996713e-05,
+      "loss": 0.0461,
+      "step": 11761
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4875332140548856e-05,
+      "loss": 0.0448,
+      "step": 11762
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.486003593443537e-05,
+      "loss": 0.0453,
+      "step": 11763
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4844741581844977e-05,
+      "loss": 0.0523,
+      "step": 11764
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4829449083291794e-05,
+      "loss": 0.0473,
+      "step": 11765
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.48141584392899e-05,
+      "loss": 0.0509,
+      "step": 11766
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.479886965035332e-05,
+      "loss": 0.0457,
+      "step": 11767
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.478358271699591e-05,
+      "loss": 0.0481,
+      "step": 11768
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.476829763973157e-05,
+      "loss": 0.0464,
+      "step": 11769
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4753014419074126e-05,
+      "loss": 0.0441,
+      "step": 11770
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4737733055537335e-05,
+      "loss": 0.0437,
+      "step": 11771
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.472245354963482e-05,
+      "loss": 0.0471,
+      "step": 11772
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.470717590188026e-05,
+      "loss": 0.0437,
+      "step": 11773
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.469190011278716e-05,
+      "loss": 0.0465,
+      "step": 11774
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.467662618286902e-05,
+      "loss": 0.0565,
+      "step": 11775
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.466135411263929e-05,
+      "loss": 0.0447,
+      "step": 11776
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4646083902611366e-05,
+      "loss": 0.0429,
+      "step": 11777
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4630815553298497e-05,
+      "loss": 0.0472,
+      "step": 11778
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4615549065213955e-05,
+      "loss": 0.0447,
+      "step": 11779
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.460028443887092e-05,
+      "loss": 0.0517,
+      "step": 11780
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4585021674782534e-05,
+      "loss": 0.106,
+      "step": 11781
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.456976077346181e-05,
+      "loss": 0.0489,
+      "step": 11782
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.455450173542175e-05,
+      "loss": 0.0448,
+      "step": 11783
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.453924456117533e-05,
+      "loss": 0.0436,
+      "step": 11784
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.452398925123535e-05,
+      "loss": 0.0462,
+      "step": 11785
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4508735806114654e-05,
+      "loss": 0.0423,
+      "step": 11786
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.449348422632601e-05,
+      "loss": 0.0467,
+      "step": 11787
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4478234512382024e-05,
+      "loss": 0.0476,
+      "step": 11788
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4462986664795367e-05,
+      "loss": 0.0436,
+      "step": 11789
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4447740684078574e-05,
+      "loss": 0.0446,
+      "step": 11790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4432496570744184e-05,
+      "loss": 0.0471,
+      "step": 11791
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4417254325304555e-05,
+      "loss": 0.0416,
+      "step": 11792
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4402013948272095e-05,
+      "loss": 0.0447,
+      "step": 11793
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.438677544015909e-05,
+      "loss": 0.0424,
+      "step": 11794
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.437153880147782e-05,
+      "loss": 0.0455,
+      "step": 11795
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4356304032740404e-05,
+      "loss": 0.0452,
+      "step": 11796
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.434107113445901e-05,
+      "loss": 0.0515,
+      "step": 11797
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4325840107145634e-05,
+      "loss": 0.0454,
+      "step": 11798
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4310610951312294e-05,
+      "loss": 0.051,
+      "step": 11799
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4295383667470924e-05,
+      "loss": 0.0433,
+      "step": 11800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.428015825613341e-05,
+      "loss": 0.0497,
+      "step": 11801
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.426493471781149e-05,
+      "loss": 0.044,
+      "step": 11802
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.424971305301694e-05,
+      "loss": 0.0531,
+      "step": 11803
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.423449326226142e-05,
+      "loss": 0.0477,
+      "step": 11804
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.421927534605659e-05,
+      "loss": 0.0462,
+      "step": 11805
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4204059304913925e-05,
+      "loss": 0.0471,
+      "step": 11806
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.4188845139344946e-05,
+      "loss": 0.0471,
+      "step": 11807
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.417363284986107e-05,
+      "loss": 0.0439,
+      "step": 11808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.415842243697369e-05,
+      "loss": 0.0478,
+      "step": 11809
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.414321390119408e-05,
+      "loss": 0.0483,
+      "step": 11810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.412800724303343e-05,
+      "loss": 0.0479,
+      "step": 11811
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.411280246300294e-05,
+      "loss": 0.0432,
+      "step": 11812
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.4097599561613723e-05,
+      "loss": 0.0483,
+      "step": 11813
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.408239853937682e-05,
+      "loss": 0.0451,
+      "step": 11814
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.406719939680326e-05,
+      "loss": 0.0487,
+      "step": 11815
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.405200213440387e-05,
+      "loss": 0.0451,
+      "step": 11816
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.403680675268954e-05,
+      "loss": 0.0451,
+      "step": 11817
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.402161325217108e-05,
+      "loss": 0.0453,
+      "step": 11818
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.400642163335924e-05,
+      "loss": 0.0476,
+      "step": 11819
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.399123189676461e-05,
+      "loss": 0.0457,
+      "step": 11820
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3976044042897835e-05,
+      "loss": 0.0482,
+      "step": 11821
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.396085807226944e-05,
+      "loss": 0.0467,
+      "step": 11822
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.394567398538996e-05,
+      "loss": 0.0464,
+      "step": 11823
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.393049178276975e-05,
+      "loss": 0.0425,
+      "step": 11824
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.391531146491912e-05,
+      "loss": 0.0463,
+      "step": 11825
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.390013303234839e-05,
+      "loss": 0.0448,
+      "step": 11826
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.388495648556781e-05,
+      "loss": 0.0432,
+      "step": 11827
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3869781825087495e-05,
+      "loss": 0.0436,
+      "step": 11828
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3854609051417614e-05,
+      "loss": 0.0506,
+      "step": 11829
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.38394381650681e-05,
+      "loss": 0.0522,
+      "step": 11830
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.382426916654897e-05,
+      "loss": 0.0501,
+      "step": 11831
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.380910205637012e-05,
+      "loss": 0.045,
+      "step": 11832
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.379393683504143e-05,
+      "loss": 0.0451,
+      "step": 11833
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.37787735030726e-05,
+      "loss": 0.052,
+      "step": 11834
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.376361206097339e-05,
+      "loss": 0.0554,
+      "step": 11835
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.374845250925348e-05,
+      "loss": 0.0468,
+      "step": 11836
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3733294848422393e-05,
+      "loss": 0.0486,
+      "step": 11837
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.371813907898966e-05,
+      "loss": 0.0519,
+      "step": 11838
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3702985201464805e-05,
+      "loss": 0.0494,
+      "step": 11839
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.368783321635714e-05,
+      "loss": 0.0482,
+      "step": 11840
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.367268312417605e-05,
+      "loss": 0.0467,
+      "step": 11841
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.365753492543078e-05,
+      "loss": 0.0484,
+      "step": 11842
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.364238862063058e-05,
+      "loss": 0.0452,
+      "step": 11843
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.362724421028451e-05,
+      "loss": 0.0423,
+      "step": 11844
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.361210169490171e-05,
+      "loss": 0.0447,
+      "step": 11845
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3596961074991164e-05,
+      "loss": 0.048,
+      "step": 11846
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3581822351061875e-05,
+      "loss": 0.0457,
+      "step": 11847
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.356668552362265e-05,
+      "loss": 0.0457,
+      "step": 11848
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3551550593182385e-05,
+      "loss": 0.0461,
+      "step": 11849
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3536417560249775e-05,
+      "loss": 0.0466,
+      "step": 11850
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3521286425333544e-05,
+      "loss": 0.0433,
+      "step": 11851
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.350615718894232e-05,
+      "loss": 0.047,
+      "step": 11852
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.349102985158472e-05,
+      "loss": 0.061,
+      "step": 11853
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.347590441376915e-05,
+      "loss": 0.0459,
+      "step": 11854
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.346078087600412e-05,
+      "loss": 0.0437,
+      "step": 11855
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3445659238797966e-05,
+      "loss": 0.0484,
+      "step": 11856
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.343053950265906e-05,
+      "loss": 0.0455,
+      "step": 11857
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.341542166809559e-05,
+      "loss": 0.0741,
+      "step": 11858
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.340030573561576e-05,
+      "loss": 0.0445,
+      "step": 11859
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.338519170572768e-05,
+      "loss": 0.047,
+      "step": 11860
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.337007957893946e-05,
+      "loss": 0.045,
+      "step": 11861
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3354969355759045e-05,
+      "loss": 0.0447,
+      "step": 11862
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.333986103669434e-05,
+      "loss": 0.0502,
+      "step": 11863
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3324754622253235e-05,
+      "loss": 0.0442,
+      "step": 11864
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3309650112943535e-05,
+      "loss": 0.0432,
+      "step": 11865
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3294547509272975e-05,
+      "loss": 0.0446,
+      "step": 11866
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.327944681174926e-05,
+      "loss": 0.0454,
+      "step": 11867
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.326434802087993e-05,
+      "loss": 0.0484,
+      "step": 11868
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.324925113717255e-05,
+      "loss": 0.0456,
+      "step": 11869
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.323415616113463e-05,
+      "loss": 0.0438,
+      "step": 11870
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.32190630932736e-05,
+      "loss": 0.0496,
+      "step": 11871
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3203971934096756e-05,
+      "loss": 0.042,
+      "step": 11872
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.31888826841114e-05,
+      "loss": 0.0451,
+      "step": 11873
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3173795343824764e-05,
+      "loss": 0.0443,
+      "step": 11874
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3158709913744044e-05,
+      "loss": 0.0454,
+      "step": 11875
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.31436263943763e-05,
+      "loss": 0.0478,
+      "step": 11876
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.312854478622853e-05,
+      "loss": 0.048,
+      "step": 11877
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.311346508980772e-05,
+      "loss": 0.0441,
+      "step": 11878
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.30983873056208e-05,
+      "loss": 0.0455,
+      "step": 11879
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.308331143417459e-05,
+      "loss": 0.0488,
+      "step": 11880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3068237475975904e-05,
+      "loss": 0.0481,
+      "step": 11881
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.305316543153137e-05,
+      "loss": 0.043,
+      "step": 11882
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.303809530134768e-05,
+      "loss": 0.0457,
+      "step": 11883
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.302302708593141e-05,
+      "loss": 0.0459,
+      "step": 11884
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3007960785789114e-05,
+      "loss": 0.0535,
+      "step": 11885
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2992896401427164e-05,
+      "loss": 0.0424,
+      "step": 11886
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2977833933352e-05,
+      "loss": 0.0424,
+      "step": 11887
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2962773382069967e-05,
+      "loss": 0.0462,
+      "step": 11888
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2947714748087255e-05,
+      "loss": 0.0706,
+      "step": 11889
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.293265803191013e-05,
+      "loss": 0.0463,
+      "step": 11890
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2917603234044655e-05,
+      "loss": 0.0459,
+      "step": 11891
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.290255035499693e-05,
+      "loss": 0.0479,
+      "step": 11892
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.288749939527294e-05,
+      "loss": 0.045,
+      "step": 11893
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2872450355378646e-05,
+      "loss": 0.0452,
+      "step": 11894
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.285740323581995e-05,
+      "loss": 0.0482,
+      "step": 11895
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.284235803710256e-05,
+      "loss": 0.0454,
+      "step": 11896
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2827314759732284e-05,
+      "loss": 0.053,
+      "step": 11897
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.28122734042148e-05,
+      "loss": 0.0469,
+      "step": 11898
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2797233971055736e-05,
+      "loss": 0.0477,
+      "step": 11899
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2782196460760615e-05,
+      "loss": 0.0474,
+      "step": 11900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.27671608738349e-05,
+      "loss": 0.0456,
+      "step": 11901
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.275212721078402e-05,
+      "loss": 0.0446,
+      "step": 11902
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.273709547211335e-05,
+      "loss": 0.0473,
+      "step": 11903
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2722065658328216e-05,
+      "loss": 0.0445,
+      "step": 11904
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2707037769933756e-05,
+      "loss": 0.0492,
+      "step": 11905
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2692011807435174e-05,
+      "loss": 0.0508,
+      "step": 11906
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.267698777133756e-05,
+      "loss": 0.0475,
+      "step": 11907
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.266196566214597e-05,
+      "loss": 0.0449,
+      "step": 11908
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.264694548036538e-05,
+      "loss": 0.0466,
+      "step": 11909
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.263192722650062e-05,
+      "loss": 0.0493,
+      "step": 11910
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.261691090105658e-05,
+      "loss": 0.0464,
+      "step": 11911
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.260189650453802e-05,
+      "loss": 0.0468,
+      "step": 11912
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.258688403744969e-05,
+      "loss": 0.0431,
+      "step": 11913
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2571873500296186e-05,
+      "loss": 0.0442,
+      "step": 11914
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.255686489358206e-05,
+      "loss": 0.0449,
+      "step": 11915
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2541858217811844e-05,
+      "loss": 0.0486,
+      "step": 11916
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2526853473490014e-05,
+      "loss": 0.0446,
+      "step": 11917
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2511850661120964e-05,
+      "loss": 0.044,
+      "step": 11918
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.249684978120895e-05,
+      "loss": 0.0488,
+      "step": 11919
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.248185083425826e-05,
+      "loss": 0.0474,
+      "step": 11920
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.246685382077308e-05,
+      "loss": 0.0475,
+      "step": 11921
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.2451858741257524e-05,
+      "loss": 0.0464,
+      "step": 11922
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.24368655962157e-05,
+      "loss": 0.0457,
+      "step": 11923
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.242187438615153e-05,
+      "loss": 0.0458,
+      "step": 11924
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.240688511156896e-05,
+      "loss": 0.0497,
+      "step": 11925
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.239189777297191e-05,
+      "loss": 0.0485,
+      "step": 11926
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2376912370864096e-05,
+      "loss": 0.0448,
+      "step": 11927
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2361928905749314e-05,
+      "loss": 0.0482,
+      "step": 11928
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.234694737813116e-05,
+      "loss": 0.0441,
+      "step": 11929
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2331967788513295e-05,
+      "loss": 0.0433,
+      "step": 11930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2316990137399235e-05,
+      "loss": 0.0435,
+      "step": 11931
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.230201442529249e-05,
+      "loss": 0.0459,
+      "step": 11932
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.228704065269641e-05,
+      "loss": 0.0492,
+      "step": 11933
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.227206882011435e-05,
+      "loss": 0.0499,
+      "step": 11934
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2257098928049596e-05,
+      "loss": 0.0444,
+      "step": 11935
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2242130977005364e-05,
+      "loss": 0.0437,
+      "step": 11936
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2227164967484837e-05,
+      "loss": 0.0467,
+      "step": 11937
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.221220089999101e-05,
+      "loss": 0.0429,
+      "step": 11938
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.219723877502697e-05,
+      "loss": 0.0456,
+      "step": 11939
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.218227859309561e-05,
+      "loss": 0.0447,
+      "step": 11940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.216732035469986e-05,
+      "loss": 0.0566,
+      "step": 11941
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.215236406034254e-05,
+      "loss": 0.0425,
+      "step": 11942
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.213740971052636e-05,
+      "loss": 0.0428,
+      "step": 11943
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2122457305754036e-05,
+      "loss": 0.0425,
+      "step": 11944
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.210750684652819e-05,
+      "loss": 0.0437,
+      "step": 11945
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2092558333351384e-05,
+      "loss": 0.0447,
+      "step": 11946
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.207761176672614e-05,
+      "loss": 0.0464,
+      "step": 11947
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.206266714715482e-05,
+      "loss": 0.045,
+      "step": 11948
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.204772447513982e-05,
+      "loss": 0.0471,
+      "step": 11949
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2032783751183426e-05,
+      "loss": 0.0435,
+      "step": 11950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.201784497578792e-05,
+      "loss": 0.0443,
+      "step": 11951
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.2002908149455426e-05,
+      "loss": 0.0463,
+      "step": 11952
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1987973272688e-05,
+      "loss": 0.043,
+      "step": 11953
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1973040345987726e-05,
+      "loss": 0.0441,
+      "step": 11954
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1958109369856556e-05,
+      "loss": 0.047,
+      "step": 11955
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.194318034479644e-05,
+      "loss": 0.048,
+      "step": 11956
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.192825327130914e-05,
+      "loss": 0.0433,
+      "step": 11957
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.191332814989646e-05,
+      "loss": 0.0437,
+      "step": 11958
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.189840498106011e-05,
+      "loss": 0.0455,
+      "step": 11959
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1883483765301735e-05,
+      "loss": 0.0472,
+      "step": 11960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.186856450312294e-05,
+      "loss": 0.0424,
+      "step": 11961
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.185364719502517e-05,
+      "loss": 0.0476,
+      "step": 11962
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1838731841509894e-05,
+      "loss": 0.0499,
+      "step": 11963
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1823818443078496e-05,
+      "loss": 0.0471,
+      "step": 11964
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.180890700023232e-05,
+      "loss": 0.0448,
+      "step": 11965
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1793997513472584e-05,
+      "loss": 0.0432,
+      "step": 11966
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1779089983300414e-05,
+      "loss": 0.0437,
+      "step": 11967
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.176418441021699e-05,
+      "loss": 0.0493,
+      "step": 11968
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.174928079472333e-05,
+      "loss": 0.0444,
+      "step": 11969
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.173437913732048e-05,
+      "loss": 0.0513,
+      "step": 11970
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.171947943850927e-05,
+      "loss": 0.046,
+      "step": 11971
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.170458169879058e-05,
+      "loss": 0.0473,
+      "step": 11972
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.168968591866521e-05,
+      "loss": 0.0464,
+      "step": 11973
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.1674792098633884e-05,
+      "loss": 0.0431,
+      "step": 11974
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.165990023919729e-05,
+      "loss": 0.0461,
+      "step": 11975
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.164501034085593e-05,
+      "loss": 0.0474,
+      "step": 11976
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.163012240411037e-05,
+      "loss": 0.0426,
+      "step": 11977
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.16152364294611e-05,
+      "loss": 0.0464,
+      "step": 11978
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.160035241740845e-05,
+      "loss": 0.0436,
+      "step": 11979
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1585470368452794e-05,
+      "loss": 0.0441,
+      "step": 11980
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1570590283094346e-05,
+      "loss": 0.0462,
+      "step": 11981
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.155571216183331e-05,
+      "loss": 0.0437,
+      "step": 11982
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.154083600516983e-05,
+      "loss": 0.0563,
+      "step": 11983
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.152596181360399e-05,
+      "loss": 0.0452,
+      "step": 11984
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.151108958763571e-05,
+      "loss": 0.0441,
+      "step": 11985
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1496219327764965e-05,
+      "loss": 0.0458,
+      "step": 11986
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1481351034491614e-05,
+      "loss": 0.0443,
+      "step": 11987
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.146648470831544e-05,
+      "loss": 0.0419,
+      "step": 11988
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1451620349736234e-05,
+      "loss": 0.0469,
+      "step": 11989
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.143675795925357e-05,
+      "loss": 0.0421,
+      "step": 11990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1421897537367116e-05,
+      "loss": 0.0469,
+      "step": 11991
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.140703908457634e-05,
+      "loss": 0.0426,
+      "step": 11992
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.139218260138074e-05,
+      "loss": 0.0439,
+      "step": 11993
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.137732808827974e-05,
+      "loss": 0.0457,
+      "step": 11994
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1362475545772616e-05,
+      "loss": 0.044,
+      "step": 11995
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1347624974358665e-05,
+      "loss": 0.0471,
+      "step": 11996
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.13327763745371e-05,
+      "loss": 0.0453,
+      "step": 11997
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.131792974680706e-05,
+      "loss": 0.0471,
+      "step": 11998
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.130308509166756e-05,
+      "loss": 0.0453,
+      "step": 11999
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1288242409617636e-05,
+      "loss": 0.053,
+      "step": 12000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1273401701156235e-05,
+      "loss": 0.0481,
+      "step": 12001
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.125856296678221e-05,
+      "loss": 0.0456,
+      "step": 12002
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.124372620699439e-05,
+      "loss": 0.0433,
+      "step": 12003
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.122889142229149e-05,
+      "loss": 0.0442,
+      "step": 12004
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.121405861317216e-05,
+      "loss": 0.0457,
+      "step": 12005
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1199227780134994e-05,
+      "loss": 0.0458,
+      "step": 12006
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1184398923678566e-05,
+      "loss": 0.0445,
+      "step": 12007
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.116957204430137e-05,
+      "loss": 0.0533,
+      "step": 12008
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.115474714250175e-05,
+      "loss": 0.0432,
+      "step": 12009
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.113992421877806e-05,
+      "loss": 0.0547,
+      "step": 12010
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.112510327362857e-05,
+      "loss": 0.0444,
+      "step": 12011
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1110284307551525e-05,
+      "loss": 0.0507,
+      "step": 12012
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.109546732104501e-05,
+      "loss": 0.0428,
+      "step": 12013
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1080652314607115e-05,
+      "loss": 0.0476,
+      "step": 12014
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.106583928873584e-05,
+      "loss": 0.0449,
+      "step": 12015
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1051028243929125e-05,
+      "loss": 0.0432,
+      "step": 12016
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.1036219180684885e-05,
+      "loss": 0.0435,
+      "step": 12017
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.102141209950089e-05,
+      "loss": 0.0464,
+      "step": 12018
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.100660700087483e-05,
+      "loss": 0.0515,
+      "step": 12019
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.099180388530443e-05,
+      "loss": 0.0463,
+      "step": 12020
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.097700275328729e-05,
+      "loss": 0.0463,
+      "step": 12021
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0962203605320956e-05,
+      "loss": 0.0462,
+      "step": 12022
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.094740644190287e-05,
+      "loss": 0.0451,
+      "step": 12023
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0932611263530454e-05,
+      "loss": 0.0468,
+      "step": 12024
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.091781807070103e-05,
+      "loss": 0.0472,
+      "step": 12025
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.090302686391193e-05,
+      "loss": 0.045,
+      "step": 12026
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.088823764366029e-05,
+      "loss": 0.0593,
+      "step": 12027
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0873450410443256e-05,
+      "loss": 0.0457,
+      "step": 12028
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.085866516475792e-05,
+      "loss": 0.0466,
+      "step": 12029
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.084388190710131e-05,
+      "loss": 0.0431,
+      "step": 12030
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0829100637970307e-05,
+      "loss": 0.0449,
+      "step": 12031
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.081432135786185e-05,
+      "loss": 0.045,
+      "step": 12032
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.079954406727265e-05,
+      "loss": 0.0475,
+      "step": 12033
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.078476876669951e-05,
+      "loss": 0.044,
+      "step": 12034
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.076999545663908e-05,
+      "loss": 0.0442,
+      "step": 12035
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0755224137588e-05,
+      "loss": 0.0472,
+      "step": 12036
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.0740454810042735e-05,
+      "loss": 0.0458,
+      "step": 12037
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.072568747449981e-05,
+      "loss": 0.0441,
+      "step": 12038
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.07109221314556e-05,
+      "loss": 0.0508,
+      "step": 12039
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.069615878140648e-05,
+      "loss": 0.0452,
+      "step": 12040
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0681397424848665e-05,
+      "loss": 0.0444,
+      "step": 12041
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.066663806227841e-05,
+      "loss": 0.043,
+      "step": 12042
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.065188069419178e-05,
+      "loss": 0.0454,
+      "step": 12043
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.063712532108489e-05,
+      "loss": 0.0458,
+      "step": 12044
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.062237194345373e-05,
+      "loss": 0.0477,
+      "step": 12045
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.060762056179426e-05,
+      "loss": 0.048,
+      "step": 12046
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.05928711766023e-05,
+      "loss": 0.0457,
+      "step": 12047
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.057812378837366e-05,
+      "loss": 0.0428,
+      "step": 12048
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.056337839760408e-05,
+      "loss": 0.045,
+      "step": 12049
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.054863500478926e-05,
+      "loss": 0.044,
+      "step": 12050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.053389361042473e-05,
+      "loss": 0.0454,
+      "step": 12051
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.051915421500605e-05,
+      "loss": 0.0446,
+      "step": 12052
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.050441681902869e-05,
+      "loss": 0.0493,
+      "step": 12053
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.048968142298808e-05,
+      "loss": 0.0436,
+      "step": 12054
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.04749480273795e-05,
+      "loss": 0.0461,
+      "step": 12055
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.046021663269819e-05,
+      "loss": 0.0482,
+      "step": 12056
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0445487239439384e-05,
+      "loss": 0.0438,
+      "step": 12057
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.043075984809819e-05,
+      "loss": 0.0431,
+      "step": 12058
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.041603445916968e-05,
+      "loss": 0.0459,
+      "step": 12059
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.040131107314887e-05,
+      "loss": 0.0483,
+      "step": 12060
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.038658969053064e-05,
+      "loss": 0.042,
+      "step": 12061
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.037187031180985e-05,
+      "loss": 0.0448,
+      "step": 12062
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0357152937481323e-05,
+      "loss": 0.0449,
+      "step": 12063
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.034243756803979e-05,
+      "loss": 0.0442,
+      "step": 12064
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.032772420397984e-05,
+      "loss": 0.0454,
+      "step": 12065
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0313012845796114e-05,
+      "loss": 0.046,
+      "step": 12066
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.029830349398311e-05,
+      "loss": 0.0477,
+      "step": 12067
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.028359614903534e-05,
+      "loss": 0.0427,
+      "step": 12068
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0268890811447124e-05,
+      "loss": 0.0471,
+      "step": 12069
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0254187481712776e-05,
+      "loss": 0.0441,
+      "step": 12070
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.023948616032657e-05,
+      "loss": 0.044,
+      "step": 12071
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.022478684778267e-05,
+      "loss": 0.0444,
+      "step": 12072
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.021008954457522e-05,
+      "loss": 0.0449,
+      "step": 12073
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0195394251198295e-05,
+      "loss": 0.0442,
+      "step": 12074
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.018070096814579e-05,
+      "loss": 0.0435,
+      "step": 12075
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.016600969591168e-05,
+      "loss": 0.0436,
+      "step": 12076
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.015132043498978e-05,
+      "loss": 0.0431,
+      "step": 12077
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.013663318587393e-05,
+      "loss": 0.0427,
+      "step": 12078
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.012194794905775e-05,
+      "loss": 0.0468,
+      "step": 12079
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.010726472503493e-05,
+      "loss": 0.0432,
+      "step": 12080
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0092583514299076e-05,
+      "loss": 0.0478,
+      "step": 12081
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.007790431734362e-05,
+      "loss": 0.0471,
+      "step": 12082
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0063227134662043e-05,
+      "loss": 0.0427,
+      "step": 12083
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.0048551966747746e-05,
+      "loss": 0.0445,
+      "step": 12084
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.003387881409397e-05,
+      "loss": 0.0547,
+      "step": 12085
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.001920767719398e-05,
+      "loss": 0.0434,
+      "step": 12086
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.000453855654096e-05,
+      "loss": 0.0436,
+      "step": 12087
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9989871452628016e-05,
+      "loss": 0.0454,
+      "step": 12088
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9975206365948136e-05,
+      "loss": 0.0442,
+      "step": 12089
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.996054329699431e-05,
+      "loss": 0.0444,
+      "step": 12090
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.994588224625944e-05,
+      "loss": 0.0486,
+      "step": 12091
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.993122321423638e-05,
+      "loss": 0.0429,
+      "step": 12092
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.991656620141784e-05,
+      "loss": 0.0474,
+      "step": 12093
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.9901911208296564e-05,
+      "loss": 0.0457,
+      "step": 12094
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.988725823536512e-05,
+      "loss": 0.0457,
+      "step": 12095
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9872607283116095e-05,
+      "loss": 0.0484,
+      "step": 12096
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.985795835204198e-05,
+      "loss": 0.0445,
+      "step": 12097
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9843311442635236e-05,
+      "loss": 0.0451,
+      "step": 12098
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.982866655538814e-05,
+      "loss": 0.0452,
+      "step": 12099
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.981402369079302e-05,
+      "loss": 0.043,
+      "step": 12100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9799382849342096e-05,
+      "loss": 0.0474,
+      "step": 12101
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.978474403152754e-05,
+      "loss": 0.0454,
+      "step": 12102
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.977010723784138e-05,
+      "loss": 0.0451,
+      "step": 12103
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.975547246877566e-05,
+      "loss": 0.0548,
+      "step": 12104
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.974083972482232e-05,
+      "loss": 0.0452,
+      "step": 12105
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.972620900647328e-05,
+      "loss": 0.0451,
+      "step": 12106
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9711580314220295e-05,
+      "loss": 0.0443,
+      "step": 12107
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.969695364855511e-05,
+      "loss": 0.048,
+      "step": 12108
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96823290099694e-05,
+      "loss": 0.0439,
+      "step": 12109
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9667706398954784e-05,
+      "loss": 0.044,
+      "step": 12110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.96530858160028e-05,
+      "loss": 0.0462,
+      "step": 12111
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.963846726160495e-05,
+      "loss": 0.0453,
+      "step": 12112
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.962385073625255e-05,
+      "loss": 0.0595,
+      "step": 12113
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9609236240436996e-05,
+      "loss": 0.0426,
+      "step": 12114
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.959462377464954e-05,
+      "loss": 0.0414,
+      "step": 12115
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.958001333938141e-05,
+      "loss": 0.0454,
+      "step": 12116
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.956540493512366e-05,
+      "loss": 0.0465,
+      "step": 12117
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9550798562367395e-05,
+      "loss": 0.0429,
+      "step": 12118
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.95361942216036e-05,
+      "loss": 0.0439,
+      "step": 12119
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.952159191332324e-05,
+      "loss": 0.0444,
+      "step": 12120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.950699163801712e-05,
+      "loss": 0.0481,
+      "step": 12121
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.949239339617601e-05,
+      "loss": 0.044,
+      "step": 12122
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9477797188290656e-05,
+      "loss": 0.0452,
+      "step": 12123
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.946320301485169e-05,
+      "loss": 0.0474,
+      "step": 12124
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.944861087634972e-05,
+      "loss": 0.0428,
+      "step": 12125
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9434020773275294e-05,
+      "loss": 0.0422,
+      "step": 12126
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.941943270611876e-05,
+      "loss": 0.0478,
+      "step": 12127
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.940484667537055e-05,
+      "loss": 0.05,
+      "step": 12128
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.939026268152097e-05,
+      "loss": 0.0477,
+      "step": 12129
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.937568072506028e-05,
+      "loss": 0.0435,
+      "step": 12130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.93611008064786e-05,
+      "loss": 0.0525,
+      "step": 12131
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.934652292626606e-05,
+      "loss": 0.0446,
+      "step": 12132
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.933194708491273e-05,
+      "loss": 0.0427,
+      "step": 12133
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.93173732829085e-05,
+      "loss": 0.0445,
+      "step": 12134
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9302801520743335e-05,
+      "loss": 0.0436,
+      "step": 12135
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9288231798906994e-05,
+      "loss": 0.0451,
+      "step": 12136
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9273664117889285e-05,
+      "loss": 0.0447,
+      "step": 12137
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.925909847817988e-05,
+      "loss": 0.05,
+      "step": 12138
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.924453488026841e-05,
+      "loss": 0.0456,
+      "step": 12139
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.922997332464447e-05,
+      "loss": 0.0467,
+      "step": 12140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.921541381179746e-05,
+      "loss": 0.0445,
+      "step": 12141
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.920085634221684e-05,
+      "loss": 0.0473,
+      "step": 12142
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9186300916391964e-05,
+      "loss": 0.0452,
+      "step": 12143
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.917174753481213e-05,
+      "loss": 0.0456,
+      "step": 12144
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.915719619796648e-05,
+      "loss": 0.0473,
+      "step": 12145
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.914264690634425e-05,
+      "loss": 0.0451,
+      "step": 12146
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9128099660434414e-05,
+      "loss": 0.0433,
+      "step": 12147
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.911355446072602e-05,
+      "loss": 0.0427,
+      "step": 12148
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9099011307708045e-05,
+      "loss": 0.0444,
+      "step": 12149
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.908447020186927e-05,
+      "loss": 0.0474,
+      "step": 12150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.9069931143698545e-05,
+      "loss": 0.0456,
+      "step": 12151
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.90553941336846e-05,
+      "loss": 0.0438,
+      "step": 12152
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.904085917231607e-05,
+      "loss": 0.0495,
+      "step": 12153
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.90263262600816e-05,
+      "loss": 0.0488,
+      "step": 12154
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.901179539746964e-05,
+      "loss": 0.043,
+      "step": 12155
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8997266584968686e-05,
+      "loss": 0.0469,
+      "step": 12156
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.898273982306711e-05,
+      "loss": 0.0471,
+      "step": 12157
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8968215112253257e-05,
+      "loss": 0.0455,
+      "step": 12158
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8953692453015356e-05,
+      "loss": 0.053,
+      "step": 12159
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8939171845841526e-05,
+      "loss": 0.0444,
+      "step": 12160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.892465329121994e-05,
+      "loss": 0.0452,
+      "step": 12161
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8910136789638615e-05,
+      "loss": 0.0481,
+      "step": 12162
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8895622341585566e-05,
+      "loss": 0.0458,
+      "step": 12163
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.888110994754862e-05,
+      "loss": 0.0481,
+      "step": 12164
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8866599608015654e-05,
+      "loss": 0.043,
+      "step": 12165
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.885209132347442e-05,
+      "loss": 0.0474,
+      "step": 12166
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.883758509441261e-05,
+      "loss": 0.0424,
+      "step": 12167
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.882308092131789e-05,
+      "loss": 0.0459,
+      "step": 12168
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.880857880467774e-05,
+      "loss": 0.0433,
+      "step": 12169
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.87940787449797e-05,
+      "loss": 0.0449,
+      "step": 12170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8779580742711206e-05,
+      "loss": 0.0439,
+      "step": 12171
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.876508479835954e-05,
+      "loss": 0.0462,
+      "step": 12172
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.875059091241205e-05,
+      "loss": 0.0546,
+      "step": 12173
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.873609908535589e-05,
+      "loss": 0.0464,
+      "step": 12174
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.872160931767822e-05,
+      "loss": 0.0449,
+      "step": 12175
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.870712160986611e-05,
+      "loss": 0.0449,
+      "step": 12176
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.869263596240661e-05,
+      "loss": 0.0474,
+      "step": 12177
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.867815237578658e-05,
+      "loss": 0.0451,
+      "step": 12178
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8663670850492906e-05,
+      "loss": 0.0463,
+      "step": 12179
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8649191387012394e-05,
+      "loss": 0.052,
+      "step": 12180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8634713985831784e-05,
+      "loss": 0.045,
+      "step": 12181
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8620238647437735e-05,
+      "loss": 0.0436,
+      "step": 12182
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.860576537231678e-05,
+      "loss": 0.045,
+      "step": 12183
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.859129416095551e-05,
+      "loss": 0.0421,
+      "step": 12184
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.85768250138403e-05,
+      "loss": 0.0446,
+      "step": 12185
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.856235793145756e-05,
+      "loss": 0.0502,
+      "step": 12186
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.854789291429365e-05,
+      "loss": 0.0455,
+      "step": 12187
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.853342996283471e-05,
+      "loss": 0.0453,
+      "step": 12188
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.851896907756697e-05,
+      "loss": 0.0455,
+      "step": 12189
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.850451025897652e-05,
+      "loss": 0.0454,
+      "step": 12190
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.84900535075494e-05,
+      "loss": 0.0476,
+      "step": 12191
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8475598823771596e-05,
+      "loss": 0.0461,
+      "step": 12192
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.846114620812894e-05,
+      "loss": 0.0454,
+      "step": 12193
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8446695661107276e-05,
+      "loss": 0.0446,
+      "step": 12194
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8432247183192386e-05,
+      "loss": 0.0442,
+      "step": 12195
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.841780077486995e-05,
+      "loss": 0.0424,
+      "step": 12196
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.840335643662558e-05,
+      "loss": 0.0419,
+      "step": 12197
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8388914168944766e-05,
+      "loss": 0.0424,
+      "step": 12198
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.837447397231303e-05,
+      "loss": 0.0534,
+      "step": 12199
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.836003584721577e-05,
+      "loss": 0.044,
+      "step": 12200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8345599794138354e-05,
+      "loss": 0.0459,
+      "step": 12201
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.833116581356599e-05,
+      "loss": 0.0439,
+      "step": 12202
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.83167339059839e-05,
+      "loss": 0.0445,
+      "step": 12203
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8302304071877215e-05,
+      "loss": 0.0423,
+      "step": 12204
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8287876311731e-05,
+      "loss": 0.0434,
+      "step": 12205
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.827345062603025e-05,
+      "loss": 0.0451,
+      "step": 12206
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.825902701525984e-05,
+      "loss": 0.0485,
+      "step": 12207
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.824460547990464e-05,
+      "loss": 0.0435,
+      "step": 12208
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.823018602044943e-05,
+      "loss": 0.0459,
+      "step": 12209
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.821576863737897e-05,
+      "loss": 0.0482,
+      "step": 12210
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8201353331177835e-05,
+      "loss": 0.0439,
+      "step": 12211
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8186940102330573e-05,
+      "loss": 0.0456,
+      "step": 12212
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8172528951321715e-05,
+      "loss": 0.0438,
+      "step": 12213
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.81581198786357e-05,
+      "loss": 0.0433,
+      "step": 12214
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.814371288475691e-05,
+      "loss": 0.0494,
+      "step": 12215
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.812930797016957e-05,
+      "loss": 0.0465,
+      "step": 12216
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8114905135357935e-05,
+      "loss": 0.043,
+      "step": 12217
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.810050438080615e-05,
+      "loss": 0.048,
+      "step": 12218
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.80861057069983e-05,
+      "loss": 0.0455,
+      "step": 12219
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.807170911441842e-05,
+      "loss": 0.0479,
+      "step": 12220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8057314603550396e-05,
+      "loss": 0.0502,
+      "step": 12221
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.8042922174878126e-05,
+      "loss": 0.0468,
+      "step": 12222
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.802853182888543e-05,
+      "loss": 0.0443,
+      "step": 12223
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.801414356605598e-05,
+      "loss": 0.0468,
+      "step": 12224
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7999757386873514e-05,
+      "loss": 0.0451,
+      "step": 12225
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.798537329182155e-05,
+      "loss": 0.0436,
+      "step": 12226
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7970991281383615e-05,
+      "loss": 0.0488,
+      "step": 12227
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.795661135604319e-05,
+      "loss": 0.0511,
+      "step": 12228
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7942233516283686e-05,
+      "loss": 0.0446,
+      "step": 12229
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7927857762588325e-05,
+      "loss": 0.0457,
+      "step": 12230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.79134840954404e-05,
+      "loss": 0.0559,
+      "step": 12231
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7899112515323065e-05,
+      "loss": 0.0487,
+      "step": 12232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.788474302271943e-05,
+      "loss": 0.0463,
+      "step": 12233
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.787037561811255e-05,
+      "loss": 0.0496,
+      "step": 12234
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.785601030198531e-05,
+      "loss": 0.0429,
+      "step": 12235
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7841647074820686e-05,
+      "loss": 0.0534,
+      "step": 12236
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.782728593710142e-05,
+      "loss": 0.0457,
+      "step": 12237
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.781292688931028e-05,
+      "loss": 0.0475,
+      "step": 12238
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.779856993192999e-05,
+      "loss": 0.0432,
+      "step": 12239
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.77842150654431e-05,
+      "loss": 0.0462,
+      "step": 12240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.776986229033216e-05,
+      "loss": 0.0432,
+      "step": 12241
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.775551160707964e-05,
+      "loss": 0.0433,
+      "step": 12242
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.774116301616798e-05,
+      "loss": 0.0453,
+      "step": 12243
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.772681651807943e-05,
+      "loss": 0.0454,
+      "step": 12244
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.771247211329629e-05,
+      "loss": 0.0489,
+      "step": 12245
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.769812980230074e-05,
+      "loss": 0.046,
+      "step": 12246
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.76837895855749e-05,
+      "loss": 0.0482,
+      "step": 12247
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.766945146360083e-05,
+      "loss": 0.0429,
+      "step": 12248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.76551154368605e-05,
+      "loss": 0.0443,
+      "step": 12249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7640781505835745e-05,
+      "loss": 0.0458,
+      "step": 12250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.762644967100846e-05,
+      "loss": 0.0486,
+      "step": 12251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.76121199328604e-05,
+      "loss": 0.0436,
+      "step": 12252
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.759779229187328e-05,
+      "loss": 0.044,
+      "step": 12253
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.758346674852867e-05,
+      "loss": 0.0417,
+      "step": 12254
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.756914330330815e-05,
+      "loss": 0.0432,
+      "step": 12255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.75548219566932e-05,
+      "loss": 0.0444,
+      "step": 12256
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.754050270916526e-05,
+      "loss": 0.0473,
+      "step": 12257
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.752618556120561e-05,
+      "loss": 0.0473,
+      "step": 12258
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.751187051329554e-05,
+      "loss": 0.0427,
+      "step": 12259
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7497557565916265e-05,
+      "loss": 0.047,
+      "step": 12260
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7483246719548905e-05,
+      "loss": 0.0433,
+      "step": 12261
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7468937974674555e-05,
+      "loss": 0.056,
+      "step": 12262
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.745463133177416e-05,
+      "loss": 0.0453,
+      "step": 12263
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.7440326791328604e-05,
+      "loss": 0.0441,
+      "step": 12264
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.742602435381877e-05,
+      "loss": 0.0517,
+      "step": 12265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.741172401972544e-05,
+      "loss": 0.0434,
+      "step": 12266
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.739742578952933e-05,
+      "loss": 0.047,
+      "step": 12267
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.738312966371103e-05,
+      "loss": 0.043,
+      "step": 12268
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.736883564275112e-05,
+      "loss": 0.0485,
+      "step": 12269
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.73545437271301e-05,
+      "loss": 0.0427,
+      "step": 12270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.734025391732842e-05,
+      "loss": 0.0453,
+      "step": 12271
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.732596621382636e-05,
+      "loss": 0.0481,
+      "step": 12272
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.731168061710424e-05,
+      "loss": 0.046,
+      "step": 12273
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.729739712764227e-05,
+      "loss": 0.0478,
+      "step": 12274
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.7283115745920615e-05,
+      "loss": 0.0458,
+      "step": 12275
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.726883647241928e-05,
+      "loss": 0.0436,
+      "step": 12276
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.725455930761833e-05,
+      "loss": 0.049,
+      "step": 12277
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.72402842519976e-05,
+      "loss": 0.0479,
+      "step": 12278
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.722601130603701e-05,
+      "loss": 0.0454,
+      "step": 12279
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.721174047021634e-05,
+      "loss": 0.044,
+      "step": 12280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.719747174501531e-05,
+      "loss": 0.0479,
+      "step": 12281
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.718320513091351e-05,
+      "loss": 0.0438,
+      "step": 12282
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.716894062839055e-05,
+      "loss": 0.0499,
+      "step": 12283
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.715467823792592e-05,
+      "loss": 0.0448,
+      "step": 12284
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.714041795999908e-05,
+      "loss": 0.0444,
+      "step": 12285
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.7126159795089334e-05,
+      "loss": 0.0447,
+      "step": 12286
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.711190374367599e-05,
+      "loss": 0.0495,
+      "step": 12287
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.709764980623831e-05,
+      "loss": 0.0455,
+      "step": 12288
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.708339798325535e-05,
+      "loss": 0.0485,
+      "step": 12289
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.7069148275206234e-05,
+      "loss": 0.0505,
+      "step": 12290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.705490068256999e-05,
+      "loss": 0.0462,
+      "step": 12291
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.704065520582549e-05,
+      "loss": 0.0465,
+      "step": 12292
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.7026411845451617e-05,
+      "loss": 0.0457,
+      "step": 12293
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.701217060192717e-05,
+      "loss": 0.0429,
+      "step": 12294
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6997931475730896e-05,
+      "loss": 0.0435,
+      "step": 12295
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.698369446734137e-05,
+      "loss": 0.0437,
+      "step": 12296
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.69694595772372e-05,
+      "loss": 0.0461,
+      "step": 12297
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.695522680589689e-05,
+      "loss": 0.042,
+      "step": 12298
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.694099615379892e-05,
+      "loss": 0.0424,
+      "step": 12299
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.692676762142161e-05,
+      "loss": 0.0465,
+      "step": 12300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6912541209243215e-05,
+      "loss": 0.0448,
+      "step": 12301
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.689831691774198e-05,
+      "loss": 0.044,
+      "step": 12302
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.688409474739607e-05,
+      "loss": 0.0427,
+      "step": 12303
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.686987469868356e-05,
+      "loss": 0.0438,
+      "step": 12304
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.685565677208247e-05,
+      "loss": 0.0485,
+      "step": 12305
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6841440968070674e-05,
+      "loss": 0.0427,
+      "step": 12306
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6827227287126076e-05,
+      "loss": 0.0425,
+      "step": 12307
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6813015729726476e-05,
+      "loss": 0.0461,
+      "step": 12308
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6798806296349595e-05,
+      "loss": 0.0448,
+      "step": 12309
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.678459898747305e-05,
+      "loss": 0.0463,
+      "step": 12310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.677039380357442e-05,
+      "loss": 0.0426,
+      "step": 12311
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6756190745131236e-05,
+      "loss": 0.0516,
+      "step": 12312
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.674198981262096e-05,
+      "loss": 0.0439,
+      "step": 12313
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6727791006520907e-05,
+      "loss": 0.0435,
+      "step": 12314
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.671359432730834e-05,
+      "loss": 0.0458,
+      "step": 12315
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.669939977546053e-05,
+      "loss": 0.0432,
+      "step": 12316
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.668520735145461e-05,
+      "loss": 0.0453,
+      "step": 12317
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6671017055767655e-05,
+      "loss": 0.0436,
+      "step": 12318
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.66568288888767e-05,
+      "loss": 0.0433,
+      "step": 12319
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.664264285125863e-05,
+      "loss": 0.0443,
+      "step": 12320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6628458943390306e-05,
+      "loss": 0.044,
+      "step": 12321
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.661427716574857e-05,
+      "loss": 0.0552,
+      "step": 12322
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.660009751881013e-05,
+      "loss": 0.0449,
+      "step": 12323
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6585920003051575e-05,
+      "loss": 0.0419,
+      "step": 12324
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.657174461894952e-05,
+      "loss": 0.0447,
+      "step": 12325
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.655757136698051e-05,
+      "loss": 0.0477,
+      "step": 12326
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.654340024762091e-05,
+      "loss": 0.0507,
+      "step": 12327
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.65292312613471e-05,
+      "loss": 0.0455,
+      "step": 12328
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.65150644086354e-05,
+      "loss": 0.0424,
+      "step": 12329
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6500899689961975e-05,
+      "loss": 0.0443,
+      "step": 12330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6486737105803006e-05,
+      "loss": 0.0452,
+      "step": 12331
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.647257665663454e-05,
+      "loss": 0.0523,
+      "step": 12332
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6458418342932645e-05,
+      "loss": 0.0434,
+      "step": 12333
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.644426216517316e-05,
+      "loss": 0.049,
+      "step": 12334
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6430108123832e-05,
+      "loss": 0.0473,
+      "step": 12335
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.641595621938492e-05,
+      "loss": 0.0489,
+      "step": 12336
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.640180645230768e-05,
+      "loss": 0.0421,
+      "step": 12337
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.638765882307589e-05,
+      "loss": 0.0423,
+      "step": 12338
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.637351333216513e-05,
+      "loss": 0.0452,
+      "step": 12339
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.635936998005087e-05,
+      "loss": 0.0432,
+      "step": 12340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.634522876720857e-05,
+      "loss": 0.0413,
+      "step": 12341
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6331089694113575e-05,
+      "loss": 0.0497,
+      "step": 12342
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6316952761241195e-05,
+      "loss": 0.0436,
+      "step": 12343
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6302817969066584e-05,
+      "loss": 0.043,
+      "step": 12344
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6288685318064906e-05,
+      "loss": 0.0456,
+      "step": 12345
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6274554808711246e-05,
+      "loss": 0.0547,
+      "step": 12346
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.626042644148062e-05,
+      "loss": 0.0502,
+      "step": 12347
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.624630021684788e-05,
+      "loss": 0.0456,
+      "step": 12348
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.623217613528793e-05,
+      "loss": 0.0452,
+      "step": 12349
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.621805419727553e-05,
+      "loss": 0.0421,
+      "step": 12350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.620393440328543e-05,
+      "loss": 0.0468,
+      "step": 12351
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6189816753792225e-05,
+      "loss": 0.0475,
+      "step": 12352
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.617570124927044e-05,
+      "loss": 0.0437,
+      "step": 12353
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.616158789019462e-05,
+      "loss": 0.0479,
+      "step": 12354
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6147476677039174e-05,
+      "loss": 0.0435,
+      "step": 12355
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.613336761027845e-05,
+      "loss": 0.0484,
+      "step": 12356
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6119260690386735e-05,
+      "loss": 0.0446,
+      "step": 12357
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.61051559178382e-05,
+      "loss": 0.0428,
+      "step": 12358
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.609105329310698e-05,
+      "loss": 0.0431,
+      "step": 12359
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6076952816667144e-05,
+      "loss": 0.0431,
+      "step": 12360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6062854488992714e-05,
+      "loss": 0.045,
+      "step": 12361
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6048758310557526e-05,
+      "loss": 0.0467,
+      "step": 12362
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.603466428183547e-05,
+      "loss": 0.0514,
+      "step": 12363
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.60205724033003e-05,
+      "loss": 0.0446,
+      "step": 12364
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6006482675425765e-05,
+      "loss": 0.043,
+      "step": 12365
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.599239509868544e-05,
+      "loss": 0.0457,
+      "step": 12366
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.5978309673552846e-05,
+      "loss": 0.0423,
+      "step": 12367
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.59642264005015e-05,
+      "loss": 0.0455,
+      "step": 12368
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.595014528000481e-05,
+      "loss": 0.0425,
+      "step": 12369
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.59360663125361e-05,
+      "loss": 0.0427,
+      "step": 12370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.5921989498568684e-05,
+      "loss": 0.0489,
+      "step": 12371
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.5907914838575676e-05,
+      "loss": 0.0474,
+      "step": 12372
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.589384233303023e-05,
+      "loss": 0.0437,
+      "step": 12373
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.5879771982405376e-05,
+      "loss": 0.0463,
+      "step": 12374
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.586570378717415e-05,
+      "loss": 0.0506,
+      "step": 12375
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.585163774780936e-05,
+      "loss": 0.0448,
+      "step": 12376
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.583757386478389e-05,
+      "loss": 0.0445,
+      "step": 12377
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.5823512138570504e-05,
+      "loss": 0.0446,
+      "step": 12378
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.580945256964183e-05,
+      "loss": 0.0452,
+      "step": 12379
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.579539515847056e-05,
+      "loss": 0.0436,
+      "step": 12380
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.578133990552914e-05,
+      "loss": 0.046,
+      "step": 12381
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.576728681129009e-05,
+      "loss": 0.0482,
+      "step": 12382
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5753235876225786e-05,
+      "loss": 0.0432,
+      "step": 12383
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.573918710080857e-05,
+      "loss": 0.0455,
+      "step": 12384
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.57251404855107e-05,
+      "loss": 0.0423,
+      "step": 12385
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.571109603080429e-05,
+      "loss": 0.0453,
+      "step": 12386
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.569705373716149e-05,
+      "loss": 0.0435,
+      "step": 12387
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5683013605054316e-05,
+      "loss": 0.0446,
+      "step": 12388
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.566897563495476e-05,
+      "loss": 0.0452,
+      "step": 12389
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5654939827334644e-05,
+      "loss": 0.0448,
+      "step": 12390
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.564090618266584e-05,
+      "loss": 0.046,
+      "step": 12391
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.562687470142004e-05,
+      "loss": 0.0452,
+      "step": 12392
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.561284538406892e-05,
+      "loss": 0.0442,
+      "step": 12393
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.559881823108412e-05,
+      "loss": 0.043,
+      "step": 12394
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5584793242937085e-05,
+      "loss": 0.0475,
+      "step": 12395
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.557077042009932e-05,
+      "loss": 0.042,
+      "step": 12396
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.555674976304216e-05,
+      "loss": 0.045,
+      "step": 12397
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.554273127223695e-05,
+      "loss": 0.0439,
+      "step": 12398
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.552871494815493e-05,
+      "loss": 0.0465,
+      "step": 12399
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5514700791267185e-05,
+      "loss": 0.0479,
+      "step": 12400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5500688802044857e-05,
+      "loss": 0.0426,
+      "step": 12401
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.548667898095893e-05,
+      "loss": 0.0489,
+      "step": 12402
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.547267132848039e-05,
+      "loss": 0.0473,
+      "step": 12403
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.545866584508007e-05,
+      "loss": 0.0444,
+      "step": 12404
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5444662531228726e-05,
+      "loss": 0.049,
+      "step": 12405
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.543066138739711e-05,
+      "loss": 0.0436,
+      "step": 12406
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.541666241405588e-05,
+      "loss": 0.0522,
+      "step": 12407
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5402665611675624e-05,
+      "loss": 0.0463,
+      "step": 12408
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.538867098072678e-05,
+      "loss": 0.0439,
+      "step": 12409
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5374678521679825e-05,
+      "loss": 0.0449,
+      "step": 12410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.53606882350051e-05,
+      "loss": 0.0421,
+      "step": 12411
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.534670012117289e-05,
+      "loss": 0.0511,
+      "step": 12412
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.533271418065344e-05,
+      "loss": 0.0436,
+      "step": 12413
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.531873041391681e-05,
+      "loss": 0.0455,
+      "step": 12414
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.530474882143311e-05,
+      "loss": 0.044,
+      "step": 12415
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.529076940367231e-05,
+      "loss": 0.0442,
+      "step": 12416
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.527679216110438e-05,
+      "loss": 0.047,
+      "step": 12417
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.526281709419912e-05,
+      "loss": 0.0432,
+      "step": 12418
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5248844203426266e-05,
+      "loss": 0.0453,
+      "step": 12419
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.523487348925556e-05,
+      "loss": 0.0457,
+      "step": 12420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5220904952156605e-05,
+      "loss": 0.0466,
+      "step": 12421
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5206938592599e-05,
+      "loss": 0.0435,
+      "step": 12422
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.519297441105216e-05,
+      "loss": 0.0472,
+      "step": 12423
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5179012407985504e-05,
+      "loss": 0.0466,
+      "step": 12424
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.516505258386837e-05,
+      "loss": 0.045,
+      "step": 12425
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.515109493917004e-05,
+      "loss": 0.0428,
+      "step": 12426
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5137139474359693e-05,
+      "loss": 0.0452,
+      "step": 12427
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.51231861899064e-05,
+      "loss": 0.0449,
+      "step": 12428
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5109235086279225e-05,
+      "loss": 0.0459,
+      "step": 12429
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.509528616394716e-05,
+      "loss": 0.0441,
+      "step": 12430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5081339423379036e-05,
+      "loss": 0.0441,
+      "step": 12431
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.506739486504373e-05,
+      "loss": 0.044,
+      "step": 12432
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5053452489409946e-05,
+      "loss": 0.043,
+      "step": 12433
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5039512296946355e-05,
+      "loss": 0.043,
+      "step": 12434
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5025574288121565e-05,
+      "loss": 0.0418,
+      "step": 12435
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.5011638463404115e-05,
+      "loss": 0.0425,
+      "step": 12436
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.499770482326247e-05,
+      "loss": 0.0467,
+      "step": 12437
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.498377336816495e-05,
+      "loss": 0.0458,
+      "step": 12438
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4969844098579896e-05,
+      "loss": 0.0418,
+      "step": 12439
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.495591701497554e-05,
+      "loss": 0.0427,
+      "step": 12440
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.494199211782006e-05,
+      "loss": 0.0515,
+      "step": 12441
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.492806940758151e-05,
+      "loss": 0.0422,
+      "step": 12442
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.491414888472788e-05,
+      "loss": 0.0446,
+      "step": 12443
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.490023054972713e-05,
+      "loss": 0.0442,
+      "step": 12444
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.488631440304711e-05,
+      "loss": 0.0429,
+      "step": 12445
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4872400445155674e-05,
+      "loss": 0.043,
+      "step": 12446
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4858488676520437e-05,
+      "loss": 0.042,
+      "step": 12447
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4844579097609106e-05,
+      "loss": 0.0464,
+      "step": 12448
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.483067170888923e-05,
+      "loss": 0.045,
+      "step": 12449
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4816766510828313e-05,
+      "loss": 0.0443,
+      "step": 12450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.48028635038938e-05,
+      "loss": 0.0447,
+      "step": 12451
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4788962688552986e-05,
+      "loss": 0.0423,
+      "step": 12452
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4775064065273165e-05,
+      "loss": 0.0446,
+      "step": 12453
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.476116763452154e-05,
+      "loss": 0.0427,
+      "step": 12454
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.474727339676529e-05,
+      "loss": 0.0431,
+      "step": 12455
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4733381352471405e-05,
+      "loss": 0.0467,
+      "step": 12456
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.471949150210685e-05,
+      "loss": 0.0456,
+      "step": 12457
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.470560384613856e-05,
+      "loss": 0.044,
+      "step": 12458
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.469171838503337e-05,
+      "loss": 0.0474,
+      "step": 12459
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.467783511925806e-05,
+      "loss": 0.0446,
+      "step": 12460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.466395404927926e-05,
+      "loss": 0.0467,
+      "step": 12461
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.465007517556361e-05,
+      "loss": 0.0461,
+      "step": 12462
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.463619849857764e-05,
+      "loss": 0.0504,
+      "step": 12463
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4622324018787836e-05,
+      "loss": 0.0574,
+      "step": 12464
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.460845173666059e-05,
+      "loss": 0.044,
+      "step": 12465
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.459458165266215e-05,
+      "loss": 0.0451,
+      "step": 12466
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.458071376725882e-05,
+      "loss": 0.0433,
+      "step": 12467
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.456684808091678e-05,
+      "loss": 0.0453,
+      "step": 12468
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.455298459410207e-05,
+      "loss": 0.0459,
+      "step": 12469
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.453912330728075e-05,
+      "loss": 0.0444,
+      "step": 12470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.452526422091873e-05,
+      "loss": 0.0421,
+      "step": 12471
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4511407335481895e-05,
+      "loss": 0.0444,
+      "step": 12472
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.449755265143605e-05,
+      "loss": 0.0465,
+      "step": 12473
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4483700169246944e-05,
+      "loss": 0.0446,
+      "step": 12474
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.446984988938018e-05,
+      "loss": 0.0414,
+      "step": 12475
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.445600181230134e-05,
+      "loss": 0.0422,
+      "step": 12476
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.444215593847594e-05,
+      "loss": 0.0444,
+      "step": 12477
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.442831226836941e-05,
+      "loss": 0.0548,
+      "step": 12478
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.441447080244713e-05,
+      "loss": 0.0471,
+      "step": 12479
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.440063154117431e-05,
+      "loss": 0.0523,
+      "step": 12480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.438679448501623e-05,
+      "loss": 0.0497,
+      "step": 12481
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.437295963443795e-05,
+      "loss": 0.0453,
+      "step": 12482
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4359126989904555e-05,
+      "loss": 0.0462,
+      "step": 12483
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.434529655188108e-05,
+      "loss": 0.0542,
+      "step": 12484
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.433146832083234e-05,
+      "loss": 0.0436,
+      "step": 12485
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.431764229722323e-05,
+      "loss": 0.0431,
+      "step": 12486
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.430381848151848e-05,
+      "loss": 0.043,
+      "step": 12487
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.428999687418284e-05,
+      "loss": 0.0464,
+      "step": 12488
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.427617747568083e-05,
+      "loss": 0.0415,
+      "step": 12489
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4262360286477036e-05,
+      "loss": 0.0469,
+      "step": 12490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.424854530703592e-05,
+      "loss": 0.0461,
+      "step": 12491
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.4234732537821856e-05,
+      "loss": 0.0459,
+      "step": 12492
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.42209219792992e-05,
+      "loss": 0.0471,
+      "step": 12493
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.420711363193218e-05,
+      "loss": 0.0448,
+      "step": 12494
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.41933074961849e-05,
+      "loss": 0.0476,
+      "step": 12495
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.417950357252149e-05,
+      "loss": 0.0467,
+      "step": 12496
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.4165701861405985e-05,
+      "loss": 0.0447,
+      "step": 12497
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.415190236330234e-05,
+      "loss": 0.0434,
+      "step": 12498
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.413810507867436e-05,
+      "loss": 0.0452,
+      "step": 12499
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.412431000798588e-05,
+      "loss": 0.0463,
+      "step": 12500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.4110517151700624e-05,
+      "loss": 0.0439,
+      "step": 12501
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.409672651028226e-05,
+      "loss": 0.0452,
+      "step": 12502
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.4082938084194305e-05,
+      "loss": 0.0432,
+      "step": 12503
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.406915187390027e-05,
+      "loss": 0.0449,
+      "step": 12504
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.405536787986359e-05,
+      "loss": 0.0436,
+      "step": 12505
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.404158610254761e-05,
+      "loss": 0.0509,
+      "step": 12506
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.4027806542415616e-05,
+      "loss": 0.0472,
+      "step": 12507
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.401402919993082e-05,
+      "loss": 0.0422,
+      "step": 12508
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.400025407555626e-05,
+      "loss": 0.0423,
+      "step": 12509
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.398648116975506e-05,
+      "loss": 0.0415,
+      "step": 12510
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.397271048299017e-05,
+      "loss": 0.044,
+      "step": 12511
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.395894201572453e-05,
+      "loss": 0.0438,
+      "step": 12512
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.394517576842089e-05,
+      "loss": 0.0496,
+      "step": 12513
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.393141174154205e-05,
+      "loss": 0.0443,
+      "step": 12514
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.391764993555068e-05,
+      "loss": 0.0456,
+      "step": 12515
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3903890350909404e-05,
+      "loss": 0.0433,
+      "step": 12516
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.38901329880807e-05,
+      "loss": 0.0504,
+      "step": 12517
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.387637784752705e-05,
+      "loss": 0.0431,
+      "step": 12518
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.386262492971082e-05,
+      "loss": 0.0456,
+      "step": 12519
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.384887423509435e-05,
+      "loss": 0.0495,
+      "step": 12520
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.383512576413981e-05,
+      "loss": 0.0449,
+      "step": 12521
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3821379517309405e-05,
+      "loss": 0.0435,
+      "step": 12522
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.380763549506516e-05,
+      "loss": 0.0427,
+      "step": 12523
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3793893697869106e-05,
+      "loss": 0.0429,
+      "step": 12524
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.378015412618318e-05,
+      "loss": 0.0501,
+      "step": 12525
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3766416780469256e-05,
+      "loss": 0.044,
+      "step": 12526
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3752681661189054e-05,
+      "loss": 0.0435,
+      "step": 12527
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.373894876880431e-05,
+      "loss": 0.0518,
+      "step": 12528
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.372521810377664e-05,
+      "loss": 0.0451,
+      "step": 12529
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.371148966656767e-05,
+      "loss": 0.042,
+      "step": 12530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3697763457638776e-05,
+      "loss": 0.0419,
+      "step": 12531
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.368403947745141e-05,
+      "loss": 0.0421,
+      "step": 12532
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3670317726466937e-05,
+      "loss": 0.0449,
+      "step": 12533
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.365659820514654e-05,
+      "loss": 0.0432,
+      "step": 12534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.364288091395144e-05,
+      "loss": 0.048,
+      "step": 12535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.362916585334276e-05,
+      "loss": 0.0474,
+      "step": 12536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.361545302378148e-05,
+      "loss": 0.043,
+      "step": 12537
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.360174242572859e-05,
+      "loss": 0.0447,
+      "step": 12538
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.358803405964495e-05,
+      "loss": 0.0465,
+      "step": 12539
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.357432792599141e-05,
+      "loss": 0.0488,
+      "step": 12540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.356062402522865e-05,
+      "loss": 0.0467,
+      "step": 12541
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.354692235781732e-05,
+      "loss": 0.0522,
+      "step": 12542
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3533222924218045e-05,
+      "loss": 0.047,
+      "step": 12543
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.351952572489132e-05,
+      "loss": 0.0424,
+      "step": 12544
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.350583076029754e-05,
+      "loss": 0.0446,
+      "step": 12545
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3492138030897114e-05,
+      "loss": 0.0445,
+      "step": 12546
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.347844753715025e-05,
+      "loss": 0.0465,
+      "step": 12547
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.34647592795172e-05,
+      "loss": 0.0493,
+      "step": 12548
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.3451073258458065e-05,
+      "loss": 0.0441,
+      "step": 12549
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.343738947443297e-05,
+      "loss": 0.0442,
+      "step": 12550
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.34237079279018e-05,
+      "loss": 0.0434,
+      "step": 12551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.341002861932451e-05,
+      "loss": 0.0433,
+      "step": 12552
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.33963515491609e-05,
+      "loss": 0.0557,
+      "step": 12553
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3382676717870786e-05,
+      "loss": 0.0469,
+      "step": 12554
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.336900412591376e-05,
+      "loss": 0.0454,
+      "step": 12555
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.335533377374948e-05,
+      "loss": 0.0464,
+      "step": 12556
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.334166566183745e-05,
+      "loss": 0.0457,
+      "step": 12557
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3327999790637135e-05,
+      "loss": 0.0424,
+      "step": 12558
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.331433616060796e-05,
+      "loss": 0.0548,
+      "step": 12559
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.330067477220912e-05,
+      "loss": 0.0474,
+      "step": 12560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.328701562589989e-05,
+      "loss": 0.0466,
+      "step": 12561
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.327335872213943e-05,
+      "loss": 0.0417,
+      "step": 12562
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.325970406138681e-05,
+      "loss": 0.0452,
+      "step": 12563
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.324605164410105e-05,
+      "loss": 0.0449,
+      "step": 12564
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3232401470741034e-05,
+      "loss": 0.0492,
+      "step": 12565
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3218753541765627e-05,
+      "loss": 0.0422,
+      "step": 12566
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.320510785763361e-05,
+      "loss": 0.0474,
+      "step": 12567
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.319146441880371e-05,
+      "loss": 0.0418,
+      "step": 12568
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.317782322573447e-05,
+      "loss": 0.0443,
+      "step": 12569
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.31641842788845e-05,
+      "loss": 0.0441,
+      "step": 12570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3150547578712285e-05,
+      "loss": 0.0441,
+      "step": 12571
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.313691312567616e-05,
+      "loss": 0.043,
+      "step": 12572
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.312328092023447e-05,
+      "loss": 0.0429,
+      "step": 12573
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.310965096284551e-05,
+      "loss": 0.0437,
+      "step": 12574
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.309602325396736e-05,
+      "loss": 0.0431,
+      "step": 12575
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.308239779405817e-05,
+      "loss": 0.0515,
+      "step": 12576
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3068774583575946e-05,
+      "loss": 0.0423,
+      "step": 12577
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3055153622978665e-05,
+      "loss": 0.0464,
+      "step": 12578
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.3041534912724125e-05,
+      "loss": 0.045,
+      "step": 12579
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.302791845327016e-05,
+      "loss": 0.0505,
+      "step": 12580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.301430424507448e-05,
+      "loss": 0.0428,
+      "step": 12581
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.300069228859476e-05,
+      "loss": 0.0431,
+      "step": 12582
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.298708258428849e-05,
+      "loss": 0.0438,
+      "step": 12583
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.297347513261323e-05,
+      "loss": 0.0445,
+      "step": 12584
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.295986993402632e-05,
+      "loss": 0.0458,
+      "step": 12585
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.294626698898515e-05,
+      "loss": 0.0488,
+      "step": 12586
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2932666297946954e-05,
+      "loss": 0.0479,
+      "step": 12587
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.291906786136898e-05,
+      "loss": 0.045,
+      "step": 12588
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2905471679708256e-05,
+      "loss": 0.0481,
+      "step": 12589
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.289187775342184e-05,
+      "loss": 0.0455,
+      "step": 12590
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.28782860829667e-05,
+      "loss": 0.0442,
+      "step": 12591
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2864696668799766e-05,
+      "loss": 0.0481,
+      "step": 12592
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.285110951137775e-05,
+      "loss": 0.0526,
+      "step": 12593
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.283752461115743e-05,
+      "loss": 0.0409,
+      "step": 12594
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2823941968595465e-05,
+      "loss": 0.0459,
+      "step": 12595
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2810361584148466e-05,
+      "loss": 0.0453,
+      "step": 12596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.279678345827288e-05,
+      "loss": 0.0438,
+      "step": 12597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2783207591425144e-05,
+      "loss": 0.0449,
+      "step": 12598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.276963398406161e-05,
+      "loss": 0.05,
+      "step": 12599
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2756062636638554e-05,
+      "loss": 0.0462,
+      "step": 12600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.274249354961218e-05,
+      "loss": 0.0434,
+      "step": 12601
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.272892672343866e-05,
+      "loss": 0.0461,
+      "step": 12602
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.271536215857395e-05,
+      "loss": 0.0433,
+      "step": 12603
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.270179985547407e-05,
+      "loss": 0.0443,
+      "step": 12604
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.268823981459491e-05,
+      "loss": 0.0497,
+      "step": 12605
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2674682036392335e-05,
+      "loss": 0.048,
+      "step": 12606
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.2661126521321995e-05,
+      "loss": 0.0471,
+      "step": 12607
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.264757326983961e-05,
+      "loss": 0.0429,
+      "step": 12608
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2634022282400764e-05,
+      "loss": 0.0493,
+      "step": 12609
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2620473559461016e-05,
+      "loss": 0.0455,
+      "step": 12610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.260692710147575e-05,
+      "loss": 0.0455,
+      "step": 12611
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.259338290890032e-05,
+      "loss": 0.0431,
+      "step": 12612
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.257984098219002e-05,
+      "loss": 0.046,
+      "step": 12613
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2566301321800085e-05,
+      "loss": 0.0449,
+      "step": 12614
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.255276392818564e-05,
+      "loss": 0.0455,
+      "step": 12615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.253922880180176e-05,
+      "loss": 0.044,
+      "step": 12616
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.252569594310338e-05,
+      "loss": 0.0419,
+      "step": 12617
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2512165352545434e-05,
+      "loss": 0.0426,
+      "step": 12618
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.249863703058274e-05,
+      "loss": 0.0458,
+      "step": 12619
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2485110977670106e-05,
+      "loss": 0.0433,
+      "step": 12620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2471587194262135e-05,
+      "loss": 0.0458,
+      "step": 12621
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.245806568081344e-05,
+      "loss": 0.0425,
+      "step": 12622
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.24445464377786e-05,
+      "loss": 0.0439,
+      "step": 12623
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.243102946561199e-05,
+      "loss": 0.0425,
+      "step": 12624
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2417514764768055e-05,
+      "loss": 0.0627,
+      "step": 12625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.240400233570101e-05,
+      "loss": 0.0471,
+      "step": 12626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.239049217886511e-05,
+      "loss": 0.044,
+      "step": 12627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.237698429471451e-05,
+      "loss": 0.0473,
+      "step": 12628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.236347868370325e-05,
+      "loss": 0.0483,
+      "step": 12629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.234997534628538e-05,
+      "loss": 0.0444,
+      "step": 12630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.233647428291473e-05,
+      "loss": 0.0433,
+      "step": 12631
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.232297549404518e-05,
+      "loss": 0.0429,
+      "step": 12632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.230947898013048e-05,
+      "loss": 0.0432,
+      "step": 12633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2295984741624366e-05,
+      "loss": 0.0458,
+      "step": 12634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.228249277898034e-05,
+      "loss": 0.049,
+      "step": 12635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.226900309265205e-05,
+      "loss": 0.0442,
+      "step": 12636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.225551568309284e-05,
+      "loss": 0.0433,
+      "step": 12637
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.224203055075615e-05,
+      "loss": 0.0441,
+      "step": 12638
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.22285476960953e-05,
+      "loss": 0.0452,
+      "step": 12639
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.221506711956345e-05,
+      "loss": 0.047,
+      "step": 12640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.220158882161378e-05,
+      "loss": 0.0437,
+      "step": 12641
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.218811280269937e-05,
+      "loss": 0.0425,
+      "step": 12642
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2174639063273194e-05,
+      "loss": 0.0491,
+      "step": 12643
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2161167603788235e-05,
+      "loss": 0.0444,
+      "step": 12644
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2147698424697236e-05,
+      "loss": 0.049,
+      "step": 12645
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.213423152645302e-05,
+      "loss": 0.0457,
+      "step": 12646
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.212076690950826e-05,
+      "loss": 0.0448,
+      "step": 12647
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.21073045743156e-05,
+      "loss": 0.0466,
+      "step": 12648
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.209384452132756e-05,
+      "loss": 0.045,
+      "step": 12649
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2080386750996535e-05,
+      "loss": 0.0458,
+      "step": 12650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.206693126377497e-05,
+      "loss": 0.0439,
+      "step": 12651
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.205347806011515e-05,
+      "loss": 0.0496,
+      "step": 12652
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.204002714046934e-05,
+      "loss": 0.0457,
+      "step": 12653
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2026578505289626e-05,
+      "loss": 0.0452,
+      "step": 12654
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.2013132155028125e-05,
+      "loss": 0.0486,
+      "step": 12655
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.199968809013681e-05,
+      "loss": 0.0446,
+      "step": 12656
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1986246311067633e-05,
+      "loss": 0.0489,
+      "step": 12657
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.197280681827245e-05,
+      "loss": 0.0423,
+      "step": 12658
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.195936961220296e-05,
+      "loss": 0.0465,
+      "step": 12659
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1945934693310896e-05,
+      "loss": 0.0436,
+      "step": 12660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1932502062047866e-05,
+      "loss": 0.0755,
+      "step": 12661
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.191907171886544e-05,
+      "loss": 0.0462,
+      "step": 12662
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.190564366421505e-05,
+      "loss": 0.0532,
+      "step": 12663
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.189221789854804e-05,
+      "loss": 0.0434,
+      "step": 12664
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.187879442231574e-05,
+      "loss": 0.0538,
+      "step": 12665
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.18653732359694e-05,
+      "loss": 0.0438,
+      "step": 12666
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.185195433996018e-05,
+      "loss": 0.0442,
+      "step": 12667
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1838537734739104e-05,
+      "loss": 0.0476,
+      "step": 12668
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.182512342075721e-05,
+      "loss": 0.0548,
+      "step": 12669
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1811711398465395e-05,
+      "loss": 0.0464,
+      "step": 12670
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1798301668314524e-05,
+      "loss": 0.0455,
+      "step": 12671
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.178489423075538e-05,
+      "loss": 0.042,
+      "step": 12672
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1771489086238595e-05,
+      "loss": 0.0433,
+      "step": 12673
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1758086235214814e-05,
+      "loss": 0.0429,
+      "step": 12674
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.174468567813461e-05,
+      "loss": 0.043,
+      "step": 12675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1731287415448366e-05,
+      "loss": 0.0463,
+      "step": 12676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.171789144760653e-05,
+      "loss": 0.0424,
+      "step": 12677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.170449777505934e-05,
+      "loss": 0.0433,
+      "step": 12678
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1691106398257065e-05,
+      "loss": 0.0444,
+      "step": 12679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1677717317649834e-05,
+      "loss": 0.0429,
+      "step": 12680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.166433053368775e-05,
+      "loss": 0.0464,
+      "step": 12681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1650946046820815e-05,
+      "loss": 0.0442,
+      "step": 12682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.163756385749889e-05,
+      "loss": 0.044,
+      "step": 12683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.162418396617185e-05,
+      "loss": 0.0435,
+      "step": 12684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.161080637328946e-05,
+      "loss": 0.0511,
+      "step": 12685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.159743107930142e-05,
+      "loss": 0.0474,
+      "step": 12686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.158405808465731e-05,
+      "loss": 0.0436,
+      "step": 12687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.157068738980669e-05,
+      "loss": 0.0441,
+      "step": 12688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.155731899519897e-05,
+      "loss": 0.0446,
+      "step": 12689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1543952901283554e-05,
+      "loss": 0.0471,
+      "step": 12690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1530589108509775e-05,
+      "loss": 0.0476,
+      "step": 12691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.151722761732679e-05,
+      "loss": 0.0494,
+      "step": 12692
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.150386842818378e-05,
+      "loss": 0.0437,
+      "step": 12693
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1490511541529796e-05,
+      "loss": 0.0524,
+      "step": 12694
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1477156957813856e-05,
+      "loss": 0.0452,
+      "step": 12695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.146380467748489e-05,
+      "loss": 0.0436,
+      "step": 12696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.145045470099166e-05,
+      "loss": 0.0457,
+      "step": 12697
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.143710702878296e-05,
+      "loss": 0.0425,
+      "step": 12698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1423761661307484e-05,
+      "loss": 0.0473,
+      "step": 12699
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1410418599013856e-05,
+      "loss": 0.0477,
+      "step": 12700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.139707784235058e-05,
+      "loss": 0.0413,
+      "step": 12701
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.138373939176604e-05,
+      "loss": 0.0454,
+      "step": 12702
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.137040324770868e-05,
+      "loss": 0.0489,
+      "step": 12703
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1357069410626774e-05,
+      "loss": 0.0442,
+      "step": 12704
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.134373788096856e-05,
+      "loss": 0.0421,
+      "step": 12705
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133040865918213e-05,
+      "loss": 0.0443,
+      "step": 12706
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.131708174571556e-05,
+      "loss": 0.0444,
+      "step": 12707
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.130375714101684e-05,
+      "loss": 0.0421,
+      "step": 12708
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.129043484553389e-05,
+      "loss": 0.0424,
+      "step": 12709
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.127711485971454e-05,
+      "loss": 0.0426,
+      "step": 12710
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.126379718400649e-05,
+      "loss": 0.0428,
+      "step": 12711
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.125048181885745e-05,
+      "loss": 0.0441,
+      "step": 12712
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.123716876471504e-05,
+      "loss": 0.0414,
+      "step": 12713
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.122385802202671e-05,
+      "loss": 0.0425,
+      "step": 12714
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.121054959123997e-05,
+      "loss": 0.0452,
+      "step": 12715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1197243472802107e-05,
+      "loss": 0.0421,
+      "step": 12716
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1183939667160444e-05,
+      "loss": 0.0429,
+      "step": 12717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.117063817476219e-05,
+      "loss": 0.0455,
+      "step": 12718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1157338996054494e-05,
+      "loss": 0.0466,
+      "step": 12719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1144042131484355e-05,
+      "loss": 0.0425,
+      "step": 12720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.113074758149876e-05,
+      "loss": 0.0441,
+      "step": 12721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.111745534654463e-05,
+      "loss": 0.0457,
+      "step": 12722
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1104165427068764e-05,
+      "loss": 0.0508,
+      "step": 12723
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1090877823517925e-05,
+      "loss": 0.0448,
+      "step": 12724
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.107759253633874e-05,
+      "loss": 0.0443,
+      "step": 12725
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1064309565977825e-05,
+      "loss": 0.0445,
+      "step": 12726
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.105102891288163e-05,
+      "loss": 0.0432,
+      "step": 12727
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.103775057749663e-05,
+      "loss": 0.0433,
+      "step": 12728
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.102447456026919e-05,
+      "loss": 0.0451,
+      "step": 12729
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.101120086164552e-05,
+      "loss": 0.0472,
+      "step": 12730
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0997929482071844e-05,
+      "loss": 0.0418,
+      "step": 12731
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.098466042199429e-05,
+      "loss": 0.0419,
+      "step": 12732
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.097139368185892e-05,
+      "loss": 0.0452,
+      "step": 12733
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0958129262111625e-05,
+      "loss": 0.0542,
+      "step": 12734
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.094486716319832e-05,
+      "loss": 0.0468,
+      "step": 12735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.093160738556482e-05,
+      "loss": 0.045,
+      "step": 12736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.091834992965684e-05,
+      "loss": 0.0433,
+      "step": 12737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.090509479592007e-05,
+      "loss": 0.0452,
+      "step": 12738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0891841984800044e-05,
+      "loss": 0.0785,
+      "step": 12739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.08785914967422e-05,
+      "loss": 0.0432,
+      "step": 12740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.086534333219201e-05,
+      "loss": 0.043,
+      "step": 12741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.08520974915948e-05,
+      "loss": 0.0476,
+      "step": 12742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.083885397539586e-05,
+      "loss": 0.0475,
+      "step": 12743
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.082561278404031e-05,
+      "loss": 0.0524,
+      "step": 12744
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.081237391797328e-05,
+      "loss": 0.0442,
+      "step": 12745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0799137377639796e-05,
+      "loss": 0.0444,
+      "step": 12746
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.078590316348482e-05,
+      "loss": 0.0435,
+      "step": 12747
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.077267127595317e-05,
+      "loss": 0.0444,
+      "step": 12748
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.075944171548965e-05,
+      "loss": 0.0439,
+      "step": 12749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0746214482538995e-05,
+      "loss": 0.0464,
+      "step": 12750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0732989577545826e-05,
+      "loss": 0.0451,
+      "step": 12751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0719767000954714e-05,
+      "loss": 0.0452,
+      "step": 12752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.070654675321012e-05,
+      "loss": 0.0446,
+      "step": 12753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0693328834756394e-05,
+      "loss": 0.0453,
+      "step": 12754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.06801132460379e-05,
+      "loss": 0.0445,
+      "step": 12755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0666899987498874e-05,
+      "loss": 0.0479,
+      "step": 12756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0653689059583514e-05,
+      "loss": 0.0437,
+      "step": 12757
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.064048046273583e-05,
+      "loss": 0.0432,
+      "step": 12758
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.062727419739987e-05,
+      "loss": 0.0446,
+      "step": 12759
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.061407026401955e-05,
+      "loss": 0.0462,
+      "step": 12760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.060086866303876e-05,
+      "loss": 0.0427,
+      "step": 12761
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.058766939490121e-05,
+      "loss": 0.0449,
+      "step": 12762
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0574472460050604e-05,
+      "loss": 0.045,
+      "step": 12763
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.056127785893057e-05,
+      "loss": 0.0428,
+      "step": 12764
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.054808559198468e-05,
+      "loss": 0.0426,
+      "step": 12765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0534895659656314e-05,
+      "loss": 0.0502,
+      "step": 12766
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.052170806238892e-05,
+      "loss": 0.0457,
+      "step": 12767
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0508522800625738e-05,
+      "loss": 0.0418,
+      "step": 12768
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0495339874810003e-05,
+      "loss": 0.0456,
+      "step": 12769
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.048215928538488e-05,
+      "loss": 0.0442,
+      "step": 12770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0468981032793454e-05,
+      "loss": 0.0431,
+      "step": 12771
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0455805117478643e-05,
+      "loss": 0.0464,
+      "step": 12772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0442631539883392e-05,
+      "loss": 0.0431,
+      "step": 12773
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.042946030045053e-05,
+      "loss": 0.0454,
+      "step": 12774
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.041629139962283e-05,
+      "loss": 0.0482,
+      "step": 12775
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.040312483784291e-05,
+      "loss": 0.0446,
+      "step": 12776
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.038996061555338e-05,
+      "loss": 0.0492,
+      "step": 12777
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.037679873319681e-05,
+      "loss": 0.0419,
+      "step": 12778
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.036363919121554e-05,
+      "loss": 0.0508,
+      "step": 12779
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0350481990051982e-05,
+      "loss": 0.0457,
+      "step": 12780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.033732713014844e-05,
+      "loss": 0.042,
+      "step": 12781
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0324174611947043e-05,
+      "loss": 0.0428,
+      "step": 12782
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0311024435889955e-05,
+      "loss": 0.0478,
+      "step": 12783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0297876602419195e-05,
+      "loss": 0.0472,
+      "step": 12784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0284731111976783e-05,
+      "loss": 0.0444,
+      "step": 12785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0271587965004533e-05,
+      "loss": 0.0451,
+      "step": 12786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.025844716194427e-05,
+      "loss": 0.0436,
+      "step": 12787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.024530870323773e-05,
+      "loss": 0.0424,
+      "step": 12788
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.023217258932659e-05,
+      "loss": 0.0419,
+      "step": 12789
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0219038820652357e-05,
+      "loss": 0.0415,
+      "step": 12790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0205907397656586e-05,
+      "loss": 0.0486,
+      "step": 12791
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0192778320780613e-05,
+      "loss": 0.0449,
+      "step": 12792
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0179651590465818e-05,
+      "loss": 0.0424,
+      "step": 12793
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.016652720715345e-05,
+      "loss": 0.0433,
+      "step": 12794
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.015340517128471e-05,
+      "loss": 0.0496,
+      "step": 12795
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0140285483300634e-05,
+      "loss": 0.0467,
+      "step": 12796
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.012716814364227e-05,
+      "loss": 0.0437,
+      "step": 12797
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0114053152750556e-05,
+      "loss": 0.0493,
+      "step": 12798
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0100940511066376e-05,
+      "loss": 0.049,
+      "step": 12799
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0087830219030468e-05,
+      "loss": 0.0474,
+      "step": 12800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0074722277083534e-05,
+      "loss": 0.0439,
+      "step": 12801
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0061616685666216e-05,
+      "loss": 0.0454,
+      "step": 12802
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.004851344521905e-05,
+      "loss": 0.0455,
+      "step": 12803
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0035412556182562e-05,
+      "loss": 0.0437,
+      "step": 12804
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.002231401899702e-05,
+      "loss": 0.0435,
+      "step": 12805
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0009217834102786e-05,
+      "loss": 0.0452,
+      "step": 12806
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9996124001940086e-05,
+      "loss": 0.0469,
+      "step": 12807
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9983032522949063e-05,
+      "loss": 0.0427,
+      "step": 12808
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.996994339756982e-05,
+      "loss": 0.0426,
+      "step": 12809
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.995685662624229e-05,
+      "loss": 0.0431,
+      "step": 12810
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9943772209406405e-05,
+      "loss": 0.0529,
+      "step": 12811
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9930690147502016e-05,
+      "loss": 0.045,
+      "step": 12812
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.991761044096888e-05,
+      "loss": 0.0476,
+      "step": 12813
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9904533090246622e-05,
+      "loss": 0.0496,
+      "step": 12814
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.989145809577486e-05,
+      "loss": 0.0408,
+      "step": 12815
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9878385457993107e-05,
+      "loss": 0.0421,
+      "step": 12816
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9865315177340847e-05,
+      "loss": 0.0466,
+      "step": 12817
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.985224725425735e-05,
+      "loss": 0.0418,
+      "step": 12818
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9839181689181973e-05,
+      "loss": 0.0462,
+      "step": 12819
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9826118482553835e-05,
+      "loss": 0.0433,
+      "step": 12820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.98130576348121e-05,
+      "loss": 0.0432,
+      "step": 12821
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9799999146395798e-05,
+      "loss": 0.0453,
+      "step": 12822
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.978694301774393e-05,
+      "loss": 0.0445,
+      "step": 12823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9773889249295294e-05,
+      "loss": 0.0437,
+      "step": 12824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9760837841488743e-05,
+      "loss": 0.043,
+      "step": 12825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9747788794762975e-05,
+      "loss": 0.0436,
+      "step": 12826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.973474210955669e-05,
+      "loss": 0.0411,
+      "step": 12827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9721697786308378e-05,
+      "loss": 0.0452,
+      "step": 12828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9708655825456578e-05,
+      "loss": 0.0436,
+      "step": 12829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9695616227439637e-05,
+      "loss": 0.0602,
+      "step": 12830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9682578992695908e-05,
+      "loss": 0.0472,
+      "step": 12831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.966954412166365e-05,
+      "loss": 0.047,
+      "step": 12832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9656511614781036e-05,
+      "loss": 0.0439,
+      "step": 12833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.96434814724861e-05,
+      "loss": 0.0436,
+      "step": 12834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9630453695216886e-05,
+      "loss": 0.0444,
+      "step": 12835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.961742828341132e-05,
+      "loss": 0.0436,
+      "step": 12836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.960440523750727e-05,
+      "loss": 0.0414,
+      "step": 12837
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9591384557942448e-05,
+      "loss": 0.0479,
+      "step": 12838
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9578366245154586e-05,
+      "loss": 0.043,
+      "step": 12839
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9565350299581272e-05,
+      "loss": 0.0486,
+      "step": 12840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9552336721660077e-05,
+      "loss": 0.0452,
+      "step": 12841
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.953932551182842e-05,
+      "loss": 0.0496,
+      "step": 12842
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9526316670523647e-05,
+      "loss": 0.0433,
+      "step": 12843
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9513310198183065e-05,
+      "loss": 0.0437,
+      "step": 12844
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.950030609524389e-05,
+      "loss": 0.0516,
+      "step": 12845
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9487304362143254e-05,
+      "loss": 0.0436,
+      "step": 12846
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9474304999318247e-05,
+      "loss": 0.0455,
+      "step": 12847
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9461308007205767e-05,
+      "loss": 0.0424,
+      "step": 12848
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9448313386242744e-05,
+      "loss": 0.0442,
+      "step": 12849
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.943532113686599e-05,
+      "loss": 0.0444,
+      "step": 12850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.942233125951227e-05,
+      "loss": 0.0416,
+      "step": 12851
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9409343754618167e-05,
+      "loss": 0.0471,
+      "step": 12852
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9396358622620303e-05,
+      "loss": 0.0495,
+      "step": 12853
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9383375863955153e-05,
+      "loss": 0.0475,
+      "step": 12854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9370395479059177e-05,
+      "loss": 0.0472,
+      "step": 12855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9357417468368665e-05,
+      "loss": 0.0511,
+      "step": 12856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9344441832319848e-05,
+      "loss": 0.0463,
+      "step": 12857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9331468571348918e-05,
+      "loss": 0.0457,
+      "step": 12858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9318497685892e-05,
+      "loss": 0.0517,
+      "step": 12859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9305529176385072e-05,
+      "loss": 0.0442,
+      "step": 12860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9292563043264122e-05,
+      "loss": 0.0419,
+      "step": 12861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9279599286964944e-05,
+      "loss": 0.0474,
+      "step": 12862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9266637907923334e-05,
+      "loss": 0.0432,
+      "step": 12863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.925367890657499e-05,
+      "loss": 0.0472,
+      "step": 12864
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.924072228335557e-05,
+      "loss": 0.0452,
+      "step": 12865
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.922776803870052e-05,
+      "loss": 0.0463,
+      "step": 12866
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9214816173045356e-05,
+      "loss": 0.0485,
+      "step": 12867
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9201866686825474e-05,
+      "loss": 0.0467,
+      "step": 12868
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9188919580476093e-05,
+      "loss": 0.0426,
+      "step": 12869
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9175974854432507e-05,
+      "loss": 0.0479,
+      "step": 12870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9163032509129783e-05,
+      "loss": 0.0439,
+      "step": 12871
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9150092545003016e-05,
+      "loss": 0.0431,
+      "step": 12872
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9137154962487177e-05,
+      "loss": 0.0423,
+      "step": 12873
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9124219762017145e-05,
+      "loss": 0.0422,
+      "step": 12874
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9111286944027793e-05,
+      "loss": 0.046,
+      "step": 12875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9098356508953772e-05,
+      "loss": 0.0428,
+      "step": 12876
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.908542845722978e-05,
+      "loss": 0.044,
+      "step": 12877
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.907250278929039e-05,
+      "loss": 0.0444,
+      "step": 12878
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9059579505570132e-05,
+      "loss": 0.0444,
+      "step": 12879
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9046658606503342e-05,
+      "loss": 0.0589,
+      "step": 12880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.903374009252444e-05,
+      "loss": 0.0515,
+      "step": 12881
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.90208239640676e-05,
+      "loss": 0.0452,
+      "step": 12882
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9007910221567025e-05,
+      "loss": 0.0466,
+      "step": 12883
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8994998865456856e-05,
+      "loss": 0.0453,
+      "step": 12884
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8982089896171038e-05,
+      "loss": 0.0435,
+      "step": 12885
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8969183314143535e-05,
+      "loss": 0.0466,
+      "step": 12886
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8956279119808194e-05,
+      "loss": 0.0426,
+      "step": 12887
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8943377313598785e-05,
+      "loss": 0.0434,
+      "step": 12888
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.893047789594905e-05,
+      "loss": 0.0439,
+      "step": 12889
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8917580867292526e-05,
+      "loss": 0.0463,
+      "step": 12890
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8904686228062784e-05,
+      "loss": 0.0453,
+      "step": 12891
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8891793978693273e-05,
+      "loss": 0.0424,
+      "step": 12892
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.887890411961738e-05,
+      "loss": 0.047,
+      "step": 12893
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.886601665126839e-05,
+      "loss": 0.044,
+      "step": 12894
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.885313157407946e-05,
+      "loss": 0.0422,
+      "step": 12895
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.884024888848377e-05,
+      "loss": 0.0472,
+      "step": 12896
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.882736859491436e-05,
+      "loss": 0.047,
+      "step": 12897
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.881449069380423e-05,
+      "loss": 0.0432,
+      "step": 12898
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8801615185586216e-05,
+      "loss": 0.0441,
+      "step": 12899
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8788742070693142e-05,
+      "loss": 0.0427,
+      "step": 12900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.877587134955776e-05,
+      "loss": 0.0443,
+      "step": 12901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.876300302261269e-05,
+      "loss": 0.0475,
+      "step": 12902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.875013709029054e-05,
+      "loss": 0.0456,
+      "step": 12903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.873727355302375e-05,
+      "loss": 0.0426,
+      "step": 12904
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.872441241124474e-05,
+      "loss": 0.0445,
+      "step": 12905
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8711553665385837e-05,
+      "loss": 0.0491,
+      "step": 12906
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.869869731587932e-05,
+      "loss": 0.0535,
+      "step": 12907
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8685843363157326e-05,
+      "loss": 0.045,
+      "step": 12908
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8672991807651916e-05,
+      "loss": 0.0439,
+      "step": 12909
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8660142649795097e-05,
+      "loss": 0.0432,
+      "step": 12910
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.864729589001882e-05,
+      "loss": 0.045,
+      "step": 12911
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8634451528754945e-05,
+      "loss": 0.0458,
+      "step": 12912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.862160956643517e-05,
+      "loss": 0.0422,
+      "step": 12913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8608770003491213e-05,
+      "loss": 0.0444,
+      "step": 12914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8595932840354678e-05,
+      "loss": 0.0496,
+      "step": 12915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8583098077457072e-05,
+      "loss": 0.0482,
+      "step": 12916
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.857026571522988e-05,
+      "loss": 0.0442,
+      "step": 12917
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8557435754104378e-05,
+      "loss": 0.0531,
+      "step": 12918
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.85446081945119e-05,
+      "loss": 0.0479,
+      "step": 12919
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8531783036883654e-05,
+      "loss": 0.044,
+      "step": 12920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8518960281650696e-05,
+      "loss": 0.0427,
+      "step": 12921
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8506139929244137e-05,
+      "loss": 0.045,
+      "step": 12922
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8493321980094845e-05,
+      "loss": 0.0416,
+      "step": 12923
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8480506434633757e-05,
+      "loss": 0.0459,
+      "step": 12924
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8467693293291642e-05,
+      "loss": 0.0455,
+      "step": 12925
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8454882556499218e-05,
+      "loss": 0.047,
+      "step": 12926
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8442074224687153e-05,
+      "loss": 0.0463,
+      "step": 12927
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8429268298285927e-05,
+      "loss": 0.0442,
+      "step": 12928
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8416464777726048e-05,
+      "loss": 0.0441,
+      "step": 12929
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8403663663437894e-05,
+      "loss": 0.044,
+      "step": 12930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8390864955851827e-05,
+      "loss": 0.0443,
+      "step": 12931
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.837806865539798e-05,
+      "loss": 0.0449,
+      "step": 12932
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8365274762506587e-05,
+      "loss": 0.0448,
+      "step": 12933
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8352483277607646e-05,
+      "loss": 0.0428,
+      "step": 12934
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.833969420113116e-05,
+      "loss": 0.0525,
+      "step": 12935
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8326907533507074e-05,
+      "loss": 0.0493,
+      "step": 12936
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8314123275165138e-05,
+      "loss": 0.0424,
+      "step": 12937
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8301341426535145e-05,
+      "loss": 0.0461,
+      "step": 12938
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8288561988046737e-05,
+      "loss": 0.0487,
+      "step": 12939
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8275784960129492e-05,
+      "loss": 0.0453,
+      "step": 12940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8263010343212947e-05,
+      "loss": 0.0434,
+      "step": 12941
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.825023813772646e-05,
+      "loss": 0.0425,
+      "step": 12942
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8237468344099394e-05,
+      "loss": 0.0441,
+      "step": 12943
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8224700962761007e-05,
+      "loss": 0.0454,
+      "step": 12944
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8211935994140492e-05,
+      "loss": 0.043,
+      "step": 12945
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8199173438666927e-05,
+      "loss": 0.0433,
+      "step": 12946
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8186413296769276e-05,
+      "loss": 0.0439,
+      "step": 12947
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8173655568876524e-05,
+      "loss": 0.0476,
+      "step": 12948
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.816090025541749e-05,
+      "loss": 0.0418,
+      "step": 12949
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8148147356821e-05,
+      "loss": 0.0416,
+      "step": 12950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8135396873515663e-05,
+      "loss": 0.0452,
+      "step": 12951
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.812264880593013e-05,
+      "loss": 0.0425,
+      "step": 12952
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8109903154492912e-05,
+      "loss": 0.0435,
+      "step": 12953
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8097159919632453e-05,
+      "loss": 0.0453,
+      "step": 12954
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.808441910177716e-05,
+      "loss": 0.0451,
+      "step": 12955
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.807168070135524e-05,
+      "loss": 0.0446,
+      "step": 12956
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.805894471879492e-05,
+      "loss": 0.0431,
+      "step": 12957
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.804621115452434e-05,
+      "loss": 0.0504,
+      "step": 12958
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8033480008971546e-05,
+      "loss": 0.0425,
+      "step": 12959
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8020751282564472e-05,
+      "loss": 0.0437,
+      "step": 12960
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.800802497573095e-05,
+      "loss": 0.0494,
+      "step": 12961
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7995301088898827e-05,
+      "loss": 0.0451,
+      "step": 12962
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7982579622495797e-05,
+      "loss": 0.0464,
+      "step": 12963
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7969860576949535e-05,
+      "loss": 0.0437,
+      "step": 12964
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7957143952687505e-05,
+      "loss": 0.0466,
+      "step": 12965
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7944429750137236e-05,
+      "loss": 0.0416,
+      "step": 12966
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.793171796972609e-05,
+      "loss": 0.0421,
+      "step": 12967
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.791900861188138e-05,
+      "loss": 0.0461,
+      "step": 12968
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7906301677030365e-05,
+      "loss": 0.0443,
+      "step": 12969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7893597165600117e-05,
+      "loss": 0.0437,
+      "step": 12970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7880895078017764e-05,
+      "loss": 0.042,
+      "step": 12971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.786819541471023e-05,
+      "loss": 0.0428,
+      "step": 12972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7855498176104434e-05,
+      "loss": 0.0452,
+      "step": 12973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7842803362627224e-05,
+      "loss": 0.0434,
+      "step": 12974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7830110974705272e-05,
+      "loss": 0.0484,
+      "step": 12975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7817421012765264e-05,
+      "loss": 0.0438,
+      "step": 12976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.780473347723377e-05,
+      "loss": 0.0454,
+      "step": 12977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7792048368537305e-05,
+      "loss": 0.0458,
+      "step": 12978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7779365687102232e-05,
+      "loss": 0.043,
+      "step": 12979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.77666854333549e-05,
+      "loss": 0.049,
+      "step": 12980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7754007607721553e-05,
+      "loss": 0.0506,
+      "step": 12981
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7741332210628345e-05,
+      "loss": 0.045,
+      "step": 12982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7728659242501408e-05,
+      "loss": 0.0456,
+      "step": 12983
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7715988703766703e-05,
+      "loss": 0.0419,
+      "step": 12984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.770332059485011e-05,
+      "loss": 0.0463,
+      "step": 12985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7690654916177517e-05,
+      "loss": 0.0431,
+      "step": 12986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7677991668174653e-05,
+      "loss": 0.0462,
+      "step": 12987
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7665330851267248e-05,
+      "loss": 0.0428,
+      "step": 12988
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7652672465880803e-05,
+      "loss": 0.0421,
+      "step": 12989
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.764001651244088e-05,
+      "loss": 0.0418,
+      "step": 12990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7627362991372906e-05,
+      "loss": 0.0449,
+      "step": 12991
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7614711903102254e-05,
+      "loss": 0.0448,
+      "step": 12992
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7602063248054134e-05,
+      "loss": 0.0454,
+      "step": 12993
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7589417026653752e-05,
+      "loss": 0.0449,
+      "step": 12994
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7576773239326204e-05,
+      "loss": 0.047,
+      "step": 12995
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7564131886496515e-05,
+      "loss": 0.0513,
+      "step": 12996
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7551492968589667e-05,
+      "loss": 0.0429,
+      "step": 12997
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.753885648603046e-05,
+      "loss": 0.0435,
+      "step": 12998
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7526222439243654e-05,
+      "loss": 0.0424,
+      "step": 12999
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.751359082865397e-05,
+      "loss": 0.0439,
+      "step": 13000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.750096165468602e-05,
+      "loss": 0.043,
+      "step": 13001
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7488334917764348e-05,
+      "loss": 0.0434,
+      "step": 13002
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7475710618313356e-05,
+      "loss": 0.0472,
+      "step": 13003
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7463088756757428e-05,
+      "loss": 0.0455,
+      "step": 13004
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7450469333520855e-05,
+      "loss": 0.0446,
+      "step": 13005
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.743785234902786e-05,
+      "loss": 0.0441,
+      "step": 13006
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7425237803702497e-05,
+      "loss": 0.0424,
+      "step": 13007
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.741262569796885e-05,
+      "loss": 0.0501,
+      "step": 13008
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7400016032250863e-05,
+      "loss": 0.0443,
+      "step": 13009
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7387408806972436e-05,
+      "loss": 0.0428,
+      "step": 13010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.73748040225573e-05,
+      "loss": 0.0447,
+      "step": 13011
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7362201679429223e-05,
+      "loss": 0.05,
+      "step": 13012
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.734960177801177e-05,
+      "loss": 0.0431,
+      "step": 13013
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.733700431872852e-05,
+      "loss": 0.0483,
+      "step": 13014
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.732440930200294e-05,
+      "loss": 0.0468,
+      "step": 13015
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7311816728258432e-05,
+      "loss": 0.0431,
+      "step": 13016
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7299226597918235e-05,
+      "loss": 0.0439,
+      "step": 13017
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7286638911405592e-05,
+      "loss": 0.0453,
+      "step": 13018
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.727405366914364e-05,
+      "loss": 0.0461,
+      "step": 13019
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.726147087155546e-05,
+      "loss": 0.0432,
+      "step": 13020
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7248890519063954e-05,
+      "loss": 0.0432,
+      "step": 13021
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.723631261209205e-05,
+      "loss": 0.0475,
+      "step": 13022
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.722373715106258e-05,
+      "loss": 0.0418,
+      "step": 13023
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7211164136398203e-05,
+      "loss": 0.0437,
+      "step": 13024
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.71985935685216e-05,
+      "loss": 0.0423,
+      "step": 13025
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7186025447855344e-05,
+      "loss": 0.0449,
+      "step": 13026
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.717345977482185e-05,
+      "loss": 0.0466,
+      "step": 13027
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7160896549843562e-05,
+      "loss": 0.046,
+      "step": 13028
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.714833577334277e-05,
+      "loss": 0.0427,
+      "step": 13029
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.713577744574175e-05,
+      "loss": 0.0491,
+      "step": 13030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7123221567462576e-05,
+      "loss": 0.0451,
+      "step": 13031
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7110668138927342e-05,
+      "loss": 0.0457,
+      "step": 13032
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7098117160558033e-05,
+      "loss": 0.0493,
+      "step": 13033
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7085568632776583e-05,
+      "loss": 0.0452,
+      "step": 13034
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7073022556004746e-05,
+      "loss": 0.0443,
+      "step": 13035
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7060478930664325e-05,
+      "loss": 0.0498,
+      "step": 13036
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.704793775717689e-05,
+      "loss": 0.0456,
+      "step": 13037
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.703539903596406e-05,
+      "loss": 0.0463,
+      "step": 13038
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7022862767447322e-05,
+      "loss": 0.0464,
+      "step": 13039
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7010328952048103e-05,
+      "loss": 0.044,
+      "step": 13040
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6997797590187668e-05,
+      "loss": 0.0436,
+      "step": 13041
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6985268682287292e-05,
+      "loss": 0.0471,
+      "step": 13042
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.697274222876812e-05,
+      "loss": 0.0443,
+      "step": 13043
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.696021823005128e-05,
+      "loss": 0.0479,
+      "step": 13044
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.694769668655769e-05,
+      "loss": 0.0514,
+      "step": 13045
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6935177598708284e-05,
+      "loss": 0.0442,
+      "step": 13046
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6922660966923898e-05,
+      "loss": 0.0493,
+      "step": 13047
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.691014679162528e-05,
+      "loss": 0.0444,
+      "step": 13048
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.689763507323315e-05,
+      "loss": 0.0452,
+      "step": 13049
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6885125812167956e-05,
+      "loss": 0.0451,
+      "step": 13050
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6872619008850274e-05,
+      "loss": 0.0454,
+      "step": 13051
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6860114663700508e-05,
+      "loss": 0.0427,
+      "step": 13052
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6847612777138997e-05,
+      "loss": 0.0467,
+      "step": 13053
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6835113349586017e-05,
+      "loss": 0.0545,
+      "step": 13054
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6822616381461664e-05,
+      "loss": 0.0471,
+      "step": 13055
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6810121873186065e-05,
+      "loss": 0.0417,
+      "step": 13056
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6797629825179214e-05,
+      "loss": 0.0449,
+      "step": 13057
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6785140237861073e-05,
+      "loss": 0.0464,
+      "step": 13058
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.67726531116514e-05,
+      "loss": 0.0429,
+      "step": 13059
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6760168446970004e-05,
+      "loss": 0.0413,
+      "step": 13060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6747686244236516e-05,
+      "loss": 0.0439,
+      "step": 13061
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.673520650387059e-05,
+      "loss": 0.0428,
+      "step": 13062
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6722729226291642e-05,
+      "loss": 0.0497,
+      "step": 13063
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.6710254411919188e-05,
+      "loss": 0.0463,
+      "step": 13064
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.669778206117247e-05,
+      "loss": 0.0481,
+      "step": 13065
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6685312174470813e-05,
+      "loss": 0.0415,
+      "step": 13066
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.667284475223335e-05,
+      "loss": 0.0435,
+      "step": 13067
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.666037979487923e-05,
+      "loss": 0.0431,
+      "step": 13068
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6647917302827397e-05,
+      "loss": 0.0421,
+      "step": 13069
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.66354572764968e-05,
+      "loss": 0.0477,
+      "step": 13070
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6622999716306286e-05,
+      "loss": 0.0466,
+      "step": 13071
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6610544622674637e-05,
+      "loss": 0.0429,
+      "step": 13072
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.659809199602048e-05,
+      "loss": 0.045,
+      "step": 13073
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6585641836762433e-05,
+      "loss": 0.0418,
+      "step": 13074
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6573194145319034e-05,
+      "loss": 0.0429,
+      "step": 13075
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6560748922108648e-05,
+      "loss": 0.0447,
+      "step": 13076
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6548306167549664e-05,
+      "loss": 0.0427,
+      "step": 13077
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6535865882060364e-05,
+      "loss": 0.0447,
+      "step": 13078
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6523428066058864e-05,
+      "loss": 0.0509,
+      "step": 13079
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.651099271996329e-05,
+      "loss": 0.0423,
+      "step": 13080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6498559844191662e-05,
+      "loss": 0.0442,
+      "step": 13081
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6486129439161934e-05,
+      "loss": 0.0454,
+      "step": 13082
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6473701505291894e-05,
+      "loss": 0.049,
+      "step": 13083
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6461276042999337e-05,
+      "loss": 0.0449,
+      "step": 13084
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6448853052701938e-05,
+      "loss": 0.0421,
+      "step": 13085
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6436432534817325e-05,
+      "loss": 0.042,
+      "step": 13086
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.642401448976296e-05,
+      "loss": 0.0416,
+      "step": 13087
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6411598917956325e-05,
+      "loss": 0.0427,
+      "step": 13088
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.639918581981472e-05,
+      "loss": 0.0429,
+      "step": 13089
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.638677519575543e-05,
+      "loss": 0.0448,
+      "step": 13090
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6374367046195634e-05,
+      "loss": 0.0441,
+      "step": 13091
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.636196137155247e-05,
+      "loss": 0.0472,
+      "step": 13092
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6349558172242884e-05,
+      "loss": 0.0447,
+      "step": 13093
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6337157448683837e-05,
+      "loss": 0.042,
+      "step": 13094
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6324759201292193e-05,
+      "loss": 0.0492,
+      "step": 13095
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6312363430484734e-05,
+      "loss": 0.0468,
+      "step": 13096
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6299970136678077e-05,
+      "loss": 0.0442,
+      "step": 13097
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6287579320288868e-05,
+      "loss": 0.043,
+      "step": 13098
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.627519098173361e-05,
+      "loss": 0.0424,
+      "step": 13099
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.626280512142878e-05,
+      "loss": 0.0456,
+      "step": 13100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6250421739790674e-05,
+      "loss": 0.0427,
+      "step": 13101
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6238040837235545e-05,
+      "loss": 0.0439,
+      "step": 13102
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6225662414179607e-05,
+      "loss": 0.0447,
+      "step": 13103
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6213286471038956e-05,
+      "loss": 0.0426,
+      "step": 13104
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.62009130082296e-05,
+      "loss": 0.0431,
+      "step": 13105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6188542026167507e-05,
+      "loss": 0.0438,
+      "step": 13106
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6176173525268466e-05,
+      "loss": 0.0529,
+      "step": 13107
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6163807505948267e-05,
+      "loss": 0.0466,
+      "step": 13108
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.61514439686226e-05,
+      "loss": 0.0529,
+      "step": 13109
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.61390829137071e-05,
+      "loss": 0.045,
+      "step": 13110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6126724341617204e-05,
+      "loss": 0.0448,
+      "step": 13111
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6114368252768395e-05,
+      "loss": 0.0459,
+      "step": 13112
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6102014647576035e-05,
+      "loss": 0.0477,
+      "step": 13113
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.608966352645533e-05,
+      "loss": 0.0494,
+      "step": 13114
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6077314889821525e-05,
+      "loss": 0.0453,
+      "step": 13115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6064968738089668e-05,
+      "loss": 0.0433,
+      "step": 13116
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6052625071674786e-05,
+      "loss": 0.0441,
+      "step": 13117
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.604028389099181e-05,
+      "loss": 0.0481,
+      "step": 13118
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6027945196455616e-05,
+      "loss": 0.0429,
+      "step": 13119
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6015608988480955e-05,
+      "loss": 0.044,
+      "step": 13120
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.6003275267482485e-05,
+      "loss": 0.0448,
+      "step": 13121
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.599094403387481e-05,
+      "loss": 0.0439,
+      "step": 13122
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.597861528807245e-05,
+      "loss": 0.0462,
+      "step": 13123
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5966289030489865e-05,
+      "loss": 0.0421,
+      "step": 13124
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5953965261541335e-05,
+      "loss": 0.0424,
+      "step": 13125
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5941643981641184e-05,
+      "loss": 0.0432,
+      "step": 13126
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5929325191203537e-05,
+      "loss": 0.0455,
+      "step": 13127
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5917008890642514e-05,
+      "loss": 0.0433,
+      "step": 13128
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5904695080372154e-05,
+      "loss": 0.0522,
+      "step": 13129
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5892383760806328e-05,
+      "loss": 0.0448,
+      "step": 13130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.588007493235891e-05,
+      "loss": 0.0672,
+      "step": 13131
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5867768595443653e-05,
+      "loss": 0.0476,
+      "step": 13132
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5855464750474235e-05,
+      "loss": 0.0424,
+      "step": 13133
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.584316339786429e-05,
+      "loss": 0.0441,
+      "step": 13134
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5830864538027242e-05,
+      "loss": 0.0425,
+      "step": 13135
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5818568171376578e-05,
+      "loss": 0.0429,
+      "step": 13136
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5806274298325607e-05,
+      "loss": 0.0433,
+      "step": 13137
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.579398291928764e-05,
+      "loss": 0.0481,
+      "step": 13138
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.578169403467581e-05,
+      "loss": 0.0444,
+      "step": 13139
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5769407644903175e-05,
+      "loss": 0.0433,
+      "step": 13140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5757123750382772e-05,
+      "loss": 0.0432,
+      "step": 13141
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.574484235152752e-05,
+      "loss": 0.0475,
+      "step": 13142
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.57325634487503e-05,
+      "loss": 0.0445,
+      "step": 13143
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5720287042463787e-05,
+      "loss": 0.0422,
+      "step": 13144
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5708013133080688e-05,
+      "loss": 0.051,
+      "step": 13145
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.56957417210136e-05,
+      "loss": 0.0485,
+      "step": 13146
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5683472806675015e-05,
+      "loss": 0.0418,
+      "step": 13147
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5671206390477376e-05,
+      "loss": 0.0423,
+      "step": 13148
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.565894247283296e-05,
+      "loss": 0.0432,
+      "step": 13149
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5646681054154053e-05,
+      "loss": 0.0439,
+      "step": 13150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.563442213485282e-05,
+      "loss": 0.0429,
+      "step": 13151
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.562216571534137e-05,
+      "loss": 0.0434,
+      "step": 13152
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.560991179603167e-05,
+      "loss": 0.0422,
+      "step": 13153
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.559766037733561e-05,
+      "loss": 0.0445,
+      "step": 13154
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5585411459665042e-05,
+      "loss": 0.0423,
+      "step": 13155
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.557316504343171e-05,
+      "loss": 0.049,
+      "step": 13156
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.556092112904732e-05,
+      "loss": 0.045,
+      "step": 13157
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5548679716923385e-05,
+      "loss": 0.044,
+      "step": 13158
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5536440807471417e-05,
+      "loss": 0.0516,
+      "step": 13159
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5524204401102836e-05,
+      "loss": 0.0478,
+      "step": 13160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5511970498228964e-05,
+      "loss": 0.0436,
+      "step": 13161
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5499739099261076e-05,
+      "loss": 0.0448,
+      "step": 13162
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5487510204610265e-05,
+      "loss": 0.0445,
+      "step": 13163
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5475283814687632e-05,
+      "loss": 0.0418,
+      "step": 13164
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.54630599299042e-05,
+      "loss": 0.056,
+      "step": 13165
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5450838550670808e-05,
+      "loss": 0.0415,
+      "step": 13166
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5438619677398336e-05,
+      "loss": 0.0416,
+      "step": 13167
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.542640331049747e-05,
+      "loss": 0.0435,
+      "step": 13168
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5414189450378888e-05,
+      "loss": 0.0428,
+      "step": 13169
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5401978097453148e-05,
+      "loss": 0.0534,
+      "step": 13170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5389769252130745e-05,
+      "loss": 0.0436,
+      "step": 13171
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5377562914822094e-05,
+      "loss": 0.0417,
+      "step": 13172
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5365359085937466e-05,
+      "loss": 0.0417,
+      "step": 13173
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5353157765887115e-05,
+      "loss": 0.0447,
+      "step": 13174
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5340958955081173e-05,
+      "loss": 0.0465,
+      "step": 13175
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.532876265392976e-05,
+      "loss": 0.0443,
+      "step": 13176
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.531656886284276e-05,
+      "loss": 0.0468,
+      "step": 13177
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5304377582230155e-05,
+      "loss": 0.0439,
+      "step": 13178
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5292188812501673e-05,
+      "loss": 0.0441,
+      "step": 13179
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.5280002554067084e-05,
+      "loss": 0.0462,
+      "step": 13180
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.526781880733604e-05,
+      "loss": 0.0437,
+      "step": 13181
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.525563757271806e-05,
+      "loss": 0.0426,
+      "step": 13182
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5243458850622614e-05,
+      "loss": 0.0437,
+      "step": 13183
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.523128264145912e-05,
+      "loss": 0.048,
+      "step": 13184
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5219108945636848e-05,
+      "loss": 0.0428,
+      "step": 13185
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5206937763565076e-05,
+      "loss": 0.0423,
+      "step": 13186
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.519476909565286e-05,
+      "loss": 0.0464,
+      "step": 13187
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5182602942309274e-05,
+      "loss": 0.0487,
+      "step": 13188
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5170439303943294e-05,
+      "loss": 0.05,
+      "step": 13189
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5158278180963825e-05,
+      "loss": 0.0427,
+      "step": 13190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5146119573779625e-05,
+      "loss": 0.0421,
+      "step": 13191
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5133963482799384e-05,
+      "loss": 0.0423,
+      "step": 13192
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.512180990843175e-05,
+      "loss": 0.049,
+      "step": 13193
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.510965885108527e-05,
+      "loss": 0.0448,
+      "step": 13194
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5097510311168425e-05,
+      "loss": 0.0471,
+      "step": 13195
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.508536428908953e-05,
+      "loss": 0.0432,
+      "step": 13196
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.50732207852569e-05,
+      "loss": 0.0439,
+      "step": 13197
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5061079800078735e-05,
+      "loss": 0.0434,
+      "step": 13198
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5048941333963162e-05,
+      "loss": 0.0451,
+      "step": 13199
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.503680538731823e-05,
+      "loss": 0.043,
+      "step": 13200
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5024671960551837e-05,
+      "loss": 0.0463,
+      "step": 13201
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.501254105407188e-05,
+      "loss": 0.0445,
+      "step": 13202
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.5000412668286134e-05,
+      "loss": 0.0447,
+      "step": 13203
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4988286803602323e-05,
+      "loss": 0.0427,
+      "step": 13204
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4976163460428027e-05,
+      "loss": 0.0499,
+      "step": 13205
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4964042639170725e-05,
+      "loss": 0.044,
+      "step": 13206
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.495192434023792e-05,
+      "loss": 0.0456,
+      "step": 13207
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.493980856403694e-05,
+      "loss": 0.0456,
+      "step": 13208
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4927695310975087e-05,
+      "loss": 0.0423,
+      "step": 13209
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4915584581459504e-05,
+      "loss": 0.047,
+      "step": 13210
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4903476375897305e-05,
+      "loss": 0.0421,
+      "step": 13211
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4891370694695517e-05,
+      "loss": 0.0431,
+      "step": 13212
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4879267538261063e-05,
+      "loss": 0.042,
+      "step": 13213
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4867166907000815e-05,
+      "loss": 0.0418,
+      "step": 13214
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.485506880132149e-05,
+      "loss": 0.0436,
+      "step": 13215
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4842973221629784e-05,
+      "loss": 0.043,
+      "step": 13216
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4830880168332316e-05,
+      "loss": 0.0501,
+      "step": 13217
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4818789641835536e-05,
+      "loss": 0.0444,
+      "step": 13218
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4806701642545926e-05,
+      "loss": 0.0438,
+      "step": 13219
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4794616170869766e-05,
+      "loss": 0.0441,
+      "step": 13220
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4782533227213334e-05,
+      "loss": 0.0457,
+      "step": 13221
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4770452811982793e-05,
+      "loss": 0.0436,
+      "step": 13222
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.475837492558426e-05,
+      "loss": 0.0415,
+      "step": 13223
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.474629956842368e-05,
+      "loss": 0.0685,
+      "step": 13224
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4734226740906975e-05,
+      "loss": 0.051,
+      "step": 13225
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.472215644343998e-05,
+      "loss": 0.0431,
+      "step": 13226
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4710088676428446e-05,
+      "loss": 0.0431,
+      "step": 13227
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.469802344027804e-05,
+      "loss": 0.0447,
+      "step": 13228
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.468596073539432e-05,
+      "loss": 0.0469,
+      "step": 13229
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4673900562182738e-05,
+      "loss": 0.0446,
+      "step": 13230
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.466184292104873e-05,
+      "loss": 0.0431,
+      "step": 13231
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4649787812397594e-05,
+      "loss": 0.0492,
+      "step": 13232
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4637735236634617e-05,
+      "loss": 0.0422,
+      "step": 13233
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.462568519416486e-05,
+      "loss": 0.0517,
+      "step": 13234
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4613637685393432e-05,
+      "loss": 0.0427,
+      "step": 13235
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.460159271072531e-05,
+      "loss": 0.0427,
+      "step": 13236
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4589550270565385e-05,
+      "loss": 0.0436,
+      "step": 13237
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4577510365318436e-05,
+      "loss": 0.0453,
+      "step": 13238
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4565472995389203e-05,
+      "loss": 0.0451,
+      "step": 13239
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4553438161182308e-05,
+      "loss": 0.0435,
+      "step": 13240
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4541405863102316e-05,
+      "loss": 0.0427,
+      "step": 13241
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4529376101553715e-05,
+      "loss": 0.0475,
+      "step": 13242
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.451734887694086e-05,
+      "loss": 0.0424,
+      "step": 13243
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4505324189668e-05,
+      "loss": 0.0459,
+      "step": 13244
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4493302040139387e-05,
+      "loss": 0.0477,
+      "step": 13245
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4481282428759143e-05,
+      "loss": 0.0418,
+      "step": 13246
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4469265355931326e-05,
+      "loss": 0.0499,
+      "step": 13247
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.445725082205984e-05,
+      "loss": 0.044,
+      "step": 13248
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4445238827548565e-05,
+      "loss": 0.0449,
+      "step": 13249
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4433229372801316e-05,
+      "loss": 0.0461,
+      "step": 13250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.442122245822178e-05,
+      "loss": 0.0431,
+      "step": 13251
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4409218084213526e-05,
+      "loss": 0.0444,
+      "step": 13252
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.439721625118012e-05,
+      "loss": 0.0425,
+      "step": 13253
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4385216959524983e-05,
+      "loss": 0.044,
+      "step": 13254
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4373220209651514e-05,
+      "loss": 0.0453,
+      "step": 13255
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4361226001962912e-05,
+      "loss": 0.0465,
+      "step": 13256
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4349234336862425e-05,
+      "loss": 0.044,
+      "step": 13257
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4337245214753103e-05,
+      "loss": 0.0472,
+      "step": 13258
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4325258636037962e-05,
+      "loss": 0.0444,
+      "step": 13259
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4313274601119952e-05,
+      "loss": 0.0493,
+      "step": 13260
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4301293110401934e-05,
+      "loss": 0.0442,
+      "step": 13261
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4289314164286614e-05,
+      "loss": 0.0453,
+      "step": 13262
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.427733776317669e-05,
+      "loss": 0.043,
+      "step": 13263
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4265363907474737e-05,
+      "loss": 0.045,
+      "step": 13264
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.425339259758329e-05,
+      "loss": 0.0457,
+      "step": 13265
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.424142383390471e-05,
+      "loss": 0.0443,
+      "step": 13266
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4229457616841344e-05,
+      "loss": 0.0459,
+      "step": 13267
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.421749394679548e-05,
+      "loss": 0.0459,
+      "step": 13268
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.420553282416921e-05,
+      "loss": 0.047,
+      "step": 13269
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4193574249364626e-05,
+      "loss": 0.0434,
+      "step": 13270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.418161822278374e-05,
+      "loss": 0.0431,
+      "step": 13271
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4169664744828412e-05,
+      "loss": 0.0425,
+      "step": 13272
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4157713815900485e-05,
+      "loss": 0.0442,
+      "step": 13273
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.414576543640168e-05,
+      "loss": 0.0453,
+      "step": 13274
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4133819606733655e-05,
+      "loss": 0.0444,
+      "step": 13275
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4121876327297933e-05,
+      "loss": 0.045,
+      "step": 13276
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4109935598496015e-05,
+      "loss": 0.0445,
+      "step": 13277
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4097997420729268e-05,
+      "loss": 0.0428,
+      "step": 13278
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4086061794399028e-05,
+      "loss": 0.0445,
+      "step": 13279
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4074128719906465e-05,
+      "loss": 0.0468,
+      "step": 13280
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4062198197652752e-05,
+      "loss": 0.0444,
+      "step": 13281
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4050270228038875e-05,
+      "loss": 0.0414,
+      "step": 13282
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.403834481146583e-05,
+      "loss": 0.048,
+      "step": 13283
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4026421948334478e-05,
+      "loss": 0.049,
+      "step": 13284
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4014501639045638e-05,
+      "loss": 0.0418,
+      "step": 13285
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.400258388399995e-05,
+      "loss": 0.0452,
+      "step": 13286
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3990668683598062e-05,
+      "loss": 0.047,
+      "step": 13287
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3978756038240503e-05,
+      "loss": 0.043,
+      "step": 13288
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3966845948327722e-05,
+      "loss": 0.0437,
+      "step": 13289
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3954938414260054e-05,
+      "loss": 0.0421,
+      "step": 13290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3943033436437766e-05,
+      "loss": 0.0433,
+      "step": 13291
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3931131015261054e-05,
+      "loss": 0.0439,
+      "step": 13292
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.391923115113002e-05,
+      "loss": 0.0438,
+      "step": 13293
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3907333844444734e-05,
+      "loss": 0.0444,
+      "step": 13294
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.389543909560501e-05,
+      "loss": 0.0432,
+      "step": 13295
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3883546905010735e-05,
+      "loss": 0.0446,
+      "step": 13296
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3871657273061667e-05,
+      "loss": 0.0437,
+      "step": 13297
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3859770200157473e-05,
+      "loss": 0.0465,
+      "step": 13298
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.384788568669778e-05,
+      "loss": 0.0441,
+      "step": 13299
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3836003733082013e-05,
+      "loss": 0.0424,
+      "step": 13300
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3824124339709608e-05,
+      "loss": 0.0439,
+      "step": 13301
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.381224750697989e-05,
+      "loss": 0.0426,
+      "step": 13302
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.380037323529214e-05,
+      "loss": 0.0444,
+      "step": 13303
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3788501525045438e-05,
+      "loss": 0.0458,
+      "step": 13304
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3776632376638885e-05,
+      "loss": 0.0423,
+      "step": 13305
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3764765790471466e-05,
+      "loss": 0.0433,
+      "step": 13306
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3752901766942093e-05,
+      "loss": 0.0449,
+      "step": 13307
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3741040306449525e-05,
+      "loss": 0.0428,
+      "step": 13308
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3729181409392532e-05,
+      "loss": 0.0453,
+      "step": 13309
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3717325076169705e-05,
+      "loss": 0.0466,
+      "step": 13310
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3705471307179605e-05,
+      "loss": 0.0425,
+      "step": 13311
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.36936201028207e-05,
+      "loss": 0.0437,
+      "step": 13312
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.368177146349141e-05,
+      "loss": 0.0443,
+      "step": 13313
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.366992538958995e-05,
+      "loss": 0.0433,
+      "step": 13314
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3658081881514572e-05,
+      "loss": 0.0553,
+      "step": 13315
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3646240939663366e-05,
+      "loss": 0.0464,
+      "step": 13316
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3634402564434422e-05,
+      "loss": 0.0476,
+      "step": 13317
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3622566756225605e-05,
+      "loss": 0.0447,
+      "step": 13318
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3610733515434823e-05,
+      "loss": 0.0459,
+      "step": 13319
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3598902842459858e-05,
+      "loss": 0.0457,
+      "step": 13320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3587074737698357e-05,
+      "loss": 0.0424,
+      "step": 13321
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3575249201547944e-05,
+      "loss": 0.0451,
+      "step": 13322
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.356342623440615e-05,
+      "loss": 0.043,
+      "step": 13323
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3551605836670353e-05,
+      "loss": 0.0446,
+      "step": 13324
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.353978800873793e-05,
+      "loss": 0.0442,
+      "step": 13325
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3527972751006122e-05,
+      "loss": 0.0433,
+      "step": 13326
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.351616006387214e-05,
+      "loss": 0.0439,
+      "step": 13327
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3504349947732997e-05,
+      "loss": 0.0415,
+      "step": 13328
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3492542402985727e-05,
+      "loss": 0.0441,
+      "step": 13329
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3480737430027235e-05,
+      "loss": 0.0456,
+      "step": 13330
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3468935029254367e-05,
+      "loss": 0.042,
+      "step": 13331
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3457135201063818e-05,
+      "loss": 0.0463,
+      "step": 13332
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3445337945852275e-05,
+      "loss": 0.0441,
+      "step": 13333
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3433543264016256e-05,
+      "loss": 0.0444,
+      "step": 13334
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3421751155952253e-05,
+      "loss": 0.0418,
+      "step": 13335
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3409961622056688e-05,
+      "loss": 0.0439,
+      "step": 13336
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3398174662725857e-05,
+      "loss": 0.0443,
+      "step": 13337
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.338639027835594e-05,
+      "loss": 0.0432,
+      "step": 13338
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3374608469343084e-05,
+      "loss": 0.0445,
+      "step": 13339
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.336282923608335e-05,
+      "loss": 0.042,
+      "step": 13340
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.335105257897271e-05,
+      "loss": 0.0412,
+      "step": 13341
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.333927849840698e-05,
+      "loss": 0.0433,
+      "step": 13342
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3327506994781978e-05,
+      "loss": 0.0459,
+      "step": 13343
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3315738068493397e-05,
+      "loss": 0.0422,
+      "step": 13344
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.330397171993688e-05,
+      "loss": 0.0422,
+      "step": 13345
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3292207949507914e-05,
+      "loss": 0.0423,
+      "step": 13346
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3280446757601925e-05,
+      "loss": 0.0421,
+      "step": 13347
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.326868814461428e-05,
+      "loss": 0.0428,
+      "step": 13348
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3256932110940254e-05,
+      "loss": 0.046,
+      "step": 13349
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.324517865697501e-05,
+      "loss": 0.0429,
+      "step": 13350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.323342778311367e-05,
+      "loss": 0.0465,
+      "step": 13351
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.322167948975118e-05,
+      "loss": 0.0434,
+      "step": 13352
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.32099337772825e-05,
+      "loss": 0.045,
+      "step": 13353
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3198190646102458e-05,
+      "loss": 0.0534,
+      "step": 13354
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3186450096605804e-05,
+      "loss": 0.0434,
+      "step": 13355
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.317471212918717e-05,
+      "loss": 0.0431,
+      "step": 13356
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3162976744241128e-05,
+      "loss": 0.0417,
+      "step": 13357
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.315124394216218e-05,
+      "loss": 0.0431,
+      "step": 13358
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.313951372334474e-05,
+      "loss": 0.0466,
+      "step": 13359
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.31277860881831e-05,
+      "loss": 0.0443,
+      "step": 13360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3116061037071444e-05,
+      "loss": 0.0455,
+      "step": 13361
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.310433857040394e-05,
+      "loss": 0.0433,
+      "step": 13362
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.309261868857463e-05,
+      "loss": 0.0542,
+      "step": 13363
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3080901391977494e-05,
+      "loss": 0.0454,
+      "step": 13364
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3069186681006427e-05,
+      "loss": 0.0528,
+      "step": 13365
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3057474556055157e-05,
+      "loss": 0.042,
+      "step": 13366
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3045765017517417e-05,
+      "loss": 0.0443,
+      "step": 13367
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.303405806578682e-05,
+      "loss": 0.0426,
+      "step": 13368
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3022353701256927e-05,
+      "loss": 0.0421,
+      "step": 13369
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3010651924321115e-05,
+      "loss": 0.0444,
+      "step": 13370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.299895273537279e-05,
+      "loss": 0.0435,
+      "step": 13371
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.298725613480519e-05,
+      "loss": 0.0455,
+      "step": 13372
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2975562123011495e-05,
+      "loss": 0.0442,
+      "step": 13373
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2963870700384825e-05,
+      "loss": 0.0474,
+      "step": 13374
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2952181867318145e-05,
+      "loss": 0.0429,
+      "step": 13375
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2940495624204405e-05,
+      "loss": 0.0476,
+      "step": 13376
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2928811971436415e-05,
+      "loss": 0.0431,
+      "step": 13377
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2917130909406936e-05,
+      "loss": 0.0463,
+      "step": 13378
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2905452438508657e-05,
+      "loss": 0.0428,
+      "step": 13379
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2893776559134074e-05,
+      "loss": 0.0435,
+      "step": 13380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2882103271675715e-05,
+      "loss": 0.0442,
+      "step": 13381
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.287043257652597e-05,
+      "loss": 0.0428,
+      "step": 13382
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2858764474077165e-05,
+      "loss": 0.0489,
+      "step": 13383
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2847098964721512e-05,
+      "loss": 0.0431,
+      "step": 13384
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.283543604885111e-05,
+      "loss": 0.0482,
+      "step": 13385
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2823775726858042e-05,
+      "loss": 0.0426,
+      "step": 13386
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.281211799913425e-05,
+      "loss": 0.0513,
+      "step": 13387
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.280046286607166e-05,
+      "loss": 0.0427,
+      "step": 13388
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.278881032806198e-05,
+      "loss": 0.0418,
+      "step": 13389
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2777160385496942e-05,
+      "loss": 0.0493,
+      "step": 13390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2765513038768173e-05,
+      "loss": 0.049,
+      "step": 13391
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2753868288267167e-05,
+      "loss": 0.0432,
+      "step": 13392
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2742226134385414e-05,
+      "loss": 0.0427,
+      "step": 13393
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2730586577514203e-05,
+      "loss": 0.0421,
+      "step": 13394
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2718949618044814e-05,
+      "loss": 0.0478,
+      "step": 13395
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2707315256368433e-05,
+      "loss": 0.0453,
+      "step": 13396
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2695683492876173e-05,
+      "loss": 0.0429,
+      "step": 13397
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2684054327959013e-05,
+      "loss": 0.0442,
+      "step": 13398
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.267242776200782e-05,
+      "loss": 0.043,
+      "step": 13399
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2660803795413455e-05,
+      "loss": 0.0439,
+      "step": 13400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.264918242856666e-05,
+      "loss": 0.0428,
+      "step": 13401
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2637563661858118e-05,
+      "loss": 0.0451,
+      "step": 13402
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2625947495678322e-05,
+      "loss": 0.0423,
+      "step": 13403
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2614333930417796e-05,
+      "loss": 0.0471,
+      "step": 13404
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2602722966466917e-05,
+      "loss": 0.0436,
+      "step": 13405
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.259111460421598e-05,
+      "loss": 0.0454,
+      "step": 13406
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.257950884405523e-05,
+      "loss": 0.0454,
+      "step": 13407
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.2567905686374746e-05,
+      "loss": 0.0419,
+      "step": 13408
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2556305131564593e-05,
+      "loss": 0.0436,
+      "step": 13409
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2544707180014736e-05,
+      "loss": 0.0447,
+      "step": 13410
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2533111832115006e-05,
+      "loss": 0.0485,
+      "step": 13411
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2521519088255215e-05,
+      "loss": 0.0425,
+      "step": 13412
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2509928948825e-05,
+      "loss": 0.0434,
+      "step": 13413
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2498341414214e-05,
+      "loss": 0.0452,
+      "step": 13414
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.248675648481172e-05,
+      "loss": 0.0449,
+      "step": 13415
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2475174161007584e-05,
+      "loss": 0.0425,
+      "step": 13416
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.246359444319096e-05,
+      "loss": 0.0469,
+      "step": 13417
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2452017331751053e-05,
+      "loss": 0.0426,
+      "step": 13418
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2440442827077045e-05,
+      "loss": 0.0498,
+      "step": 13419
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.242887092955801e-05,
+      "loss": 0.0427,
+      "step": 13420
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2417301639582966e-05,
+      "loss": 0.0464,
+      "step": 13421
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.240573495754076e-05,
+      "loss": 0.0426,
+      "step": 13422
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2394170883820253e-05,
+      "loss": 0.0441,
+      "step": 13423
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2382609418810117e-05,
+      "loss": 0.0432,
+      "step": 13424
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.237105056289903e-05,
+      "loss": 0.0471,
+      "step": 13425
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.235949431647556e-05,
+      "loss": 0.0426,
+      "step": 13426
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2347940679928105e-05,
+      "loss": 0.0442,
+      "step": 13427
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2336389653645075e-05,
+      "loss": 0.0448,
+      "step": 13428
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.232484123801475e-05,
+      "loss": 0.044,
+      "step": 13429
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.231329543342534e-05,
+      "loss": 0.0413,
+      "step": 13430
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2301752240264974e-05,
+      "loss": 0.0432,
+      "step": 13431
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2290211658921623e-05,
+      "loss": 0.0437,
+      "step": 13432
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.227867368978326e-05,
+      "loss": 0.0465,
+      "step": 13433
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2267138333237713e-05,
+      "loss": 0.0422,
+      "step": 13434
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.225560558967278e-05,
+      "loss": 0.0459,
+      "step": 13435
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.22440754594761e-05,
+      "loss": 0.0433,
+      "step": 13436
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.223254794303523e-05,
+      "loss": 0.0428,
+      "step": 13437
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.222102304073771e-05,
+      "loss": 0.0421,
+      "step": 13438
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2209500752970926e-05,
+      "loss": 0.0494,
+      "step": 13439
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2197981080122234e-05,
+      "loss": 0.0431,
+      "step": 13440
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2186464022578812e-05,
+      "loss": 0.0478,
+      "step": 13441
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2174949580727832e-05,
+      "loss": 0.0443,
+      "step": 13442
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2163437754956353e-05,
+      "loss": 0.0426,
+      "step": 13443
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2151928545651334e-05,
+      "loss": 0.0431,
+      "step": 13444
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2140421953199697e-05,
+      "loss": 0.0417,
+      "step": 13445
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2128917977988172e-05,
+      "loss": 0.0434,
+      "step": 13446
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.21174166204035e-05,
+      "loss": 0.0462,
+      "step": 13447
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2105917880832284e-05,
+      "loss": 0.0452,
+      "step": 13448
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2094421759661076e-05,
+      "loss": 0.0417,
+      "step": 13449
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2082928257276315e-05,
+      "loss": 0.0425,
+      "step": 13450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2071437374064296e-05,
+      "loss": 0.0444,
+      "step": 13451
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2059949110411327e-05,
+      "loss": 0.0445,
+      "step": 13452
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2048463466703597e-05,
+      "loss": 0.0476,
+      "step": 13453
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2036980443327193e-05,
+      "loss": 0.0422,
+      "step": 13454
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.202550004066808e-05,
+      "loss": 0.0426,
+      "step": 13455
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.20140222591122e-05,
+      "loss": 0.0443,
+      "step": 13456
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2002547099045355e-05,
+      "loss": 0.0462,
+      "step": 13457
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1991074560853307e-05,
+      "loss": 0.043,
+      "step": 13458
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1979604644921724e-05,
+      "loss": 0.0444,
+      "step": 13459
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1968137351636108e-05,
+      "loss": 0.0439,
+      "step": 13460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1956672681381952e-05,
+      "loss": 0.0429,
+      "step": 13461
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.194521063454468e-05,
+      "loss": 0.0445,
+      "step": 13462
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1933751211509523e-05,
+      "loss": 0.0412,
+      "step": 13463
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.192229441266174e-05,
+      "loss": 0.0418,
+      "step": 13464
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.1910840238386398e-05,
+      "loss": 0.0426,
+      "step": 13465
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1899388689068566e-05,
+      "loss": 0.0416,
+      "step": 13466
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.188793976509318e-05,
+      "loss": 0.0476,
+      "step": 13467
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1876493466845106e-05,
+      "loss": 0.0435,
+      "step": 13468
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.186504979470908e-05,
+      "loss": 0.0444,
+      "step": 13469
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1853608749069798e-05,
+      "loss": 0.0418,
+      "step": 13470
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1842170330311838e-05,
+      "loss": 0.0424,
+      "step": 13471
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1830734538819708e-05,
+      "loss": 0.0414,
+      "step": 13472
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1819301374977863e-05,
+      "loss": 0.0477,
+      "step": 13473
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1807870839170553e-05,
+      "loss": 0.051,
+      "step": 13474
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1796442931782068e-05,
+      "loss": 0.0427,
+      "step": 13475
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1785017653196517e-05,
+      "loss": 0.042,
+      "step": 13476
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1773595003797974e-05,
+      "loss": 0.0502,
+      "step": 13477
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1762174983970452e-05,
+      "loss": 0.0488,
+      "step": 13478
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.175075759409776e-05,
+      "loss": 0.0416,
+      "step": 13479
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1739342834563726e-05,
+      "loss": 0.0422,
+      "step": 13480
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1727930705752054e-05,
+      "loss": 0.0433,
+      "step": 13481
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.17165212080464e-05,
+      "loss": 0.045,
+      "step": 13482
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1705114341830234e-05,
+      "loss": 0.0457,
+      "step": 13483
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.169371010748701e-05,
+      "loss": 0.0428,
+      "step": 13484
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1682308505400097e-05,
+      "loss": 0.0421,
+      "step": 13485
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1670909535952743e-05,
+      "loss": 0.0464,
+      "step": 13486
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.165951319952817e-05,
+      "loss": 0.0522,
+      "step": 13487
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.164811949650942e-05,
+      "loss": 0.044,
+      "step": 13488
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1636728427279474e-05,
+      "loss": 0.0442,
+      "step": 13489
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.162533999222126e-05,
+      "loss": 0.043,
+      "step": 13490
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.16139541917176e-05,
+      "loss": 0.0442,
+      "step": 13491
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1602571026151264e-05,
+      "loss": 0.0452,
+      "step": 13492
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1591190495904824e-05,
+      "loss": 0.0414,
+      "step": 13493
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1579812601360883e-05,
+      "loss": 0.0409,
+      "step": 13494
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.156843734290189e-05,
+      "loss": 0.0455,
+      "step": 13495
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.155706472091026e-05,
+      "loss": 0.0441,
+      "step": 13496
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.154569473576823e-05,
+      "loss": 0.0412,
+      "step": 13497
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.153432738785801e-05,
+      "loss": 0.045,
+      "step": 13498
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1522962677561732e-05,
+      "loss": 0.0448,
+      "step": 13499
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.151160060526144e-05,
+      "loss": 0.0422,
+      "step": 13500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1500241171339007e-05,
+      "loss": 0.0433,
+      "step": 13501
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1488884376176344e-05,
+      "loss": 0.0574,
+      "step": 13502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.147753022015515e-05,
+      "loss": 0.0424,
+      "step": 13503
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1466178703657125e-05,
+      "loss": 0.0427,
+      "step": 13504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1454829827063836e-05,
+      "loss": 0.0461,
+      "step": 13505
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.144348359075682e-05,
+      "loss": 0.0437,
+      "step": 13506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1432139995117408e-05,
+      "loss": 0.0461,
+      "step": 13507
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.142079904052694e-05,
+      "loss": 0.045,
+      "step": 13508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.140946072736666e-05,
+      "loss": 0.0443,
+      "step": 13509
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1398125056017714e-05,
+      "loss": 0.0424,
+      "step": 13510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1386792026861103e-05,
+      "loss": 0.0433,
+      "step": 13511
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1375461640277804e-05,
+      "loss": 0.047,
+      "step": 13512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.136413389664872e-05,
+      "loss": 0.0432,
+      "step": 13513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1352808796354585e-05,
+      "loss": 0.043,
+      "step": 13514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1341486339776107e-05,
+      "loss": 0.0451,
+      "step": 13515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1330166527293915e-05,
+      "loss": 0.0481,
+      "step": 13516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1318849359288473e-05,
+      "loss": 0.0415,
+      "step": 13517
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.130753483614023e-05,
+      "loss": 0.0417,
+      "step": 13518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.1296222958229526e-05,
+      "loss": 0.0419,
+      "step": 13519
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.128491372593664e-05,
+      "loss": 0.0505,
+      "step": 13520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.127360713964167e-05,
+      "loss": 0.0418,
+      "step": 13521
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.126230319972471e-05,
+      "loss": 0.0443,
+      "step": 13522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.125100190656575e-05,
+      "loss": 0.0423,
+      "step": 13523
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1239703260544696e-05,
+      "loss": 0.0442,
+      "step": 13524
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1228407262041315e-05,
+      "loss": 0.0454,
+      "step": 13525
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.121711391143535e-05,
+      "loss": 0.0421,
+      "step": 13526
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.120582320910639e-05,
+      "loss": 0.042,
+      "step": 13527
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1194535155433992e-05,
+      "loss": 0.0446,
+      "step": 13528
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1183249750797605e-05,
+      "loss": 0.0435,
+      "step": 13529
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1171966995576607e-05,
+      "loss": 0.0432,
+      "step": 13530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1160686890150218e-05,
+      "loss": 0.053,
+      "step": 13531
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1149409434897648e-05,
+      "loss": 0.0413,
+      "step": 13532
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.113813463019797e-05,
+      "loss": 0.0441,
+      "step": 13533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.112686247643024e-05,
+      "loss": 0.0467,
+      "step": 13534
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1115592973973285e-05,
+      "loss": 0.043,
+      "step": 13535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1104326123205985e-05,
+      "loss": 0.0474,
+      "step": 13536
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1093061924507053e-05,
+      "loss": 0.0422,
+      "step": 13537
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1081800378255135e-05,
+      "loss": 0.0416,
+      "step": 13538
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1070541484828855e-05,
+      "loss": 0.0425,
+      "step": 13539
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1059285244606553e-05,
+      "loss": 0.0418,
+      "step": 13540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1048031657966673e-05,
+      "loss": 0.0433,
+      "step": 13541
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1036780725287496e-05,
+      "loss": 0.0458,
+      "step": 13542
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.1025532446947226e-05,
+      "loss": 0.042,
+      "step": 13543
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.101428682332399e-05,
+      "loss": 0.0418,
+      "step": 13544
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.100304385479578e-05,
+      "loss": 0.0492,
+      "step": 13545
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0991803541740517e-05,
+      "loss": 0.0532,
+      "step": 13546
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0980565884536064e-05,
+      "loss": 0.0451,
+      "step": 13547
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.09693308835602e-05,
+      "loss": 0.0427,
+      "step": 13548
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0958098539190542e-05,
+      "loss": 0.0427,
+      "step": 13549
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.094686885180467e-05,
+      "loss": 0.0442,
+      "step": 13550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.093564182178007e-05,
+      "loss": 0.0452,
+      "step": 13551
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0924417449494184e-05,
+      "loss": 0.0464,
+      "step": 13552
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0913195735324255e-05,
+      "loss": 0.0478,
+      "step": 13553
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.090197667964755e-05,
+      "loss": 0.0416,
+      "step": 13554
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0890760282841137e-05,
+      "loss": 0.0412,
+      "step": 13555
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0879546545282103e-05,
+      "loss": 0.0444,
+      "step": 13556
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0868335467347366e-05,
+      "loss": 0.0427,
+      "step": 13557
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.085712704941384e-05,
+      "loss": 0.0408,
+      "step": 13558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0845921291858216e-05,
+      "loss": 0.044,
+      "step": 13559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0834718195057212e-05,
+      "loss": 0.0446,
+      "step": 13560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0823517759387424e-05,
+      "loss": 0.0434,
+      "step": 13561
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0812319985225383e-05,
+      "loss": 0.0444,
+      "step": 13562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.080112487294743e-05,
+      "loss": 0.0433,
+      "step": 13563
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0789932422929927e-05,
+      "loss": 0.0426,
+      "step": 13564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0778742635549132e-05,
+      "loss": 0.0453,
+      "step": 13565
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0767555511181136e-05,
+      "loss": 0.0412,
+      "step": 13566
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0756371050202007e-05,
+      "loss": 0.0464,
+      "step": 13567
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.074518925298775e-05,
+      "loss": 0.0462,
+      "step": 13568
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0734010119914192e-05,
+      "loss": 0.0415,
+      "step": 13569
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.072283365135712e-05,
+      "loss": 0.0434,
+      "step": 13570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0711659847692255e-05,
+      "loss": 0.0422,
+      "step": 13571
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.070048870929522e-05,
+      "loss": 0.0419,
+      "step": 13572
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0689320236541476e-05,
+      "loss": 0.0438,
+      "step": 13573
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0678154429806475e-05,
+      "loss": 0.0435,
+      "step": 13574
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0666991289465554e-05,
+      "loss": 0.0425,
+      "step": 13575
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0655830815893995e-05,
+      "loss": 0.0507,
+      "step": 13576
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0644673009466896e-05,
+      "loss": 0.0436,
+      "step": 13577
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0633517870559394e-05,
+      "loss": 0.045,
+      "step": 13578
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0622365399546385e-05,
+      "loss": 0.0428,
+      "step": 13579
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.06112155968028e-05,
+      "loss": 0.0435,
+      "step": 13580
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.060006846270345e-05,
+      "loss": 0.0422,
+      "step": 13581
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0588923997623054e-05,
+      "loss": 0.0437,
+      "step": 13582
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.057778220193619e-05,
+      "loss": 0.0439,
+      "step": 13583
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.05666430760174e-05,
+      "loss": 0.0448,
+      "step": 13584
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.055550662024114e-05,
+      "loss": 0.0446,
+      "step": 13585
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0544372834981785e-05,
+      "loss": 0.0464,
+      "step": 13586
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0533241720613528e-05,
+      "loss": 0.0445,
+      "step": 13587
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.052211327751059e-05,
+      "loss": 0.0425,
+      "step": 13588
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.051098750604703e-05,
+      "loss": 0.0419,
+      "step": 13589
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.049986440659688e-05,
+      "loss": 0.0448,
+      "step": 13590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.048874397953401e-05,
+      "loss": 0.042,
+      "step": 13591
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0477626225232193e-05,
+      "loss": 0.0458,
+      "step": 13592
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0466511144065205e-05,
+      "loss": 0.0436,
+      "step": 13593
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0455398736406662e-05,
+      "loss": 0.0435,
+      "step": 13594
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.04442890026301e-05,
+      "loss": 0.0427,
+      "step": 13595
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0433181943109015e-05,
+      "loss": 0.0427,
+      "step": 13596
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.04220775582167e-05,
+      "loss": 0.0438,
+      "step": 13597
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0410975848326465e-05,
+      "loss": 0.0439,
+      "step": 13598
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.039987681381148e-05,
+      "loss": 0.0463,
+      "step": 13599
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0388780455044885e-05,
+      "loss": 0.0436,
+      "step": 13600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0377686772399605e-05,
+      "loss": 0.0442,
+      "step": 13601
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.036659576624861e-05,
+      "loss": 0.0428,
+      "step": 13602
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.035550743696468e-05,
+      "loss": 0.0482,
+      "step": 13603
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.034442178492062e-05,
+      "loss": 0.0437,
+      "step": 13604
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.033333881048901e-05,
+      "loss": 0.0502,
+      "step": 13605
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0322258514042404e-05,
+      "loss": 0.0472,
+      "step": 13606
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0311180895953276e-05,
+      "loss": 0.0424,
+      "step": 13607
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0300105956593995e-05,
+      "loss": 0.042,
+      "step": 13608
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0289033696336847e-05,
+      "loss": 0.0427,
+      "step": 13609
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.027796411555406e-05,
+      "loss": 0.0422,
+      "step": 13610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0266897214617687e-05,
+      "loss": 0.0493,
+      "step": 13611
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0255832993899747e-05,
+      "loss": 0.0435,
+      "step": 13612
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0244771453772172e-05,
+      "loss": 0.0438,
+      "step": 13613
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0233712594606834e-05,
+      "loss": 0.0447,
+      "step": 13614
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0222656416775397e-05,
+      "loss": 0.0481,
+      "step": 13615
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0211602920649563e-05,
+      "loss": 0.0417,
+      "step": 13616
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0200552106600902e-05,
+      "loss": 0.041,
+      "step": 13617
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0189503975000844e-05,
+      "loss": 0.0458,
+      "step": 13618
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0178458526220823e-05,
+      "loss": 0.0448,
+      "step": 13619
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.016741576063208e-05,
+      "loss": 0.0418,
+      "step": 13620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0156375678605833e-05,
+      "loss": 0.0426,
+      "step": 13621
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0145338280513194e-05,
+      "loss": 0.0423,
+      "step": 13622
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0134303566725187e-05,
+      "loss": 0.0443,
+      "step": 13623
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0123271537612775e-05,
+      "loss": 0.0521,
+      "step": 13624
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.011224219354674e-05,
+      "loss": 0.0478,
+      "step": 13625
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0101215534897855e-05,
+      "loss": 0.0423,
+      "step": 13626
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0090191562036788e-05,
+      "loss": 0.0423,
+      "step": 13627
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0079170275334125e-05,
+      "loss": 0.0447,
+      "step": 13628
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0068151675160328e-05,
+      "loss": 0.0424,
+      "step": 13629
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0057135761885747e-05,
+      "loss": 0.0429,
+      "step": 13630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0046122535880728e-05,
+      "loss": 0.0434,
+      "step": 13631
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0035111997515466e-05,
+      "loss": 0.0439,
+      "step": 13632
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0024104147160094e-05,
+      "loss": 0.042,
+      "step": 13633
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0013098985184608e-05,
+      "loss": 0.043,
+      "step": 13634
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0002096511958957e-05,
+      "loss": 0.0432,
+      "step": 13635
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9991096727852997e-05,
+      "loss": 0.0488,
+      "step": 13636
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.998009963323648e-05,
+      "loss": 0.0428,
+      "step": 13637
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9969105228479103e-05,
+      "loss": 0.0423,
+      "step": 13638
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9958113513950373e-05,
+      "loss": 0.0422,
+      "step": 13639
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.994712449001982e-05,
+      "loss": 0.0412,
+      "step": 13640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.993613815705684e-05,
+      "loss": 0.045,
+      "step": 13641
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9925154515430744e-05,
+      "loss": 0.0435,
+      "step": 13642
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9914173565510728e-05,
+      "loss": 0.0461,
+      "step": 13643
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.99031953076659e-05,
+      "loss": 0.0551,
+      "step": 13644
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9892219742265317e-05,
+      "loss": 0.0422,
+      "step": 13645
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9881246869677907e-05,
+      "loss": 0.0438,
+      "step": 13646
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.987027669027255e-05,
+      "loss": 0.0432,
+      "step": 13647
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9859309204417964e-05,
+      "loss": 0.0484,
+      "step": 13648
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9848344412482854e-05,
+      "loss": 0.0427,
+      "step": 13649
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.983738231483577e-05,
+      "loss": 0.0463,
+      "step": 13650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9826422911845243e-05,
+      "loss": 0.0428,
+      "step": 13651
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9815466203879662e-05,
+      "loss": 0.0423,
+      "step": 13652
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.98045121913073e-05,
+      "loss": 0.0507,
+      "step": 13653
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.97935608744964e-05,
+      "loss": 0.0432,
+      "step": 13654
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9782612253815102e-05,
+      "loss": 0.0428,
+      "step": 13655
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9771666329631412e-05,
+      "loss": 0.0411,
+      "step": 13656
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9760723102313318e-05,
+      "loss": 0.0443,
+      "step": 13657
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9749782572228614e-05,
+      "loss": 0.045,
+      "step": 13658
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.97388447397451e-05,
+      "loss": 0.0482,
+      "step": 13659
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9727909605230454e-05,
+      "loss": 0.0499,
+      "step": 13660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9716977169052253e-05,
+      "loss": 0.0456,
+      "step": 13661
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9706047431578024e-05,
+      "loss": 0.0546,
+      "step": 13662
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.969512039317509e-05,
+      "loss": 0.0423,
+      "step": 13663
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9684196054210823e-05,
+      "loss": 0.0437,
+      "step": 13664
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.967327441505242e-05,
+      "loss": 0.0427,
+      "step": 13665
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9662355476067062e-05,
+      "loss": 0.0452,
+      "step": 13666
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9651439237621706e-05,
+      "loss": 0.0443,
+      "step": 13667
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9640525700083357e-05,
+      "loss": 0.0462,
+      "step": 13668
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9629614863818834e-05,
+      "loss": 0.0417,
+      "step": 13669
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9618706729194925e-05,
+      "loss": 0.0424,
+      "step": 13670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9607801296578322e-05,
+      "loss": 0.0427,
+      "step": 13671
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9596898566335576e-05,
+      "loss": 0.0439,
+      "step": 13672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9585998538833196e-05,
+      "loss": 0.0419,
+      "step": 13673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9575101214437576e-05,
+      "loss": 0.0419,
+      "step": 13674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9564206593515032e-05,
+      "loss": 0.0446,
+      "step": 13675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9553314676431834e-05,
+      "loss": 0.0433,
+      "step": 13676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9542425463554037e-05,
+      "loss": 0.0442,
+      "step": 13677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9531538955247706e-05,
+      "loss": 0.0463,
+      "step": 13678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.952065515187881e-05,
+      "loss": 0.0421,
+      "step": 13679
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9509774053813202e-05,
+      "loss": 0.0421,
+      "step": 13680
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9498895661416648e-05,
+      "loss": 0.0429,
+      "step": 13681
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9488019975054793e-05,
+      "loss": 0.0441,
+      "step": 13682
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9477146995093242e-05,
+      "loss": 0.0427,
+      "step": 13683
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.946627672189749e-05,
+      "loss": 0.0456,
+      "step": 13684
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9455409155832973e-05,
+      "loss": 0.0441,
+      "step": 13685
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9444544297264943e-05,
+      "loss": 0.0418,
+      "step": 13686
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9433682146558652e-05,
+      "loss": 0.0433,
+      "step": 13687
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9422822704079214e-05,
+      "loss": 0.0446,
+      "step": 13688
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.941196597019169e-05,
+      "loss": 0.0442,
+      "step": 13689
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9401111945261054e-05,
+      "loss": 0.0432,
+      "step": 13690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.939026062965209e-05,
+      "loss": 0.0416,
+      "step": 13691
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9379412023729593e-05,
+      "loss": 0.0458,
+      "step": 13692
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9368566127858258e-05,
+      "loss": 0.0437,
+      "step": 13693
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.935772294240268e-05,
+      "loss": 0.042,
+      "step": 13694
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9346882467727325e-05,
+      "loss": 0.0435,
+      "step": 13695
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9336044704196578e-05,
+      "loss": 0.045,
+      "step": 13696
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.932520965217476e-05,
+      "loss": 0.0427,
+      "step": 13697
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.93143773120261e-05,
+      "loss": 0.0441,
+      "step": 13698
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9303547684114743e-05,
+      "loss": 0.0491,
+      "step": 13699
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9292720768804695e-05,
+      "loss": 0.0419,
+      "step": 13700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9281896566459902e-05,
+      "loss": 0.0442,
+      "step": 13701
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9271075077444223e-05,
+      "loss": 0.0436,
+      "step": 13702
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.926025630212144e-05,
+      "loss": 0.0438,
+      "step": 13703
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9249440240855233e-05,
+      "loss": 0.043,
+      "step": 13704
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9238626894009127e-05,
+      "loss": 0.0481,
+      "step": 13705
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9227816261946653e-05,
+      "loss": 0.0476,
+      "step": 13706
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.921700834503123e-05,
+      "loss": 0.0434,
+      "step": 13707
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.92062031436261e-05,
+      "loss": 0.0433,
+      "step": 13708
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9195400658094553e-05,
+      "loss": 0.0422,
+      "step": 13709
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9184600888799643e-05,
+      "loss": 0.0456,
+      "step": 13710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.917380383610443e-05,
+      "loss": 0.0441,
+      "step": 13711
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9163009500371863e-05,
+      "loss": 0.0421,
+      "step": 13712
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.915221788196482e-05,
+      "loss": 0.0446,
+      "step": 13713
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9141428981246e-05,
+      "loss": 0.043,
+      "step": 13714
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9130642798578103e-05,
+      "loss": 0.0432,
+      "step": 13715
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.911985933432371e-05,
+      "loss": 0.0472,
+      "step": 13716
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9109078588845287e-05,
+      "loss": 0.0417,
+      "step": 13717
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 0.0413,
+      "step": 13718
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9087525255665895e-05,
+      "loss": 0.0415,
+      "step": 13719
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9076752668689434e-05,
+      "loss": 0.0464,
+      "step": 13720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9065982801937964e-05,
+      "loss": 0.0441,
+      "step": 13721
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.905521565577353e-05,
+      "loss": 0.042,
+      "step": 13722
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9044451230558082e-05,
+      "loss": 0.0432,
+      "step": 13723
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.903368952665343e-05,
+      "loss": 0.0427,
+      "step": 13724
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9022930544421336e-05,
+      "loss": 0.0442,
+      "step": 13725
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9012174284223484e-05,
+      "loss": 0.0425,
+      "step": 13726
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.900142074642146e-05,
+      "loss": 0.0434,
+      "step": 13727
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8990669931376683e-05,
+      "loss": 0.0412,
+      "step": 13728
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8979921839450578e-05,
+      "loss": 0.0428,
+      "step": 13729
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8969176471004436e-05,
+      "loss": 0.0463,
+      "step": 13730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8958433826399457e-05,
+      "loss": 0.043,
+      "step": 13731
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8947693905996787e-05,
+      "loss": 0.0426,
+      "step": 13732
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8936956710157415e-05,
+      "loss": 0.0507,
+      "step": 13733
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8926222239242252e-05,
+      "loss": 0.0423,
+      "step": 13734
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8915490493612155e-05,
+      "loss": 0.0442,
+      "step": 13735
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8904761473627873e-05,
+      "loss": 0.0436,
+      "step": 13736
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8894035179650082e-05,
+      "loss": 0.0439,
+      "step": 13737
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8883311612039302e-05,
+      "loss": 0.0413,
+      "step": 13738
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.887259077115603e-05,
+      "loss": 0.043,
+      "step": 13739
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8861872657360634e-05,
+      "loss": 0.0455,
+      "step": 13740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8851157271013442e-05,
+      "loss": 0.0494,
+      "step": 13741
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8840444612474583e-05,
+      "loss": 0.0427,
+      "step": 13742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8829734682104204e-05,
+      "loss": 0.0462,
+      "step": 13743
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.881902748026231e-05,
+      "loss": 0.0415,
+      "step": 13744
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8808323007308824e-05,
+      "loss": 0.0556,
+      "step": 13745
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8797621263603594e-05,
+      "loss": 0.044,
+      "step": 13746
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8786922249506343e-05,
+      "loss": 0.0439,
+      "step": 13747
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8776225965376694e-05,
+      "loss": 0.0423,
+      "step": 13748
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8765532411574206e-05,
+      "loss": 0.0457,
+      "step": 13749
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.875484158845836e-05,
+      "loss": 0.0429,
+      "step": 13750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.8744153496388552e-05,
+      "loss": 0.0457,
+      "step": 13751
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8733468135724008e-05,
+      "loss": 0.0422,
+      "step": 13752
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8722785506823924e-05,
+      "loss": 0.0418,
+      "step": 13753
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8712105610047427e-05,
+      "loss": 0.0438,
+      "step": 13754
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.870142844575352e-05,
+      "loss": 0.042,
+      "step": 13755
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8690754014301072e-05,
+      "loss": 0.0423,
+      "step": 13756
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.868008231604892e-05,
+      "loss": 0.0416,
+      "step": 13757
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8669413351355836e-05,
+      "loss": 0.0427,
+      "step": 13758
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8658747120580388e-05,
+      "loss": 0.0434,
+      "step": 13759
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.864808362408115e-05,
+      "loss": 0.0433,
+      "step": 13760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.86374228622166e-05,
+      "loss": 0.0503,
+      "step": 13761
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8626764835345056e-05,
+      "loss": 0.0424,
+      "step": 13762
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.86161095438248e-05,
+      "loss": 0.0432,
+      "step": 13763
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8605456988014015e-05,
+      "loss": 0.045,
+      "step": 13764
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8594807168270813e-05,
+      "loss": 0.0453,
+      "step": 13765
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8584160084953127e-05,
+      "loss": 0.0425,
+      "step": 13766
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.857351573841888e-05,
+      "loss": 0.0581,
+      "step": 13767
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8562874129025897e-05,
+      "loss": 0.044,
+      "step": 13768
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.85522352571319e-05,
+      "loss": 0.0443,
+      "step": 13769
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8541599123094476e-05,
+      "loss": 0.0449,
+      "step": 13770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.853096572727121e-05,
+      "loss": 0.0472,
+      "step": 13771
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.852033507001948e-05,
+      "loss": 0.0425,
+      "step": 13772
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8509707151696664e-05,
+      "loss": 0.0422,
+      "step": 13773
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8499081972660027e-05,
+      "loss": 0.0465,
+      "step": 13774
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8488459533266745e-05,
+      "loss": 0.042,
+      "step": 13775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8477839833873844e-05,
+      "loss": 0.0428,
+      "step": 13776
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.846722287483834e-05,
+      "loss": 0.0419,
+      "step": 13777
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8456608656517105e-05,
+      "loss": 0.0415,
+      "step": 13778
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8445997179266972e-05,
+      "loss": 0.0421,
+      "step": 13779
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8435388443444577e-05,
+      "loss": 0.0443,
+      "step": 13780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8424782449406585e-05,
+      "loss": 0.0449,
+      "step": 13781
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8414179197509486e-05,
+      "loss": 0.0426,
+      "step": 13782
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8403578688109723e-05,
+      "loss": 0.0451,
+      "step": 13783
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8392980921563675e-05,
+      "loss": 0.0453,
+      "step": 13784
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8382385898227494e-05,
+      "loss": 0.0438,
+      "step": 13785
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8371793618457366e-05,
+      "loss": 0.042,
+      "step": 13786
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8361204082609352e-05,
+      "loss": 0.0415,
+      "step": 13787
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.835061729103943e-05,
+      "loss": 0.0452,
+      "step": 13788
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8340033244103495e-05,
+      "loss": 0.0445,
+      "step": 13789
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8329451942157262e-05,
+      "loss": 0.0434,
+      "step": 13790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8318873385556466e-05,
+      "loss": 0.0424,
+      "step": 13791
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8308297574656697e-05,
+      "loss": 0.0748,
+      "step": 13792
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8297724509813473e-05,
+      "loss": 0.0418,
+      "step": 13793
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8287154191382172e-05,
+      "loss": 0.0412,
+      "step": 13794
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.827658661971814e-05,
+      "loss": 0.0425,
+      "step": 13795
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8266021795176592e-05,
+      "loss": 0.0417,
+      "step": 13796
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.825545971811269e-05,
+      "loss": 0.0469,
+      "step": 13797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8244900388881437e-05,
+      "loss": 0.0434,
+      "step": 13798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.823434380783783e-05,
+      "loss": 0.0452,
+      "step": 13799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.822378997533667e-05,
+      "loss": 0.042,
+      "step": 13800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8213238891732754e-05,
+      "loss": 0.0442,
+      "step": 13801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.820269055738075e-05,
+      "loss": 0.0512,
+      "step": 13802
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8192144972635274e-05,
+      "loss": 0.043,
+      "step": 13803
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.818160213785075e-05,
+      "loss": 0.0436,
+      "step": 13804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8171062053381614e-05,
+      "loss": 0.0439,
+      "step": 13805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8160524719582162e-05,
+      "loss": 0.0419,
+      "step": 13806
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8149990136806626e-05,
+      "loss": 0.0438,
+      "step": 13807
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.8139458305409096e-05,
+      "loss": 0.0445,
+      "step": 13808
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8128929225743596e-05,
+      "loss": 0.0446,
+      "step": 13809
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.811840289816409e-05,
+      "loss": 0.0425,
+      "step": 13810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.810787932302439e-05,
+      "loss": 0.044,
+      "step": 13811
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.809735850067824e-05,
+      "loss": 0.0425,
+      "step": 13812
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8086840431479346e-05,
+      "loss": 0.0418,
+      "step": 13813
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.807632511578121e-05,
+      "loss": 0.0435,
+      "step": 13814
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8065812553937323e-05,
+      "loss": 0.0459,
+      "step": 13815
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8055302746301084e-05,
+      "loss": 0.0424,
+      "step": 13816
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.804479569322578e-05,
+      "loss": 0.052,
+      "step": 13817
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8034291395064562e-05,
+      "loss": 0.0422,
+      "step": 13818
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.802378985217057e-05,
+      "loss": 0.0441,
+      "step": 13819
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8013291064896786e-05,
+      "loss": 0.0419,
+      "step": 13820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.8002795033596176e-05,
+      "loss": 0.058,
+      "step": 13821
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7992301758621488e-05,
+      "loss": 0.0443,
+      "step": 13822
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.798181124032553e-05,
+      "loss": 0.0465,
+      "step": 13823
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7971323479060864e-05,
+      "loss": 0.0503,
+      "step": 13824
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7960838475180076e-05,
+      "loss": 0.0421,
+      "step": 13825
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7950356229035615e-05,
+      "loss": 0.0431,
+      "step": 13826
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7939876740979867e-05,
+      "loss": 0.0417,
+      "step": 13827
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.792940001136503e-05,
+      "loss": 0.0431,
+      "step": 13828
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7918926040543327e-05,
+      "loss": 0.0474,
+      "step": 13829
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.790845482886683e-05,
+      "loss": 0.0437,
+      "step": 13830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7897986376687558e-05,
+      "loss": 0.046,
+      "step": 13831
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7887520684357338e-05,
+      "loss": 0.0428,
+      "step": 13832
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.787705775222802e-05,
+      "loss": 0.0428,
+      "step": 13833
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7866597580651312e-05,
+      "loss": 0.0419,
+      "step": 13834
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.785614016997884e-05,
+      "loss": 0.0437,
+      "step": 13835
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7845685520562116e-05,
+      "loss": 0.0418,
+      "step": 13836
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7835233632752546e-05,
+      "loss": 0.0427,
+      "step": 13837
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7824784506901492e-05,
+      "loss": 0.0411,
+      "step": 13838
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7814338143360198e-05,
+      "loss": 0.0425,
+      "step": 13839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7803894542479826e-05,
+      "loss": 0.0434,
+      "step": 13840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.779345370461145e-05,
+      "loss": 0.0434,
+      "step": 13841
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7783015630105993e-05,
+      "loss": 0.0457,
+      "step": 13842
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7772580319314357e-05,
+      "loss": 0.0425,
+      "step": 13843
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7762147772587312e-05,
+      "loss": 0.0444,
+      "step": 13844
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7751717990275595e-05,
+      "loss": 0.0437,
+      "step": 13845
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7741290972729718e-05,
+      "loss": 0.0457,
+      "step": 13846
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7730866720300244e-05,
+      "loss": 0.043,
+      "step": 13847
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7720445233337557e-05,
+      "loss": 0.0424,
+      "step": 13848
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7710026512192024e-05,
+      "loss": 0.0436,
+      "step": 13849
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.769961055721382e-05,
+      "loss": 0.0443,
+      "step": 13850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.768919736875305e-05,
+      "loss": 0.0419,
+      "step": 13851
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7678786947159796e-05,
+      "loss": 0.0432,
+      "step": 13852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7668379292784e-05,
+      "loss": 0.0446,
+      "step": 13853
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7657974405975497e-05,
+      "loss": 0.0436,
+      "step": 13854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.764757228708409e-05,
+      "loss": 0.0442,
+      "step": 13855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.763717293645939e-05,
+      "loss": 0.0437,
+      "step": 13856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7626776354450992e-05,
+      "loss": 0.0443,
+      "step": 13857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7616382541408373e-05,
+      "loss": 0.0444,
+      "step": 13858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.760599149768094e-05,
+      "loss": 0.0446,
+      "step": 13859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.759560322361795e-05,
+      "loss": 0.042,
+      "step": 13860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7585217719568626e-05,
+      "loss": 0.0426,
+      "step": 13861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7574834985882094e-05,
+      "loss": 0.0422,
+      "step": 13862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7564455022907333e-05,
+      "loss": 0.0484,
+      "step": 13863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7554077830993286e-05,
+      "loss": 0.0428,
+      "step": 13864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7543703410488756e-05,
+      "loss": 0.0495,
+      "step": 13865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.75333317617425e-05,
+      "loss": 0.0415,
+      "step": 13866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7522962885103145e-05,
+      "loss": 0.0428,
+      "step": 13867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7512596780919256e-05,
+      "loss": 0.0469,
+      "step": 13868
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7502233449539308e-05,
+      "loss": 0.0423,
+      "step": 13869
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.74918728913116e-05,
+      "loss": 0.0425,
+      "step": 13870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7481515106584446e-05,
+      "loss": 0.042,
+      "step": 13871
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7471160095706006e-05,
+      "loss": 0.0436,
+      "step": 13872
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.746080785902441e-05,
+      "loss": 0.0448,
+      "step": 13873
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.745045839688756e-05,
+      "loss": 0.0419,
+      "step": 13874
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7440111709643437e-05,
+      "loss": 0.0652,
+      "step": 13875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7429767797639783e-05,
+      "loss": 0.0426,
+      "step": 13876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7419426661224326e-05,
+      "loss": 0.0425,
+      "step": 13877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7409088300744702e-05,
+      "loss": 0.048,
+      "step": 13878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7398752716548395e-05,
+      "loss": 0.0421,
+      "step": 13879
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7388419908982867e-05,
+      "loss": 0.0469,
+      "step": 13880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7378089878395432e-05,
+      "loss": 0.0485,
+      "step": 13881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7367762625133354e-05,
+      "loss": 0.0413,
+      "step": 13882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7357438149543794e-05,
+      "loss": 0.0425,
+      "step": 13883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7347116451973754e-05,
+      "loss": 0.0419,
+      "step": 13884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7336797532770244e-05,
+      "loss": 0.054,
+      "step": 13885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7326481392280103e-05,
+      "loss": 0.0424,
+      "step": 13886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7316168030850154e-05,
+      "loss": 0.0438,
+      "step": 13887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7305857448827057e-05,
+      "loss": 0.0423,
+      "step": 13888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7295549646557353e-05,
+      "loss": 0.0434,
+      "step": 13889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7285244624387575e-05,
+      "loss": 0.0482,
+      "step": 13890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.727494238266414e-05,
+      "loss": 0.0413,
+      "step": 13891
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7264642921733354e-05,
+      "loss": 0.0414,
+      "step": 13892
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.725434624194141e-05,
+      "loss": 0.0421,
+      "step": 13893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.724405234363443e-05,
+      "loss": 0.0428,
+      "step": 13894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7233761227158452e-05,
+      "loss": 0.0495,
+      "step": 13895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7223472892859417e-05,
+      "loss": 0.0425,
+      "step": 13896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.721318734108318e-05,
+      "loss": 0.0431,
+      "step": 13897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.720290457217545e-05,
+      "loss": 0.0421,
+      "step": 13898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7192624586481887e-05,
+      "loss": 0.0432,
+      "step": 13899
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7182347384348096e-05,
+      "loss": 0.0436,
+      "step": 13900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.717207296611949e-05,
+      "loss": 0.044,
+      "step": 13901
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.716180133214149e-05,
+      "loss": 0.0461,
+      "step": 13902
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7151532482759324e-05,
+      "loss": 0.0427,
+      "step": 13903
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7141266418318204e-05,
+      "loss": 0.0419,
+      "step": 13904
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7131003139163228e-05,
+      "loss": 0.0431,
+      "step": 13905
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7120742645639387e-05,
+      "loss": 0.0439,
+      "step": 13906
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.711048493809161e-05,
+      "loss": 0.0441,
+      "step": 13907
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.710023001686466e-05,
+      "loss": 0.0411,
+      "step": 13908
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.708997788230329e-05,
+      "loss": 0.0431,
+      "step": 13909
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7079728534752115e-05,
+      "loss": 0.0429,
+      "step": 13910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.706948197455569e-05,
+      "loss": 0.0422,
+      "step": 13911
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7059238202058403e-05,
+      "loss": 0.0432,
+      "step": 13912
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7048997217604645e-05,
+      "loss": 0.0435,
+      "step": 13913
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.703875902153863e-05,
+      "loss": 0.047,
+      "step": 13914
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7028523614204516e-05,
+      "loss": 0.0428,
+      "step": 13915
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7018290995946407e-05,
+      "loss": 0.0418,
+      "step": 13916
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7008061167108214e-05,
+      "loss": 0.0416,
+      "step": 13917
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.699783412803384e-05,
+      "loss": 0.0434,
+      "step": 13918
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6987609879067067e-05,
+      "loss": 0.042,
+      "step": 13919
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6977388420551565e-05,
+      "loss": 0.043,
+      "step": 13920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6967169752830982e-05,
+      "loss": 0.0431,
+      "step": 13921
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.695695387624874e-05,
+      "loss": 0.0442,
+      "step": 13922
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.6946740791148286e-05,
+      "loss": 0.0433,
+      "step": 13923
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6936530497872917e-05,
+      "loss": 0.0433,
+      "step": 13924
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6926322996765897e-05,
+      "loss": 0.0433,
+      "step": 13925
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.69161182881703e-05,
+      "loss": 0.0428,
+      "step": 13926
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6905916372429153e-05,
+      "loss": 0.0466,
+      "step": 13927
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.68957172498854e-05,
+      "loss": 0.0433,
+      "step": 13928
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6885520920881902e-05,
+      "loss": 0.042,
+      "step": 13929
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6875327385761408e-05,
+      "loss": 0.0448,
+      "step": 13930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6865136644866553e-05,
+      "loss": 0.0429,
+      "step": 13931
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6854948698539884e-05,
+      "loss": 0.0418,
+      "step": 13932
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.684476354712391e-05,
+      "loss": 0.0423,
+      "step": 13933
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6834581190960963e-05,
+      "loss": 0.0434,
+      "step": 13934
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6824401630393382e-05,
+      "loss": 0.0522,
+      "step": 13935
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6814224865763273e-05,
+      "loss": 0.0449,
+      "step": 13936
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.680405089741277e-05,
+      "loss": 0.0421,
+      "step": 13937
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.679387972568387e-05,
+      "loss": 0.0417,
+      "step": 13938
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6783711350918486e-05,
+      "loss": 0.0431,
+      "step": 13939
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.677354577345841e-05,
+      "loss": 0.0472,
+      "step": 13940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6763382993645338e-05,
+      "loss": 0.0504,
+      "step": 13941
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6753223011820908e-05,
+      "loss": 0.047,
+      "step": 13942
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.674306582832664e-05,
+      "loss": 0.0424,
+      "step": 13943
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6732911443504005e-05,
+      "loss": 0.0427,
+      "step": 13944
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6722759857694293e-05,
+      "loss": 0.0441,
+      "step": 13945
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.671261107123876e-05,
+      "loss": 0.0431,
+      "step": 13946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.670246508447857e-05,
+      "loss": 0.046,
+      "step": 13947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6692321897754758e-05,
+      "loss": 0.0427,
+      "step": 13948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6682181511408335e-05,
+      "loss": 0.0423,
+      "step": 13949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.66720439257801e-05,
+      "loss": 0.0434,
+      "step": 13950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6661909141210874e-05,
+      "loss": 0.0497,
+      "step": 13951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6651777158041338e-05,
+      "loss": 0.0418,
+      "step": 13952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.664164797661204e-05,
+      "loss": 0.0446,
+      "step": 13953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.663152159726351e-05,
+      "loss": 0.0422,
+      "step": 13954
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.662139802033611e-05,
+      "loss": 0.0438,
+      "step": 13955
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6611277246170165e-05,
+      "loss": 0.0423,
+      "step": 13956
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.660115927510587e-05,
+      "loss": 0.0466,
+      "step": 13957
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.659104410748339e-05,
+      "loss": 0.0413,
+      "step": 13958
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.658093174364267e-05,
+      "loss": 0.0423,
+      "step": 13959
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.657082218392366e-05,
+      "loss": 0.0475,
+      "step": 13960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6560715428666217e-05,
+      "loss": 0.0449,
+      "step": 13961
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6550611478210056e-05,
+      "loss": 0.0423,
+      "step": 13962
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6540510332894854e-05,
+      "loss": 0.0426,
+      "step": 13963
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6530411993060103e-05,
+      "loss": 0.0419,
+      "step": 13964
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6520316459045317e-05,
+      "loss": 0.0418,
+      "step": 13965
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.65102237311898e-05,
+      "loss": 0.0442,
+      "step": 13966
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.650013380983284e-05,
+      "loss": 0.0425,
+      "step": 13967
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6490046695313645e-05,
+      "loss": 0.0461,
+      "step": 13968
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6479962387971225e-05,
+      "loss": 0.0438,
+      "step": 13969
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6469880888144608e-05,
+      "loss": 0.0416,
+      "step": 13970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6459802196172668e-05,
+      "loss": 0.0433,
+      "step": 13971
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6449726312394232e-05,
+      "loss": 0.0454,
+      "step": 13972
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6439653237147933e-05,
+      "loss": 0.0462,
+      "step": 13973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6429582970772426e-05,
+      "loss": 0.042,
+      "step": 13974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.64195155136062e-05,
+      "loss": 0.0463,
+      "step": 13975
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6409450865987687e-05,
+      "loss": 0.0442,
+      "step": 13976
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.639938902825523e-05,
+      "loss": 0.043,
+      "step": 13977
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6389330000747027e-05,
+      "loss": 0.0438,
+      "step": 13978
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6379273783801186e-05,
+      "loss": 0.0431,
+      "step": 13979
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.6369220377755766e-05,
+      "loss": 0.0411,
+      "step": 13980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6359169782948724e-05,
+      "loss": 0.0426,
+      "step": 13981
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.634912199971793e-05,
+      "loss": 0.0447,
+      "step": 13982
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6339077028401084e-05,
+      "loss": 0.0434,
+      "step": 13983
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.632903486933588e-05,
+      "loss": 0.0413,
+      "step": 13984
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.631899552285987e-05,
+      "loss": 0.0443,
+      "step": 13985
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6308958989310563e-05,
+      "loss": 0.046,
+      "step": 13986
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.629892526902529e-05,
+      "loss": 0.0425,
+      "step": 13987
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6288894362341334e-05,
+      "loss": 0.0422,
+      "step": 13988
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6278866269595904e-05,
+      "loss": 0.0448,
+      "step": 13989
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.62688409911261e-05,
+      "loss": 0.0423,
+      "step": 13990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6258818527268916e-05,
+      "loss": 0.0416,
+      "step": 13991
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.624879887836126e-05,
+      "loss": 0.0446,
+      "step": 13992
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6238782044739896e-05,
+      "loss": 0.0489,
+      "step": 13993
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.622876802674158e-05,
+      "loss": 0.0431,
+      "step": 13994
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.621875682470292e-05,
+      "loss": 0.0414,
+      "step": 13995
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6208748438960463e-05,
+      "loss": 0.0426,
+      "step": 13996
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6198742869850615e-05,
+      "loss": 0.0458,
+      "step": 13997
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.61887401177097e-05,
+      "loss": 0.0427,
+      "step": 13998
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6178740182873987e-05,
+      "loss": 0.0416,
+      "step": 13999
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6168743065679637e-05,
+      "loss": 0.0418,
+      "step": 14000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.615874876646266e-05,
+      "loss": 0.0419,
+      "step": 14001
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6148757285559024e-05,
+      "loss": 0.0415,
+      "step": 14002
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.61387686233046e-05,
+      "loss": 0.0425,
+      "step": 14003
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.612878278003519e-05,
+      "loss": 0.0437,
+      "step": 14004
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.61187997560864e-05,
+      "loss": 0.0428,
+      "step": 14005
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6108819551793863e-05,
+      "loss": 0.0447,
+      "step": 14006
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.609884216749301e-05,
+      "loss": 0.0412,
+      "step": 14007
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6088867603519263e-05,
+      "loss": 0.043,
+      "step": 14008
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.607889586020792e-05,
+      "loss": 0.0425,
+      "step": 14009
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.606892693789418e-05,
+      "loss": 0.0414,
+      "step": 14010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.605896083691313e-05,
+      "loss": 0.0429,
+      "step": 14011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6048997557599777e-05,
+      "loss": 0.0421,
+      "step": 14012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6039037100289055e-05,
+      "loss": 0.042,
+      "step": 14013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.602907946531579e-05,
+      "loss": 0.0456,
+      "step": 14014
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.6019124653014673e-05,
+      "loss": 0.0508,
+      "step": 14015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.600917266372035e-05,
+      "loss": 0.0411,
+      "step": 14016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.599922349776738e-05,
+      "loss": 0.0437,
+      "step": 14017
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5989277155490157e-05,
+      "loss": 0.0416,
+      "step": 14018
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.597933363722306e-05,
+      "loss": 0.0457,
+      "step": 14019
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5969392943300344e-05,
+      "loss": 0.0423,
+      "step": 14020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.595945507405613e-05,
+      "loss": 0.0422,
+      "step": 14021
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5949520029824493e-05,
+      "loss": 0.0441,
+      "step": 14022
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5939587810939406e-05,
+      "loss": 0.0409,
+      "step": 14023
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.592965841773476e-05,
+      "loss": 0.0441,
+      "step": 14024
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5919731850544283e-05,
+      "loss": 0.0438,
+      "step": 14025
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5909808109701673e-05,
+      "loss": 0.046,
+      "step": 14026
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5899887195540532e-05,
+      "loss": 0.0416,
+      "step": 14027
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5889969108394332e-05,
+      "loss": 0.0418,
+      "step": 14028
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.588005384859653e-05,
+      "loss": 0.042,
+      "step": 14029
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.587014141648032e-05,
+      "loss": 0.0433,
+      "step": 14030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.586023181237897e-05,
+      "loss": 0.0445,
+      "step": 14031
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5850325036625557e-05,
+      "loss": 0.0439,
+      "step": 14032
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.5840421089553137e-05,
+      "loss": 0.045,
+      "step": 14033
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.583051997149463e-05,
+      "loss": 0.0414,
+      "step": 14034
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.582062168278282e-05,
+      "loss": 0.0434,
+      "step": 14035
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.581072622375046e-05,
+      "loss": 0.0437,
+      "step": 14036
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.580083359473018e-05,
+      "loss": 0.0441,
+      "step": 14037
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5790943796054546e-05,
+      "loss": 0.0427,
+      "step": 14038
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5781056828055964e-05,
+      "loss": 0.0409,
+      "step": 14039
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5771172691066794e-05,
+      "loss": 0.0422,
+      "step": 14040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5761291385419296e-05,
+      "loss": 0.0436,
+      "step": 14041
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.575141291144565e-05,
+      "loss": 0.0416,
+      "step": 14042
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.574153726947787e-05,
+      "loss": 0.0465,
+      "step": 14043
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5731664459847982e-05,
+      "loss": 0.0462,
+      "step": 14044
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5721794482887807e-05,
+      "loss": 0.0449,
+      "step": 14045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5711927338929133e-05,
+      "loss": 0.0452,
+      "step": 14046
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.570206302830367e-05,
+      "loss": 0.0442,
+      "step": 14047
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.569220155134301e-05,
+      "loss": 0.0433,
+      "step": 14048
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.56823429083786e-05,
+      "loss": 0.0428,
+      "step": 14049
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.567248709974186e-05,
+      "loss": 0.0436,
+      "step": 14050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.56626341257641e-05,
+      "loss": 0.0424,
+      "step": 14051
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.565278398677654e-05,
+      "loss": 0.0498,
+      "step": 14052
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5642936683110253e-05,
+      "loss": 0.0475,
+      "step": 14053
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5633092215096278e-05,
+      "loss": 0.0447,
+      "step": 14054
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.562325058306555e-05,
+      "loss": 0.0425,
+      "step": 14055
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5613411787348863e-05,
+      "loss": 0.0418,
+      "step": 14056
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.560357582827695e-05,
+      "loss": 0.0437,
+      "step": 14057
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.559374270618048e-05,
+      "loss": 0.0427,
+      "step": 14058
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5583912421389944e-05,
+      "loss": 0.0431,
+      "step": 14059
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5574084974235815e-05,
+      "loss": 0.0439,
+      "step": 14060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5564260365048432e-05,
+      "loss": 0.043,
+      "step": 14061
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5554438594158083e-05,
+      "loss": 0.0421,
+      "step": 14062
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5544619661894864e-05,
+      "loss": 0.0413,
+      "step": 14063
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5534803568588874e-05,
+      "loss": 0.0431,
+      "step": 14064
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5524990314570063e-05,
+      "loss": 0.0445,
+      "step": 14065
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5515179900168353e-05,
+      "loss": 0.0415,
+      "step": 14066
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.550537232571344e-05,
+      "loss": 0.0463,
+      "step": 14067
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5495567591535075e-05,
+      "loss": 0.041,
+      "step": 14068
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5485765697962797e-05,
+      "loss": 0.0423,
+      "step": 14069
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.54759666453261e-05,
+      "loss": 0.0417,
+      "step": 14070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5466170433954395e-05,
+      "loss": 0.0447,
+      "step": 14071
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5456377064176997e-05,
+      "loss": 0.0512,
+      "step": 14072
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5446586536323058e-05,
+      "loss": 0.0421,
+      "step": 14073
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5436798850721724e-05,
+      "loss": 0.0428,
+      "step": 14074
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.542701400770199e-05,
+      "loss": 0.0418,
+      "step": 14075
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5417232007592796e-05,
+      "loss": 0.0414,
+      "step": 14076
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5407452850722924e-05,
+      "loss": 0.0421,
+      "step": 14077
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5397676537421125e-05,
+      "loss": 0.0424,
+      "step": 14078
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.538790306801603e-05,
+      "loss": 0.0432,
+      "step": 14079
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.537813244283618e-05,
+      "loss": 0.0417,
+      "step": 14080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.536836466221001e-05,
+      "loss": 0.0432,
+      "step": 14081
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5358599726465816e-05,
+      "loss": 0.0408,
+      "step": 14082
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5348837635931902e-05,
+      "loss": 0.0438,
+      "step": 14083
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5339078390936378e-05,
+      "loss": 0.0472,
+      "step": 14084
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5329321991807334e-05,
+      "loss": 0.042,
+      "step": 14085
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5319568438872745e-05,
+      "loss": 0.0427,
+      "step": 14086
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5309817732460418e-05,
+      "loss": 0.0418,
+      "step": 14087
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.530006987289815e-05,
+      "loss": 0.0421,
+      "step": 14088
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5290324860513626e-05,
+      "loss": 0.0452,
+      "step": 14089
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5280582695634425e-05,
+      "loss": 0.0415,
+      "step": 14090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5270843378588007e-05,
+      "loss": 0.0424,
+      "step": 14091
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5261106909701762e-05,
+      "loss": 0.043,
+      "step": 14092
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.5251373289302984e-05,
+      "loss": 0.0435,
+      "step": 14093
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.524164251771889e-05,
+      "loss": 0.0425,
+      "step": 14094
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5231914595276563e-05,
+      "loss": 0.0436,
+      "step": 14095
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5222189522302988e-05,
+      "loss": 0.0456,
+      "step": 14096
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5212467299125078e-05,
+      "loss": 0.0414,
+      "step": 14097
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5202747926069649e-05,
+      "loss": 0.0411,
+      "step": 14098
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.519303140346342e-05,
+      "loss": 0.0422,
+      "step": 14099
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5183317731633028e-05,
+      "loss": 0.0425,
+      "step": 14100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5173606910904969e-05,
+      "loss": 0.0451,
+      "step": 14101
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.516389894160567e-05,
+      "loss": 0.0517,
+      "step": 14102
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.515419382406148e-05,
+      "loss": 0.0504,
+      "step": 14103
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.514449155859865e-05,
+      "loss": 0.0466,
+      "step": 14104
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5134792145543285e-05,
+      "loss": 0.0437,
+      "step": 14105
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5125095585221438e-05,
+      "loss": 0.042,
+      "step": 14106
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.511540187795909e-05,
+      "loss": 0.043,
+      "step": 14107
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5105711024082036e-05,
+      "loss": 0.0417,
+      "step": 14108
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5096023023916094e-05,
+      "loss": 0.042,
+      "step": 14109
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5086337877786882e-05,
+      "loss": 0.0441,
+      "step": 14110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5076655586019973e-05,
+      "loss": 0.0441,
+      "step": 14111
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5066976148940836e-05,
+      "loss": 0.0432,
+      "step": 14112
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.505729956687485e-05,
+      "loss": 0.0417,
+      "step": 14113
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5047625840147328e-05,
+      "loss": 0.0431,
+      "step": 14114
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5037954969083378e-05,
+      "loss": 0.0424,
+      "step": 14115
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5028286954008131e-05,
+      "loss": 0.0411,
+      "step": 14116
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5018621795246569e-05,
+      "loss": 0.045,
+      "step": 14117
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.5008959493123597e-05,
+      "loss": 0.0544,
+      "step": 14118
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4999300047963981e-05,
+      "loss": 0.0449,
+      "step": 14119
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4989643460092461e-05,
+      "loss": 0.0437,
+      "step": 14120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.497998972983361e-05,
+      "loss": 0.0431,
+      "step": 14121
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4970338857511934e-05,
+      "loss": 0.0438,
+      "step": 14122
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4960690843451885e-05,
+      "loss": 0.0416,
+      "step": 14123
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4951045687977739e-05,
+      "loss": 0.0428,
+      "step": 14124
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4941403391413722e-05,
+      "loss": 0.0421,
+      "step": 14125
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4931763954083977e-05,
+      "loss": 0.0422,
+      "step": 14126
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4922127376312523e-05,
+      "loss": 0.0425,
+      "step": 14127
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.491249365842331e-05,
+      "loss": 0.0415,
+      "step": 14128
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4902862800740147e-05,
+      "loss": 0.0414,
+      "step": 14129
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4893234803586775e-05,
+      "loss": 0.0423,
+      "step": 14130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4883609667286858e-05,
+      "loss": 0.0452,
+      "step": 14131
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4873987392163947e-05,
+      "loss": 0.0504,
+      "step": 14132
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4864367978541483e-05,
+      "loss": 0.0434,
+      "step": 14133
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4854751426742796e-05,
+      "loss": 0.0422,
+      "step": 14134
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4845137737091163e-05,
+      "loss": 0.0434,
+      "step": 14135
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4835526909909758e-05,
+      "loss": 0.0437,
+      "step": 14136
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4825918945521667e-05,
+      "loss": 0.0427,
+      "step": 14137
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4816313844249819e-05,
+      "loss": 0.0415,
+      "step": 14138
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.480671160641709e-05,
+      "loss": 0.0424,
+      "step": 14139
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4797112232346278e-05,
+      "loss": 0.0427,
+      "step": 14140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4787515722360056e-05,
+      "loss": 0.0448,
+      "step": 14141
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4777922076781048e-05,
+      "loss": 0.0418,
+      "step": 14142
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.476833129593167e-05,
+      "loss": 0.0427,
+      "step": 14143
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4758743380134366e-05,
+      "loss": 0.0424,
+      "step": 14144
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.474915832971141e-05,
+      "loss": 0.0433,
+      "step": 14145
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4739576144985035e-05,
+      "loss": 0.0448,
+      "step": 14146
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4729996826277325e-05,
+      "loss": 0.0426,
+      "step": 14147
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.472042037391027e-05,
+      "loss": 0.0436,
+      "step": 14148
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.471084678820579e-05,
+      "loss": 0.042,
+      "step": 14149
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4701276069485703e-05,
+      "loss": 0.0412,
+      "step": 14150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.4691708218071731e-05,
+      "loss": 0.0422,
+      "step": 14151
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4682143234285517e-05,
+      "loss": 0.0427,
+      "step": 14152
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4672581118448547e-05,
+      "loss": 0.042,
+      "step": 14153
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4663021870882265e-05,
+      "loss": 0.0484,
+      "step": 14154
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4653465491908003e-05,
+      "loss": 0.0472,
+      "step": 14155
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4643911981847035e-05,
+      "loss": 0.0448,
+      "step": 14156
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4634361341020442e-05,
+      "loss": 0.0449,
+      "step": 14157
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4624813569749307e-05,
+      "loss": 0.0417,
+      "step": 14158
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4615268668354554e-05,
+      "loss": 0.0427,
+      "step": 14159
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4605726637157035e-05,
+      "loss": 0.0415,
+      "step": 14160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4596187476477541e-05,
+      "loss": 0.0412,
+      "step": 14161
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.458665118663668e-05,
+      "loss": 0.0479,
+      "step": 14162
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4577117767955028e-05,
+      "loss": 0.0458,
+      "step": 14163
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4567587220753055e-05,
+      "loss": 0.0433,
+      "step": 14164
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4558059545351143e-05,
+      "loss": 0.0416,
+      "step": 14165
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4548534742069564e-05,
+      "loss": 0.0413,
+      "step": 14166
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4539012811228458e-05,
+      "loss": 0.0425,
+      "step": 14167
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4529493753147927e-05,
+      "loss": 0.0469,
+      "step": 14168
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4519977568147946e-05,
+      "loss": 0.0427,
+      "step": 14169
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4510464256548439e-05,
+      "loss": 0.0449,
+      "step": 14170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.450095381866916e-05,
+      "loss": 0.0448,
+      "step": 14171
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4491446254829776e-05,
+      "loss": 0.0525,
+      "step": 14172
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.448194156534991e-05,
+      "loss": 0.0427,
+      "step": 14173
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4472439750549061e-05,
+      "loss": 0.042,
+      "step": 14174
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4462940810746662e-05,
+      "loss": 0.0419,
+      "step": 14175
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4453444746261958e-05,
+      "loss": 0.0425,
+      "step": 14176
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4443951557414182e-05,
+      "loss": 0.0415,
+      "step": 14177
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4434461244522458e-05,
+      "loss": 0.0452,
+      "step": 14178
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4424973807905795e-05,
+      "loss": 0.0427,
+      "step": 14179
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4415489247883141e-05,
+      "loss": 0.044,
+      "step": 14180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4406007564773261e-05,
+      "loss": 0.0419,
+      "step": 14181
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4396528758894912e-05,
+      "loss": 0.0413,
+      "step": 14182
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4387052830566727e-05,
+      "loss": 0.0421,
+      "step": 14183
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4377579780107241e-05,
+      "loss": 0.045,
+      "step": 14184
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.436810960783489e-05,
+      "loss": 0.0415,
+      "step": 14185
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4358642314067982e-05,
+      "loss": 0.0409,
+      "step": 14186
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4349177899124766e-05,
+      "loss": 0.0458,
+      "step": 14187
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4339716363323419e-05,
+      "loss": 0.0429,
+      "step": 14188
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4330257706981987e-05,
+      "loss": 0.0421,
+      "step": 14189
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.432080193041837e-05,
+      "loss": 0.0443,
+      "step": 14190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4311349033950472e-05,
+      "loss": 0.0444,
+      "step": 14191
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4301899017896036e-05,
+      "loss": 0.0421,
+      "step": 14192
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4292451882572732e-05,
+      "loss": 0.0449,
+      "step": 14193
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4283007628298128e-05,
+      "loss": 0.0433,
+      "step": 14194
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4273566255389658e-05,
+      "loss": 0.0419,
+      "step": 14195
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4264127764164726e-05,
+      "loss": 0.0433,
+      "step": 14196
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4254692154940607e-05,
+      "loss": 0.0417,
+      "step": 14197
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4245259428034453e-05,
+      "loss": 0.0457,
+      "step": 14198
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4235829583763372e-05,
+      "loss": 0.0535,
+      "step": 14199
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4226402622444313e-05,
+      "loss": 0.0443,
+      "step": 14200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4216978544394177e-05,
+      "loss": 0.0443,
+      "step": 14201
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4207557349929756e-05,
+      "loss": 0.0415,
+      "step": 14202
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4198139039367764e-05,
+      "loss": 0.0433,
+      "step": 14203
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4188723613024757e-05,
+      "loss": 0.0429,
+      "step": 14204
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4179311071217261e-05,
+      "loss": 0.0421,
+      "step": 14205
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4169901414261655e-05,
+      "loss": 0.0426,
+      "step": 14206
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4160494642474265e-05,
+      "loss": 0.0452,
+      "step": 14207
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.4151090756171303e-05,
+      "loss": 0.0446,
+      "step": 14208
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4141689755668852e-05,
+      "loss": 0.0436,
+      "step": 14209
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4132291641282968e-05,
+      "loss": 0.044,
+      "step": 14210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4122896413329512e-05,
+      "loss": 0.0419,
+      "step": 14211
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4113504072124328e-05,
+      "loss": 0.0426,
+      "step": 14212
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4104114617983166e-05,
+      "loss": 0.0411,
+      "step": 14213
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4094728051221606e-05,
+      "loss": 0.0433,
+      "step": 14214
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4085344372155195e-05,
+      "loss": 0.0493,
+      "step": 14215
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4075963581099372e-05,
+      "loss": 0.0423,
+      "step": 14216
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4066585678369481e-05,
+      "loss": 0.0437,
+      "step": 14217
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.405721066428073e-05,
+      "loss": 0.0428,
+      "step": 14218
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4047838539148262e-05,
+      "loss": 0.0453,
+      "step": 14219
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4038469303287138e-05,
+      "loss": 0.0429,
+      "step": 14220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4029102957012296e-05,
+      "loss": 0.0468,
+      "step": 14221
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4019739500638607e-05,
+      "loss": 0.0417,
+      "step": 14222
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4010378934480795e-05,
+      "loss": 0.0436,
+      "step": 14223
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.4001021258853509e-05,
+      "loss": 0.0511,
+      "step": 14224
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3991666474071307e-05,
+      "loss": 0.0416,
+      "step": 14225
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3982314580448653e-05,
+      "loss": 0.0464,
+      "step": 14226
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3972965578299946e-05,
+      "loss": 0.042,
+      "step": 14227
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3963619467939404e-05,
+      "loss": 0.0478,
+      "step": 14228
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3954276249681209e-05,
+      "loss": 0.0465,
+      "step": 14229
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.394493592383943e-05,
+      "loss": 0.045,
+      "step": 14230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3935598490728064e-05,
+      "loss": 0.0469,
+      "step": 14231
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3926263950660957e-05,
+      "loss": 0.0415,
+      "step": 14232
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3916932303951901e-05,
+      "loss": 0.0416,
+      "step": 14233
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3907603550914571e-05,
+      "loss": 0.0464,
+      "step": 14234
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3898277691862572e-05,
+      "loss": 0.0454,
+      "step": 14235
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3888954727109394e-05,
+      "loss": 0.0424,
+      "step": 14236
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.387963465696841e-05,
+      "loss": 0.0415,
+      "step": 14237
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3870317481752892e-05,
+      "loss": 0.0448,
+      "step": 14238
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3861003201776069e-05,
+      "loss": 0.0432,
+      "step": 14239
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.385169181735102e-05,
+      "loss": 0.0419,
+      "step": 14240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3842383328790775e-05,
+      "loss": 0.0433,
+      "step": 14241
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3833077736408207e-05,
+      "loss": 0.0437,
+      "step": 14242
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.382377504051613e-05,
+      "loss": 0.0428,
+      "step": 14243
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.381447524142725e-05,
+      "loss": 0.0439,
+      "step": 14244
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3805178339454216e-05,
+      "loss": 0.0433,
+      "step": 14245
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3795884334909481e-05,
+      "loss": 0.0416,
+      "step": 14246
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3786593228105494e-05,
+      "loss": 0.0416,
+      "step": 14247
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.377730501935458e-05,
+      "loss": 0.0422,
+      "step": 14248
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3768019708968972e-05,
+      "loss": 0.0488,
+      "step": 14249
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3758737297260748e-05,
+      "loss": 0.0421,
+      "step": 14250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3749457784541985e-05,
+      "loss": 0.0456,
+      "step": 14251
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3740181171124578e-05,
+      "loss": 0.0467,
+      "step": 14252
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3730907457320363e-05,
+      "loss": 0.0418,
+      "step": 14253
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3721636643441093e-05,
+      "loss": 0.0439,
+      "step": 14254
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3712368729798409e-05,
+      "loss": 0.0437,
+      "step": 14255
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3703103716703824e-05,
+      "loss": 0.0416,
+      "step": 14256
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3693841604468793e-05,
+      "loss": 0.0423,
+      "step": 14257
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3684582393404654e-05,
+      "loss": 0.0422,
+      "step": 14258
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3675326083822693e-05,
+      "loss": 0.0421,
+      "step": 14259
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3666072676034003e-05,
+      "loss": 0.0409,
+      "step": 14260
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3656822170349659e-05,
+      "loss": 0.0417,
+      "step": 14261
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3647574567080645e-05,
+      "loss": 0.0418,
+      "step": 14262
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3638329866537769e-05,
+      "loss": 0.0429,
+      "step": 14263
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3629088069031815e-05,
+      "loss": 0.0475,
+      "step": 14264
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3619849174873467e-05,
+      "loss": 0.0477,
+      "step": 14265
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.3610613184373233e-05,
+      "loss": 0.0479,
+      "step": 14266
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3601380097841621e-05,
+      "loss": 0.043,
+      "step": 14267
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3592149915588993e-05,
+      "loss": 0.042,
+      "step": 14268
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3582922637925632e-05,
+      "loss": 0.0426,
+      "step": 14269
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3573698265161683e-05,
+      "loss": 0.042,
+      "step": 14270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.356447679760724e-05,
+      "loss": 0.0423,
+      "step": 14271
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3555258235572265e-05,
+      "loss": 0.0418,
+      "step": 14272
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3546042579366669e-05,
+      "loss": 0.0423,
+      "step": 14273
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3536829829300234e-05,
+      "loss": 0.0411,
+      "step": 14274
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3527619985682626e-05,
+      "loss": 0.0425,
+      "step": 14275
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.351841304882342e-05,
+      "loss": 0.0421,
+      "step": 14276
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3509209019032131e-05,
+      "loss": 0.0433,
+      "step": 14277
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3500007896618128e-05,
+      "loss": 0.0465,
+      "step": 14278
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.349080968189076e-05,
+      "loss": 0.044,
+      "step": 14279
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3481614375159146e-05,
+      "loss": 0.0431,
+      "step": 14280
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.347242197673243e-05,
+      "loss": 0.0418,
+      "step": 14281
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.346323248691962e-05,
+      "loss": 0.0421,
+      "step": 14282
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.345404590602961e-05,
+      "loss": 0.0415,
+      "step": 14283
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.344486223437119e-05,
+      "loss": 0.045,
+      "step": 14284
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3435681472253081e-05,
+      "loss": 0.0441,
+      "step": 14285
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.34265036199839e-05,
+      "loss": 0.0427,
+      "step": 14286
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3417328677872142e-05,
+      "loss": 0.0414,
+      "step": 14287
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3408156646226255e-05,
+      "loss": 0.0435,
+      "step": 14288
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3398987525354534e-05,
+      "loss": 0.0422,
+      "step": 14289
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3389821315565176e-05,
+      "loss": 0.0428,
+      "step": 14290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3380658017166326e-05,
+      "loss": 0.0426,
+      "step": 14291
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.337149763046599e-05,
+      "loss": 0.0472,
+      "step": 14292
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3362340155772146e-05,
+      "loss": 0.0432,
+      "step": 14293
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3353185593392548e-05,
+      "loss": 0.0413,
+      "step": 14294
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3344033943634971e-05,
+      "loss": 0.0423,
+      "step": 14295
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3334885206807023e-05,
+      "loss": 0.0444,
+      "step": 14296
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3325739383216274e-05,
+      "loss": 0.0422,
+      "step": 14297
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3316596473170117e-05,
+      "loss": 0.0427,
+      "step": 14298
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.330745647697591e-05,
+      "loss": 0.0493,
+      "step": 14299
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3298319394940917e-05,
+      "loss": 0.0499,
+      "step": 14300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3289185227372226e-05,
+      "loss": 0.0454,
+      "step": 14301
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3280053974576912e-05,
+      "loss": 0.0441,
+      "step": 14302
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3270925636861942e-05,
+      "loss": 0.0429,
+      "step": 14303
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3261800214534115e-05,
+      "loss": 0.0431,
+      "step": 14304
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3252677707900208e-05,
+      "loss": 0.0418,
+      "step": 14305
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3243558117266875e-05,
+      "loss": 0.0425,
+      "step": 14306
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3234441442940681e-05,
+      "loss": 0.042,
+      "step": 14307
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.322532768522805e-05,
+      "loss": 0.0424,
+      "step": 14308
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3216216844435347e-05,
+      "loss": 0.0415,
+      "step": 14309
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.320710892086885e-05,
+      "loss": 0.0414,
+      "step": 14310
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3198003914834734e-05,
+      "loss": 0.0419,
+      "step": 14311
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3188901826639023e-05,
+      "loss": 0.0426,
+      "step": 14312
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3179802656587714e-05,
+      "loss": 0.0486,
+      "step": 14313
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3170706404986644e-05,
+      "loss": 0.043,
+      "step": 14314
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3161613072141598e-05,
+      "loss": 0.041,
+      "step": 14315
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3152522658358245e-05,
+      "loss": 0.0433,
+      "step": 14316
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3143435163942186e-05,
+      "loss": 0.0424,
+      "step": 14317
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3134350589198851e-05,
+      "loss": 0.0409,
+      "step": 14318
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3125268934433632e-05,
+      "loss": 0.0467,
+      "step": 14319
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3116190199951828e-05,
+      "loss": 0.0418,
+      "step": 14320
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3107114386058617e-05,
+      "loss": 0.0433,
+      "step": 14321
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.3098041493059044e-05,
+      "loss": 0.0462,
+      "step": 14322
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.308897152125812e-05,
+      "loss": 0.0426,
+      "step": 14323
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3079904470960735e-05,
+      "loss": 0.0412,
+      "step": 14324
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.307084034247169e-05,
+      "loss": 0.0422,
+      "step": 14325
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3061779136095653e-05,
+      "loss": 0.0419,
+      "step": 14326
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3052720852137201e-05,
+      "loss": 0.0405,
+      "step": 14327
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3043665490900836e-05,
+      "loss": 0.0416,
+      "step": 14328
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.303461305269097e-05,
+      "loss": 0.0454,
+      "step": 14329
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3025563537811891e-05,
+      "loss": 0.0418,
+      "step": 14330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3016516946567815e-05,
+      "loss": 0.0452,
+      "step": 14331
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.3007473279262805e-05,
+      "loss": 0.0427,
+      "step": 14332
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2998432536200877e-05,
+      "loss": 0.0419,
+      "step": 14333
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2989394717685942e-05,
+      "loss": 0.0445,
+      "step": 14334
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2980359824021815e-05,
+      "loss": 0.042,
+      "step": 14335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2971327855512184e-05,
+      "loss": 0.0454,
+      "step": 14336
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.296229881246066e-05,
+      "loss": 0.0439,
+      "step": 14337
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2953272695170748e-05,
+      "loss": 0.0422,
+      "step": 14338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2944249503945894e-05,
+      "loss": 0.0425,
+      "step": 14339
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2935229239089397e-05,
+      "loss": 0.0428,
+      "step": 14340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2926211900904427e-05,
+      "loss": 0.0416,
+      "step": 14341
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2917197489694134e-05,
+      "loss": 0.0451,
+      "step": 14342
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2908186005761546e-05,
+      "loss": 0.0419,
+      "step": 14343
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2899177449409572e-05,
+      "loss": 0.0435,
+      "step": 14344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2890171820941054e-05,
+      "loss": 0.0423,
+      "step": 14345
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2881169120658676e-05,
+      "loss": 0.0411,
+      "step": 14346
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2872169348865094e-05,
+      "loss": 0.0442,
+      "step": 14347
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2863172505862809e-05,
+      "loss": 0.0444,
+      "step": 14348
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.285417859195429e-05,
+      "loss": 0.0432,
+      "step": 14349
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2845187607441822e-05,
+      "loss": 0.0448,
+      "step": 14350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2836199552627647e-05,
+      "loss": 0.0482,
+      "step": 14351
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2827214427813927e-05,
+      "loss": 0.0432,
+      "step": 14352
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2818232233302663e-05,
+      "loss": 0.0418,
+      "step": 14353
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2809252969395803e-05,
+      "loss": 0.0429,
+      "step": 14354
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.280027663639517e-05,
+      "loss": 0.0446,
+      "step": 14355
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2791303234602514e-05,
+      "loss": 0.0426,
+      "step": 14356
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.278233276431947e-05,
+      "loss": 0.0443,
+      "step": 14357
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.277336522584759e-05,
+      "loss": 0.0437,
+      "step": 14358
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2764400619488326e-05,
+      "loss": 0.0423,
+      "step": 14359
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2755438945542986e-05,
+      "loss": 0.0447,
+      "step": 14360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2746480204312839e-05,
+      "loss": 0.0409,
+      "step": 14361
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2737524396099032e-05,
+      "loss": 0.0416,
+      "step": 14362
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2728571521202626e-05,
+      "loss": 0.0425,
+      "step": 14363
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2719621579924535e-05,
+      "loss": 0.0451,
+      "step": 14364
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.271067457256565e-05,
+      "loss": 0.0424,
+      "step": 14365
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2701730499426678e-05,
+      "loss": 0.0421,
+      "step": 14366
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.269278936080831e-05,
+      "loss": 0.043,
+      "step": 14367
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2683851157011106e-05,
+      "loss": 0.0423,
+      "step": 14368
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.267491588833548e-05,
+      "loss": 0.0462,
+      "step": 14369
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2665983555081818e-05,
+      "loss": 0.0416,
+      "step": 14370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2657054157550385e-05,
+      "loss": 0.0422,
+      "step": 14371
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2648127696041334e-05,
+      "loss": 0.044,
+      "step": 14372
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2639204170854747e-05,
+      "loss": 0.0412,
+      "step": 14373
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2630283582290536e-05,
+      "loss": 0.0429,
+      "step": 14374
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.262136593064861e-05,
+      "loss": 0.0417,
+      "step": 14375
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2612451216228715e-05,
+      "loss": 0.0451,
+      "step": 14376
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2603539439330547e-05,
+      "loss": 0.0408,
+      "step": 14377
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2594630600253654e-05,
+      "loss": 0.0414,
+      "step": 14378
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2585724699297485e-05,
+      "loss": 0.0415,
+      "step": 14379
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.2576821736761423e-05,
+      "loss": 0.0414,
+      "step": 14380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2567921712944753e-05,
+      "loss": 0.0451,
+      "step": 14381
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2559024628146666e-05,
+      "loss": 0.0439,
+      "step": 14382
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2550130482666178e-05,
+      "loss": 0.0432,
+      "step": 14383
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2541239276802308e-05,
+      "loss": 0.0417,
+      "step": 14384
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2532351010853916e-05,
+      "loss": 0.0418,
+      "step": 14385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2523465685119795e-05,
+      "loss": 0.0444,
+      "step": 14386
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2514583299898631e-05,
+      "loss": 0.0437,
+      "step": 14387
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.250570385548897e-05,
+      "loss": 0.0432,
+      "step": 14388
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2496827352189323e-05,
+      "loss": 0.0452,
+      "step": 14389
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2487953790298056e-05,
+      "loss": 0.0419,
+      "step": 14390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2479083170113481e-05,
+      "loss": 0.0415,
+      "step": 14391
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2470215491933756e-05,
+      "loss": 0.0428,
+      "step": 14392
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2461350756056956e-05,
+      "loss": 0.0431,
+      "step": 14393
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2452488962781084e-05,
+      "loss": 0.0411,
+      "step": 14394
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2443630112404026e-05,
+      "loss": 0.0442,
+      "step": 14395
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2434774205223576e-05,
+      "loss": 0.0462,
+      "step": 14396
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2425921241537442e-05,
+      "loss": 0.0491,
+      "step": 14397
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2417071221643173e-05,
+      "loss": 0.0427,
+      "step": 14398
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2408224145838287e-05,
+      "loss": 0.0419,
+      "step": 14399
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.239938001442017e-05,
+      "loss": 0.042,
+      "step": 14400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2390538827686149e-05,
+      "loss": 0.042,
+      "step": 14401
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2381700585933365e-05,
+      "loss": 0.0432,
+      "step": 14402
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2372865289458935e-05,
+      "loss": 0.0421,
+      "step": 14403
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2364032938559888e-05,
+      "loss": 0.0485,
+      "step": 14404
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2355203533533067e-05,
+      "loss": 0.0422,
+      "step": 14405
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.234637707467533e-05,
+      "loss": 0.0414,
+      "step": 14406
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2337553562283322e-05,
+      "loss": 0.0418,
+      "step": 14407
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2328732996653669e-05,
+      "loss": 0.0455,
+      "step": 14408
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.231991537808287e-05,
+      "loss": 0.0423,
+      "step": 14409
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.231110070686734e-05,
+      "loss": 0.0409,
+      "step": 14410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2302288983303389e-05,
+      "loss": 0.0453,
+      "step": 14411
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2293480207687192e-05,
+      "loss": 0.0415,
+      "step": 14412
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2284674380314876e-05,
+      "loss": 0.0436,
+      "step": 14413
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2275871501482439e-05,
+      "loss": 0.0425,
+      "step": 14414
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2267071571485812e-05,
+      "loss": 0.0441,
+      "step": 14415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2258274590620766e-05,
+      "loss": 0.0424,
+      "step": 14416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2249480559183046e-05,
+      "loss": 0.0412,
+      "step": 14417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2240689477468226e-05,
+      "loss": 0.042,
+      "step": 14418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2231901345771824e-05,
+      "loss": 0.0421,
+      "step": 14419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2223116164389304e-05,
+      "loss": 0.0464,
+      "step": 14420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2214333933615908e-05,
+      "loss": 0.0421,
+      "step": 14421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2205554653746876e-05,
+      "loss": 0.0459,
+      "step": 14422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.219677832507734e-05,
+      "loss": 0.0414,
+      "step": 14423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2188004947902288e-05,
+      "loss": 0.0424,
+      "step": 14424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.217923452251668e-05,
+      "loss": 0.0424,
+      "step": 14425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2170467049215273e-05,
+      "loss": 0.043,
+      "step": 14426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.216170252829283e-05,
+      "loss": 0.0419,
+      "step": 14427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2152940960043935e-05,
+      "loss": 0.0447,
+      "step": 14428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2144182344763156e-05,
+      "loss": 0.0426,
+      "step": 14429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2135426682744888e-05,
+      "loss": 0.0449,
+      "step": 14430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.212667397428342e-05,
+      "loss": 0.0465,
+      "step": 14431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2117924219672994e-05,
+      "loss": 0.0428,
+      "step": 14432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.210917741920774e-05,
+      "loss": 0.0452,
+      "step": 14433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2100433573181702e-05,
+      "loss": 0.042,
+      "step": 14434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2091692681888756e-05,
+      "loss": 0.0428,
+      "step": 14435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2082954745622743e-05,
+      "loss": 0.044,
+      "step": 14436
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.2074219764677396e-05,
+      "loss": 0.043,
+      "step": 14437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.2065487739346348e-05,
+      "loss": 0.0438,
+      "step": 14438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.205675866992313e-05,
+      "loss": 0.0461,
+      "step": 14439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.2048032556701127e-05,
+      "loss": 0.0447,
+      "step": 14440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.2039309399973697e-05,
+      "loss": 0.0417,
+      "step": 14441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.203058920003408e-05,
+      "loss": 0.0427,
+      "step": 14442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.2021871957175379e-05,
+      "loss": 0.0437,
+      "step": 14443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.2013157671690645e-05,
+      "loss": 0.0498,
+      "step": 14444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.2004446343872767e-05,
+      "loss": 0.0429,
+      "step": 14445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.199573797401461e-05,
+      "loss": 0.0417,
+      "step": 14446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1987032562408906e-05,
+      "loss": 0.0437,
+      "step": 14447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.197833010934829e-05,
+      "loss": 0.0437,
+      "step": 14448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.196963061512526e-05,
+      "loss": 0.0428,
+      "step": 14449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1960934080032271e-05,
+      "loss": 0.0418,
+      "step": 14450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1952240504361655e-05,
+      "loss": 0.0439,
+      "step": 14451
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1943549888405647e-05,
+      "loss": 0.0414,
+      "step": 14452
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1934862232456401e-05,
+      "loss": 0.0431,
+      "step": 14453
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1926177536805905e-05,
+      "loss": 0.0417,
+      "step": 14454
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1917495801746148e-05,
+      "loss": 0.045,
+      "step": 14455
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1908817027568908e-05,
+      "loss": 0.0526,
+      "step": 14456
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1900141214565951e-05,
+      "loss": 0.0433,
+      "step": 14457
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1891468363028935e-05,
+      "loss": 0.0417,
+      "step": 14458
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1882798473249356e-05,
+      "loss": 0.0426,
+      "step": 14459
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1874131545518663e-05,
+      "loss": 0.0413,
+      "step": 14460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1865467580128208e-05,
+      "loss": 0.0605,
+      "step": 14461
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1856806577369229e-05,
+      "loss": 0.041,
+      "step": 14462
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1848148537532843e-05,
+      "loss": 0.0441,
+      "step": 14463
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.18394934609101e-05,
+      "loss": 0.042,
+      "step": 14464
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1830841347791954e-05,
+      "loss": 0.0455,
+      "step": 14465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1822192198469218e-05,
+      "loss": 0.0425,
+      "step": 14466
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1813546013232668e-05,
+      "loss": 0.0412,
+      "step": 14467
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.180490279237293e-05,
+      "loss": 0.0441,
+      "step": 14468
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1796262536180514e-05,
+      "loss": 0.0429,
+      "step": 14469
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1787625244945888e-05,
+      "loss": 0.0415,
+      "step": 14470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1778990918959386e-05,
+      "loss": 0.0435,
+      "step": 14471
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1770359558511278e-05,
+      "loss": 0.0421,
+      "step": 14472
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1761731163891654e-05,
+      "loss": 0.0427,
+      "step": 14473
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1753105735390591e-05,
+      "loss": 0.0443,
+      "step": 14474
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1744483273298024e-05,
+      "loss": 0.0423,
+      "step": 14475
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1735863777903821e-05,
+      "loss": 0.0447,
+      "step": 14476
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1727247249497685e-05,
+      "loss": 0.048,
+      "step": 14477
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1718633688369263e-05,
+      "loss": 0.0416,
+      "step": 14478
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1710023094808121e-05,
+      "loss": 0.0414,
+      "step": 14479
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1701415469103694e-05,
+      "loss": 0.0418,
+      "step": 14480
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1692810811545352e-05,
+      "loss": 0.0411,
+      "step": 14481
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1684209122422307e-05,
+      "loss": 0.0458,
+      "step": 14482
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1675610402023685e-05,
+      "loss": 0.0421,
+      "step": 14483
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1667014650638564e-05,
+      "loss": 0.0424,
+      "step": 14484
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1658421868555879e-05,
+      "loss": 0.0419,
+      "step": 14485
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1649832056064502e-05,
+      "loss": 0.0465,
+      "step": 14486
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1641245213453134e-05,
+      "loss": 0.0415,
+      "step": 14487
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1632661341010443e-05,
+      "loss": 0.0444,
+      "step": 14488
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1624080439024964e-05,
+      "loss": 0.0426,
+      "step": 14489
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.161550250778518e-05,
+      "loss": 0.0428,
+      "step": 14490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1606927547579383e-05,
+      "loss": 0.0446,
+      "step": 14491
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.159835555869585e-05,
+      "loss": 0.0434,
+      "step": 14492
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.158978654142272e-05,
+      "loss": 0.0423,
+      "step": 14493
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.1581220496048062e-05,
+      "loss": 0.0429,
+      "step": 14494
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.157265742285979e-05,
+      "loss": 0.041,
+      "step": 14495
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1564097322145773e-05,
+      "loss": 0.0419,
+      "step": 14496
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1555540194193737e-05,
+      "loss": 0.0414,
+      "step": 14497
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.154698603929133e-05,
+      "loss": 0.0438,
+      "step": 14498
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.153843485772612e-05,
+      "loss": 0.0476,
+      "step": 14499
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.152988664978556e-05,
+      "loss": 0.0463,
+      "step": 14500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1521341415756947e-05,
+      "loss": 0.0426,
+      "step": 14501
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1512799155927578e-05,
+      "loss": 0.0431,
+      "step": 14502
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1504259870584567e-05,
+      "loss": 0.0428,
+      "step": 14503
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1495723560015003e-05,
+      "loss": 0.0415,
+      "step": 14504
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1487190224505795e-05,
+      "loss": 0.0471,
+      "step": 14505
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1478659864343789e-05,
+      "loss": 0.0426,
+      "step": 14506
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1470132479815777e-05,
+      "loss": 0.0431,
+      "step": 14507
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1461608071208341e-05,
+      "loss": 0.0417,
+      "step": 14508
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1453086638808075e-05,
+      "loss": 0.0415,
+      "step": 14509
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1444568182901416e-05,
+      "loss": 0.0408,
+      "step": 14510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1436052703774703e-05,
+      "loss": 0.0436,
+      "step": 14511
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.142754020171417e-05,
+      "loss": 0.0436,
+      "step": 14512
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1419030677005981e-05,
+      "loss": 0.0458,
+      "step": 14513
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1410524129936217e-05,
+      "loss": 0.0441,
+      "step": 14514
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.140202056079076e-05,
+      "loss": 0.0426,
+      "step": 14515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1393519969855481e-05,
+      "loss": 0.0409,
+      "step": 14516
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.138502235741613e-05,
+      "loss": 0.043,
+      "step": 14517
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1376527723758356e-05,
+      "loss": 0.0439,
+      "step": 14518
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1368036069167732e-05,
+      "loss": 0.0437,
+      "step": 14519
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1359547393929671e-05,
+      "loss": 0.0433,
+      "step": 14520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1351061698329502e-05,
+      "loss": 0.0424,
+      "step": 14521
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1342578982652485e-05,
+      "loss": 0.0445,
+      "step": 14522
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1334099247183783e-05,
+      "loss": 0.0451,
+      "step": 14523
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1325622492208442e-05,
+      "loss": 0.0424,
+      "step": 14524
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1317148718011373e-05,
+      "loss": 0.0444,
+      "step": 14525
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1308677924877443e-05,
+      "loss": 0.0422,
+      "step": 14526
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1300210113091402e-05,
+      "loss": 0.0439,
+      "step": 14527
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1291745282937904e-05,
+      "loss": 0.0464,
+      "step": 14528
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1283283434701463e-05,
+      "loss": 0.0443,
+      "step": 14529
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1274824568666541e-05,
+      "loss": 0.0423,
+      "step": 14530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1266368685117478e-05,
+      "loss": 0.0419,
+      "step": 14531
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1257915784338513e-05,
+      "loss": 0.0424,
+      "step": 14532
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1249465866613817e-05,
+      "loss": 0.0413,
+      "step": 14533
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.124101893222741e-05,
+      "loss": 0.0417,
+      "step": 14534
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1232574981463217e-05,
+      "loss": 0.0418,
+      "step": 14535
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1224134014605104e-05,
+      "loss": 0.045,
+      "step": 14536
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1215696031936795e-05,
+      "loss": 0.0438,
+      "step": 14537
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1207261033741978e-05,
+      "loss": 0.0459,
+      "step": 14538
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1198829020304137e-05,
+      "loss": 0.0414,
+      "step": 14539
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1190399991906731e-05,
+      "loss": 0.0426,
+      "step": 14540
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1181973948833113e-05,
+      "loss": 0.0415,
+      "step": 14541
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1173550891366536e-05,
+      "loss": 0.0414,
+      "step": 14542
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1165130819790104e-05,
+      "loss": 0.0413,
+      "step": 14543
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.115671373438687e-05,
+      "loss": 0.0422,
+      "step": 14544
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1148299635439774e-05,
+      "loss": 0.0436,
+      "step": 14545
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1139888523231678e-05,
+      "loss": 0.0425,
+      "step": 14546
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.113148039804529e-05,
+      "loss": 0.0418,
+      "step": 14547
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1123075260163273e-05,
+      "loss": 0.0416,
+      "step": 14548
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1114673109868134e-05,
+      "loss": 0.0411,
+      "step": 14549
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1106273947442325e-05,
+      "loss": 0.0425,
+      "step": 14550
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.1097877773168176e-05,
+      "loss": 0.0412,
+      "step": 14551
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1089484587327959e-05,
+      "loss": 0.048,
+      "step": 14552
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.108109439020376e-05,
+      "loss": 0.0414,
+      "step": 14553
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1072707182077646e-05,
+      "loss": 0.0416,
+      "step": 14554
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1064322963231533e-05,
+      "loss": 0.0421,
+      "step": 14555
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1055941733947295e-05,
+      "loss": 0.0428,
+      "step": 14556
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1047563494506608e-05,
+      "loss": 0.0426,
+      "step": 14557
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1039188245191157e-05,
+      "loss": 0.0422,
+      "step": 14558
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1030815986282439e-05,
+      "loss": 0.0427,
+      "step": 14559
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1022446718061897e-05,
+      "loss": 0.0434,
+      "step": 14560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.1014080440810859e-05,
+      "loss": 0.0417,
+      "step": 14561
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.100571715481058e-05,
+      "loss": 0.0424,
+      "step": 14562
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0997356860342168e-05,
+      "loss": 0.0432,
+      "step": 14563
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0988999557686641e-05,
+      "loss": 0.0451,
+      "step": 14564
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0980645247124965e-05,
+      "loss": 0.0438,
+      "step": 14565
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.097229392893796e-05,
+      "loss": 0.0436,
+      "step": 14566
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0963945603406322e-05,
+      "loss": 0.0439,
+      "step": 14567
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0955600270810706e-05,
+      "loss": 0.042,
+      "step": 14568
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0947257931431642e-05,
+      "loss": 0.0454,
+      "step": 14569
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0938918585549562e-05,
+      "loss": 0.0421,
+      "step": 14570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0930582233444787e-05,
+      "loss": 0.0415,
+      "step": 14571
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0922248875397512e-05,
+      "loss": 0.0424,
+      "step": 14572
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0913918511687892e-05,
+      "loss": 0.0449,
+      "step": 14573
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0905591142595938e-05,
+      "loss": 0.0417,
+      "step": 14574
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0897266768401581e-05,
+      "loss": 0.0418,
+      "step": 14575
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0888945389384664e-05,
+      "loss": 0.0415,
+      "step": 14576
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0880627005824862e-05,
+      "loss": 0.0423,
+      "step": 14577
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.087231161800183e-05,
+      "loss": 0.0421,
+      "step": 14578
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0863999226195087e-05,
+      "loss": 0.043,
+      "step": 14579
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0855689830684058e-05,
+      "loss": 0.0422,
+      "step": 14580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0847383431748037e-05,
+      "loss": 0.0415,
+      "step": 14581
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.083908002966625e-05,
+      "loss": 0.0431,
+      "step": 14582
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0830779624717824e-05,
+      "loss": 0.0426,
+      "step": 14583
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0822482217181807e-05,
+      "loss": 0.0411,
+      "step": 14584
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0814187807337062e-05,
+      "loss": 0.0448,
+      "step": 14585
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0805896395462422e-05,
+      "loss": 0.0474,
+      "step": 14586
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.07976079818366e-05,
+      "loss": 0.0442,
+      "step": 14587
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0789322566738214e-05,
+      "loss": 0.0459,
+      "step": 14588
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0781040150445776e-05,
+      "loss": 0.0411,
+      "step": 14589
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0772760733237718e-05,
+      "loss": 0.0438,
+      "step": 14590
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0764484315392321e-05,
+      "loss": 0.0411,
+      "step": 14591
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0756210897187812e-05,
+      "loss": 0.0445,
+      "step": 14592
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0747940478902296e-05,
+      "loss": 0.0425,
+      "step": 14593
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0739673060813794e-05,
+      "loss": 0.0446,
+      "step": 14594
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0731408643200192e-05,
+      "loss": 0.0424,
+      "step": 14595
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0723147226339314e-05,
+      "loss": 0.0412,
+      "step": 14596
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0714888810508872e-05,
+      "loss": 0.0413,
+      "step": 14597
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0706633395986443e-05,
+      "loss": 0.0451,
+      "step": 14598
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0698380983049572e-05,
+      "loss": 0.0446,
+      "step": 14599
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0690131571975626e-05,
+      "loss": 0.0426,
+      "step": 14600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0681885163041927e-05,
+      "loss": 0.0431,
+      "step": 14601
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0673641756525665e-05,
+      "loss": 0.042,
+      "step": 14602
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0665401352703952e-05,
+      "loss": 0.0416,
+      "step": 14603
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0657163951853799e-05,
+      "loss": 0.0418,
+      "step": 14604
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0648929554252086e-05,
+      "loss": 0.0422,
+      "step": 14605
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.06406981601756e-05,
+      "loss": 0.0411,
+      "step": 14606
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0632469769901065e-05,
+      "loss": 0.0637,
+      "step": 14607
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0624244383705084e-05,
+      "loss": 0.0421,
+      "step": 14608
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0616022001864123e-05,
+      "loss": 0.0421,
+      "step": 14609
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0607802624654605e-05,
+      "loss": 0.0414,
+      "step": 14610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0599586252352788e-05,
+      "loss": 0.0414,
+      "step": 14611
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0591372885234885e-05,
+      "loss": 0.0427,
+      "step": 14612
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0583162523577006e-05,
+      "loss": 0.0419,
+      "step": 14613
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0574955167655098e-05,
+      "loss": 0.0444,
+      "step": 14614
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0566750817745074e-05,
+      "loss": 0.0435,
+      "step": 14615
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0558549474122726e-05,
+      "loss": 0.0428,
+      "step": 14616
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0550351137063742e-05,
+      "loss": 0.0415,
+      "step": 14617
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0542155806843711e-05,
+      "loss": 0.0428,
+      "step": 14618
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0533963483738097e-05,
+      "loss": 0.0424,
+      "step": 14619
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0525774168022296e-05,
+      "loss": 0.0422,
+      "step": 14620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0517587859971601e-05,
+      "loss": 0.0411,
+      "step": 14621
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0509404559861191e-05,
+      "loss": 0.0425,
+      "step": 14622
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0501224267966148e-05,
+      "loss": 0.0492,
+      "step": 14623
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0493046984561428e-05,
+      "loss": 0.0415,
+      "step": 14624
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0484872709921922e-05,
+      "loss": 0.0467,
+      "step": 14625
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0476701444322402e-05,
+      "loss": 0.046,
+      "step": 14626
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0468533188037587e-05,
+      "loss": 0.0437,
+      "step": 14627
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0460367941341986e-05,
+      "loss": 0.0426,
+      "step": 14628
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.045220570451011e-05,
+      "loss": 0.0427,
+      "step": 14629
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0444046477816327e-05,
+      "loss": 0.0421,
+      "step": 14630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.043589026153491e-05,
+      "loss": 0.0453,
+      "step": 14631
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0427737055940034e-05,
+      "loss": 0.0462,
+      "step": 14632
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0419586861305752e-05,
+      "loss": 0.0468,
+      "step": 14633
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0411439677906031e-05,
+      "loss": 0.0418,
+      "step": 14634
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0403295506014754e-05,
+      "loss": 0.0447,
+      "step": 14635
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.03951543459057e-05,
+      "loss": 0.0433,
+      "step": 14636
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0387016197852517e-05,
+      "loss": 0.0428,
+      "step": 14637
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0378881062128731e-05,
+      "loss": 0.0423,
+      "step": 14638
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0370748939007845e-05,
+      "loss": 0.0435,
+      "step": 14639
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0362619828763209e-05,
+      "loss": 0.0421,
+      "step": 14640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035449373166808e-05,
+      "loss": 0.0411,
+      "step": 14641
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.034637064799564e-05,
+      "loss": 0.0415,
+      "step": 14642
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0338250578018904e-05,
+      "loss": 0.0428,
+      "step": 14643
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0330133522010854e-05,
+      "loss": 0.0436,
+      "step": 14644
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0322019480244327e-05,
+      "loss": 0.0431,
+      "step": 14645
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0313908452992116e-05,
+      "loss": 0.0426,
+      "step": 14646
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0305800440526813e-05,
+      "loss": 0.0419,
+      "step": 14647
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0297695443121003e-05,
+      "loss": 0.0468,
+      "step": 14648
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0289593461047153e-05,
+      "loss": 0.0415,
+      "step": 14649
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0281494494577559e-05,
+      "loss": 0.0416,
+      "step": 14650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0273398543984513e-05,
+      "loss": 0.0427,
+      "step": 14651
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0265305609540122e-05,
+      "loss": 0.0422,
+      "step": 14652
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0257215691516441e-05,
+      "loss": 0.0431,
+      "step": 14653
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0249128790185425e-05,
+      "loss": 0.0453,
+      "step": 14654
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0241044905818897e-05,
+      "loss": 0.0412,
+      "step": 14655
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0232964038688641e-05,
+      "loss": 0.0457,
+      "step": 14656
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.022488618906623e-05,
+      "loss": 0.0415,
+      "step": 14657
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0216811357223222e-05,
+      "loss": 0.0415,
+      "step": 14658
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0208739543431068e-05,
+      "loss": 0.0432,
+      "step": 14659
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0200670747961106e-05,
+      "loss": 0.0458,
+      "step": 14660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.019260497108453e-05,
+      "loss": 0.0416,
+      "step": 14661
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0184542213072524e-05,
+      "loss": 0.0417,
+      "step": 14662
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0176482474196048e-05,
+      "loss": 0.0437,
+      "step": 14663
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0168425754726086e-05,
+      "loss": 0.0492,
+      "step": 14664
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0160372054933453e-05,
+      "loss": 0.0417,
+      "step": 14665
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0152321375088847e-05,
+      "loss": 0.0422,
+      "step": 14666
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0144273715462904e-05,
+      "loss": 0.042,
+      "step": 14667
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0136229076326153e-05,
+      "loss": 0.0446,
+      "step": 14668
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0128187457949002e-05,
+      "loss": 0.0432,
+      "step": 14669
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0120148860601808e-05,
+      "loss": 0.043,
+      "step": 14670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0112113284554725e-05,
+      "loss": 0.0475,
+      "step": 14671
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0104080730077913e-05,
+      "loss": 0.0416,
+      "step": 14672
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0096051197441358e-05,
+      "loss": 0.0417,
+      "step": 14673
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.008802468691501e-05,
+      "loss": 0.0458,
+      "step": 14674
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0080001198768652e-05,
+      "loss": 0.0433,
+      "step": 14675
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.007198073327198e-05,
+      "loss": 0.0535,
+      "step": 14676
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0063963290694612e-05,
+      "loss": 0.0408,
+      "step": 14677
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0055948871306065e-05,
+      "loss": 0.0435,
+      "step": 14678
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0047937475375746e-05,
+      "loss": 0.0422,
+      "step": 14679
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0039929103172919e-05,
+      "loss": 0.0436,
+      "step": 14680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0031923754966821e-05,
+      "loss": 0.0425,
+      "step": 14681
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.002392143102654e-05,
+      "loss": 0.0417,
+      "step": 14682
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.001592213162108e-05,
+      "loss": 0.0417,
+      "step": 14683
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.000792585701934e-05,
+      "loss": 0.0411,
+      "step": 14684
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.999932607490092e-06,
+      "loss": 0.0424,
+      "step": 14685
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.991942383302033e-06,
+      "loss": 0.0419,
+      "step": 14686
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.98395518472377e-06,
+      "loss": 0.0426,
+      "step": 14687
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.975971012023798e-06,
+      "loss": 0.047,
+      "step": 14688
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.967989865470485e-06,
+      "loss": 0.0446,
+      "step": 14689
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.96001174533211e-06,
+      "loss": 0.0422,
+      "step": 14690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.952036651876862e-06,
+      "loss": 0.0442,
+      "step": 14691
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.944064585372825e-06,
+      "loss": 0.041,
+      "step": 14692
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.936095546087998e-06,
+      "loss": 0.0419,
+      "step": 14693
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.928129534290231e-06,
+      "loss": 0.0431,
+      "step": 14694
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.920166550247301e-06,
+      "loss": 0.0414,
+      "step": 14695
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.912206594226902e-06,
+      "loss": 0.0448,
+      "step": 14696
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.9042496664966e-06,
+      "loss": 0.0423,
+      "step": 14697
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.896295767323883e-06,
+      "loss": 0.0437,
+      "step": 14698
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.888344896976077e-06,
+      "loss": 0.0416,
+      "step": 14699
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.880397055720503e-06,
+      "loss": 0.0461,
+      "step": 14700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.872452243824271e-06,
+      "loss": 0.0429,
+      "step": 14701
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.864510461554478e-06,
+      "loss": 0.0419,
+      "step": 14702
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.856571709178097e-06,
+      "loss": 0.044,
+      "step": 14703
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.848635986961952e-06,
+      "loss": 0.0419,
+      "step": 14704
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.840703295172825e-06,
+      "loss": 0.0431,
+      "step": 14705
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.832773634077363e-06,
+      "loss": 0.0416,
+      "step": 14706
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.82484700394215e-06,
+      "loss": 0.0421,
+      "step": 14707
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.816923405033585e-06,
+      "loss": 0.0432,
+      "step": 14708
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.809002837618063e-06,
+      "loss": 0.0415,
+      "step": 14709
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.80108530196181e-06,
+      "loss": 0.0469,
+      "step": 14710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.793170798330975e-06,
+      "loss": 0.0456,
+      "step": 14711
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.785259326991635e-06,
+      "loss": 0.0514,
+      "step": 14712
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.777350888209713e-06,
+      "loss": 0.0426,
+      "step": 14713
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.769445482251027e-06,
+      "loss": 0.0426,
+      "step": 14714
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.76154310938132e-06,
+      "loss": 0.0415,
+      "step": 14715
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.75364376986625e-06,
+      "loss": 0.0423,
+      "step": 14716
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.745747463971366e-06,
+      "loss": 0.0424,
+      "step": 14717
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.73785419196207e-06,
+      "loss": 0.0435,
+      "step": 14718
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.729963954103694e-06,
+      "loss": 0.0416,
+      "step": 14719
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.72207675066148e-06,
+      "loss": 0.0422,
+      "step": 14720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.714192581900572e-06,
+      "loss": 0.0415,
+      "step": 14721
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.70631144808596e-06,
+      "loss": 0.0427,
+      "step": 14722
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.698433349482572e-06,
+      "loss": 0.0414,
+      "step": 14723
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.690558286355256e-06,
+      "loss": 0.0449,
+      "step": 14724
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.682686258968709e-06,
+      "loss": 0.0436,
+      "step": 14725
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.674817267587577e-06,
+      "loss": 0.043,
+      "step": 14726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.666951312476346e-06,
+      "loss": 0.0423,
+      "step": 14727
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.659088393899429e-06,
+      "loss": 0.0491,
+      "step": 14728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.651228512121146e-06,
+      "loss": 0.0431,
+      "step": 14729
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.643371667405698e-06,
+      "loss": 0.0418,
+      "step": 14730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.635517860017218e-06,
+      "loss": 0.0413,
+      "step": 14731
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.627667090219672e-06,
+      "loss": 0.0422,
+      "step": 14732
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.619819358276982e-06,
+      "loss": 0.0429,
+      "step": 14733
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.61197466445295e-06,
+      "loss": 0.0502,
+      "step": 14734
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.604133009011296e-06,
+      "loss": 0.0447,
+      "step": 14735
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.596294392215566e-06,
+      "loss": 0.0417,
+      "step": 14736
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.588458814329293e-06,
+      "loss": 0.0423,
+      "step": 14737
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.580626275615845e-06,
+      "loss": 0.0438,
+      "step": 14738
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.572796776338556e-06,
+      "loss": 0.0445,
+      "step": 14739
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.564970316760558e-06,
+      "loss": 0.0429,
+      "step": 14740
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.557146897144974e-06,
+      "loss": 0.0408,
+      "step": 14741
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.549326517754754e-06,
+      "loss": 0.049,
+      "step": 14742
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.541509178852804e-06,
+      "loss": 0.0435,
+      "step": 14743
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.533694880701904e-06,
+      "loss": 0.0415,
+      "step": 14744
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.525883623564736e-06,
+      "loss": 0.0417,
+      "step": 14745
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.518075407703842e-06,
+      "loss": 0.0432,
+      "step": 14746
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.510270233381713e-06,
+      "loss": 0.044,
+      "step": 14747
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.502468100860728e-06,
+      "loss": 0.042,
+      "step": 14748
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.494669010403168e-06,
+      "loss": 0.0421,
+      "step": 14749
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.486872962271154e-06,
+      "loss": 0.0438,
+      "step": 14750
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.479079956726778e-06,
+      "loss": 0.0435,
+      "step": 14751
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.471289994032007e-06,
+      "loss": 0.0423,
+      "step": 14752
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.463503074448677e-06,
+      "loss": 0.0446,
+      "step": 14753
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.455719198238554e-06,
+      "loss": 0.0424,
+      "step": 14754
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.44793836566331e-06,
+      "loss": 0.0417,
+      "step": 14755
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.44016057698447e-06,
+      "loss": 0.0467,
+      "step": 14756
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.432385832463498e-06,
+      "loss": 0.0413,
+      "step": 14757
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.424614132361732e-06,
+      "loss": 0.0413,
+      "step": 14758
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.416845476940438e-06,
+      "loss": 0.0439,
+      "step": 14759
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.409079866460724e-06,
+      "loss": 0.0428,
+      "step": 14760
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.401317301183655e-06,
+      "loss": 0.0417,
+      "step": 14761
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.393557781370155e-06,
+      "loss": 0.0413,
+      "step": 14762
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.385801307281061e-06,
+      "loss": 0.0428,
+      "step": 14763
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.378047879177144e-06,
+      "loss": 0.0417,
+      "step": 14764
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.370297497319002e-06,
+      "loss": 0.0473,
+      "step": 14765
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.362550161967132e-06,
+      "loss": 0.0413,
+      "step": 14766
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.354805873382e-06,
+      "loss": 0.0461,
+      "step": 14767
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.34706463182392e-06,
+      "loss": 0.0413,
+      "step": 14768
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.339326437553131e-06,
+      "loss": 0.0413,
+      "step": 14769
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.331591290829722e-06,
+      "loss": 0.0435,
+      "step": 14770
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.323859191913709e-06,
+      "loss": 0.0431,
+      "step": 14771
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.316130141065027e-06,
+      "loss": 0.0421,
+      "step": 14772
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.308404138543492e-06,
+      "loss": 0.0416,
+      "step": 14773
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.300681184608783e-06,
+      "loss": 0.052,
+      "step": 14774
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.292961279520517e-06,
+      "loss": 0.0433,
+      "step": 14775
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.285244423538197e-06,
+      "loss": 0.0482,
+      "step": 14776
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.277530616921237e-06,
+      "loss": 0.0426,
+      "step": 14777
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.269819859928942e-06,
+      "loss": 0.0435,
+      "step": 14778
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.26211215282049e-06,
+      "loss": 0.042,
+      "step": 14779
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.254407495854966e-06,
+      "loss": 0.042,
+      "step": 14780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.246705889291373e-06,
+      "loss": 0.0409,
+      "step": 14781
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.239007333388605e-06,
+      "loss": 0.042,
+      "step": 14782
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.231311828405454e-06,
+      "loss": 0.0429,
+      "step": 14783
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.223619374600578e-06,
+      "loss": 0.0449,
+      "step": 14784
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.215929972232572e-06,
+      "loss": 0.0426,
+      "step": 14785
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.208243621559909e-06,
+      "loss": 0.0426,
+      "step": 14786
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.200560322841e-06,
+      "loss": 0.0449,
+      "step": 14787
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.192880076334065e-06,
+      "loss": 0.0416,
+      "step": 14788
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.185202882297294e-06,
+      "loss": 0.0417,
+      "step": 14789
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.17752874098876e-06,
+      "loss": 0.0425,
+      "step": 14790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.169857652666447e-06,
+      "loss": 0.0418,
+      "step": 14791
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.16218961758818e-06,
+      "loss": 0.0422,
+      "step": 14792
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.154524636011752e-06,
+      "loss": 0.0418,
+      "step": 14793
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.146862708194792e-06,
+      "loss": 0.0415,
+      "step": 14794
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.139203834394872e-06,
+      "loss": 0.0416,
+      "step": 14795
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.13154801486944e-06,
+      "loss": 0.0419,
+      "step": 14796
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.123895249875858e-06,
+      "loss": 0.0418,
+      "step": 14797
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.116245539671342e-06,
+      "loss": 0.0427,
+      "step": 14798
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.108598884513053e-06,
+      "loss": 0.0413,
+      "step": 14799
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.100955284658041e-06,
+      "loss": 0.0437,
+      "step": 14800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.093314740363257e-06,
+      "loss": 0.0418,
+      "step": 14801
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.085677251885493e-06,
+      "loss": 0.0422,
+      "step": 14802
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.078042819481503e-06,
+      "loss": 0.0438,
+      "step": 14803
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.070411443407955e-06,
+      "loss": 0.0432,
+      "step": 14804
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.062783123921314e-06,
+      "loss": 0.044,
+      "step": 14805
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.055157861278029e-06,
+      "loss": 0.0416,
+      "step": 14806
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.047535655734463e-06,
+      "loss": 0.0415,
+      "step": 14807
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.039916507546763e-06,
+      "loss": 0.0491,
+      "step": 14808
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.032300416971096e-06,
+      "loss": 0.0456,
+      "step": 14809
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.024687384263452e-06,
+      "loss": 0.0421,
+      "step": 14810
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.017077409679786e-06,
+      "loss": 0.0417,
+      "step": 14811
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.009470493475847e-06,
+      "loss": 0.0414,
+      "step": 14812
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.001866635907364e-06,
+      "loss": 0.0415,
+      "step": 14813
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.994265837229942e-06,
+      "loss": 0.0416,
+      "step": 14814
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.986668097699102e-06,
+      "loss": 0.0423,
+      "step": 14815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.979073417570205e-06,
+      "loss": 0.0416,
+      "step": 14816
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.97148179709857e-06,
+      "loss": 0.0432,
+      "step": 14817
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.963893236539366e-06,
+      "loss": 0.0433,
+      "step": 14818
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.956307736147695e-06,
+      "loss": 0.0421,
+      "step": 14819
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.948725296178551e-06,
+      "loss": 0.0535,
+      "step": 14820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.941145916886817e-06,
+      "loss": 0.0426,
+      "step": 14821
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.933569598527247e-06,
+      "loss": 0.045,
+      "step": 14822
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.925996341354537e-06,
+      "loss": 0.0447,
+      "step": 14823
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.918426145623248e-06,
+      "loss": 0.0416,
+      "step": 14824
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.910859011587891e-06,
+      "loss": 0.0414,
+      "step": 14825
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.903294939502792e-06,
+      "loss": 0.0422,
+      "step": 14826
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.89573392962222e-06,
+      "loss": 0.0441,
+      "step": 14827
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.888175982200353e-06,
+      "loss": 0.0411,
+      "step": 14828
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.88062109749126e-06,
+      "loss": 0.0431,
+      "step": 14829
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.873069275748879e-06,
+      "loss": 0.0429,
+      "step": 14830
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.865520517227056e-06,
+      "loss": 0.0429,
+      "step": 14831
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.857974822179548e-06,
+      "loss": 0.0421,
+      "step": 14832
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.850432190860015e-06,
+      "loss": 0.0415,
+      "step": 14833
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.84289262352198e-06,
+      "loss": 0.0457,
+      "step": 14834
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.835356120418925e-06,
+      "loss": 0.042,
+      "step": 14835
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.827822681804143e-06,
+      "loss": 0.042,
+      "step": 14836
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.820292307930889e-06,
+      "loss": 0.0415,
+      "step": 14837
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.812764999052293e-06,
+      "loss": 0.0499,
+      "step": 14838
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.805240755421407e-06,
+      "loss": 0.0418,
+      "step": 14839
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.79771957729112e-06,
+      "loss": 0.0419,
+      "step": 14840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.790201464914272e-06,
+      "loss": 0.0412,
+      "step": 14841
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.782686418543596e-06,
+      "loss": 0.0416,
+      "step": 14842
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.775174438431688e-06,
+      "loss": 0.0431,
+      "step": 14843
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.767665524831092e-06,
+      "loss": 0.042,
+      "step": 14844
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.760159677994172e-06,
+      "loss": 0.0434,
+      "step": 14845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.75265689817326e-06,
+      "loss": 0.0417,
+      "step": 14846
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.745157185620578e-06,
+      "loss": 0.048,
+      "step": 14847
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.737660540588199e-06,
+      "loss": 0.0411,
+      "step": 14848
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.730166963328167e-06,
+      "loss": 0.0421,
+      "step": 14849
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.722676454092326e-06,
+      "loss": 0.0435,
+      "step": 14850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.715189013132486e-06,
+      "loss": 0.0418,
+      "step": 14851
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.707704640700353e-06,
+      "loss": 0.0423,
+      "step": 14852
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.700223337047508e-06,
+      "loss": 0.0426,
+      "step": 14853
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.692745102425403e-06,
+      "loss": 0.0452,
+      "step": 14854
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.685269937085472e-06,
+      "loss": 0.0424,
+      "step": 14855
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.677797841278934e-06,
+      "loss": 0.0427,
+      "step": 14856
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.670328815257e-06,
+      "loss": 0.0415,
+      "step": 14857
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.662862859270748e-06,
+      "loss": 0.0415,
+      "step": 14858
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.655399973571098e-06,
+      "loss": 0.0424,
+      "step": 14859
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.647940158408951e-06,
+      "loss": 0.0419,
+      "step": 14860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.640483414035061e-06,
+      "loss": 0.0425,
+      "step": 14861
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.633029740700083e-06,
+      "loss": 0.0418,
+      "step": 14862
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.625579138654594e-06,
+      "loss": 0.0434,
+      "step": 14863
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.618131608149005e-06,
+      "loss": 0.043,
+      "step": 14864
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.610687149433683e-06,
+      "loss": 0.0415,
+      "step": 14865
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.603245762758882e-06,
+      "loss": 0.0421,
+      "step": 14866
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.595807448374738e-06,
+      "loss": 0.041,
+      "step": 14867
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.588372206531292e-06,
+      "loss": 0.0421,
+      "step": 14868
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.580940037478447e-06,
+      "loss": 0.0411,
+      "step": 14869
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.573510941466068e-06,
+      "loss": 0.0448,
+      "step": 14870
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.566084918743877e-06,
+      "loss": 0.0421,
+      "step": 14871
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.5586619695615e-06,
+      "loss": 0.046,
+      "step": 14872
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.551242094168443e-06,
+      "loss": 0.0435,
+      "step": 14873
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.543825292814144e-06,
+      "loss": 0.0422,
+      "step": 14874
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.536411565747903e-06,
+      "loss": 0.0446,
+      "step": 14875
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.529000913218944e-06,
+      "loss": 0.0568,
+      "step": 14876
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.521593335476375e-06,
+      "loss": 0.0426,
+      "step": 14877
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.514188832769188e-06,
+      "loss": 0.0431,
+      "step": 14878
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.506787405346295e-06,
+      "loss": 0.0429,
+      "step": 14879
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.499389053456485e-06,
+      "loss": 0.0416,
+      "step": 14880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.491993777348472e-06,
+      "loss": 0.0416,
+      "step": 14881
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.484601577270846e-06,
+      "loss": 0.0421,
+      "step": 14882
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.47721245347205e-06,
+      "loss": 0.0421,
+      "step": 14883
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.469826406200509e-06,
+      "loss": 0.0415,
+      "step": 14884
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.462443435704503e-06,
+      "loss": 0.0436,
+      "step": 14885
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.455063542232189e-06,
+      "loss": 0.0455,
+      "step": 14886
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.447686726031679e-06,
+      "loss": 0.0409,
+      "step": 14887
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.440312987350906e-06,
+      "loss": 0.0421,
+      "step": 14888
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.432942326437753e-06,
+      "loss": 0.0437,
+      "step": 14889
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.425574743539977e-06,
+      "loss": 0.0419,
+      "step": 14890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.418210238905256e-06,
+      "loss": 0.0427,
+      "step": 14891
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.410848812781124e-06,
+      "loss": 0.0463,
+      "step": 14892
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.403490465415042e-06,
+      "loss": 0.0414,
+      "step": 14893
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.396135197054388e-06,
+      "loss": 0.0445,
+      "step": 14894
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.388783007946355e-06,
+      "loss": 0.045,
+      "step": 14895
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.381433898338142e-06,
+      "loss": 0.044,
+      "step": 14896
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.374087868476733e-06,
+      "loss": 0.043,
+      "step": 14897
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.366744918609104e-06,
+      "loss": 0.0422,
+      "step": 14898
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.35940504898206e-06,
+      "loss": 0.0407,
+      "step": 14899
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.352068259842361e-06,
+      "loss": 0.0428,
+      "step": 14900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.344734551436617e-06,
+      "loss": 0.0426,
+      "step": 14901
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.337403924011345e-06,
+      "loss": 0.0409,
+      "step": 14902
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.330076377812957e-06,
+      "loss": 0.0423,
+      "step": 14903
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.322751913087779e-06,
+      "loss": 0.0428,
+      "step": 14904
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.315430530082036e-06,
+      "loss": 0.045,
+      "step": 14905
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.308112229041809e-06,
+      "loss": 0.0413,
+      "step": 14906
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.300797010213124e-06,
+      "loss": 0.0434,
+      "step": 14907
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.293484873841861e-06,
+      "loss": 0.0434,
+      "step": 14908
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.286175820173814e-06,
+      "loss": 0.041,
+      "step": 14909
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.278869849454718e-06,
+      "loss": 0.0423,
+      "step": 14910
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.2715669619301e-06,
+      "loss": 0.0421,
+      "step": 14911
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.264267157845484e-06,
+      "loss": 0.0426,
+      "step": 14912
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.256970437446242e-06,
+      "loss": 0.0434,
+      "step": 14913
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.249676800977658e-06,
+      "loss": 0.0417,
+      "step": 14914
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.24238624868493e-06,
+      "loss": 0.0415,
+      "step": 14915
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.235098780813078e-06,
+      "loss": 0.0427,
+      "step": 14916
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.227814397607092e-06,
+      "loss": 0.0419,
+      "step": 14917
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.220533099311834e-06,
+      "loss": 0.042,
+      "step": 14918
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.213254886172094e-06,
+      "loss": 0.0419,
+      "step": 14919
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.205979758432503e-06,
+      "loss": 0.0435,
+      "step": 14920
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.198707716337594e-06,
+      "loss": 0.0432,
+      "step": 14921
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.19143876013183e-06,
+      "loss": 0.043,
+      "step": 14922
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.184172890059572e-06,
+      "loss": 0.0425,
+      "step": 14923
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.176910106365055e-06,
+      "loss": 0.0497,
+      "step": 14924
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.169650409292396e-06,
+      "loss": 0.0414,
+      "step": 14925
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.162393799085655e-06,
+      "loss": 0.0425,
+      "step": 14926
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.155140275988737e-06,
+      "loss": 0.0419,
+      "step": 14927
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.147889840245493e-06,
+      "loss": 0.0425,
+      "step": 14928
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.140642492099659e-06,
+      "loss": 0.0456,
+      "step": 14929
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.133398231794798e-06,
+      "loss": 0.0426,
+      "step": 14930
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.12615705957447e-06,
+      "loss": 0.0451,
+      "step": 14931
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.118918975682067e-06,
+      "loss": 0.0446,
+      "step": 14932
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.111683980360918e-06,
+      "loss": 0.0423,
+      "step": 14933
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.104452073854207e-06,
+      "loss": 0.0411,
+      "step": 14934
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.097223256405029e-06,
+      "loss": 0.0414,
+      "step": 14935
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.089997528256388e-06,
+      "loss": 0.0415,
+      "step": 14936
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.082774889651168e-06,
+      "loss": 0.0434,
+      "step": 14937
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.075555340832187e-06,
+      "loss": 0.0419,
+      "step": 14938
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.068338882042092e-06,
+      "loss": 0.0418,
+      "step": 14939
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.061125513523482e-06,
+      "loss": 0.0442,
+      "step": 14940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.053915235518828e-06,
+      "loss": 0.041,
+      "step": 14941
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.0467080482705e-06,
+      "loss": 0.0421,
+      "step": 14942
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.039503952020799e-06,
+      "loss": 0.0415,
+      "step": 14943
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.032302947011839e-06,
+      "loss": 0.0425,
+      "step": 14944
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.025105033485702e-06,
+      "loss": 0.0432,
+      "step": 14945
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.017910211684366e-06,
+      "loss": 0.0417,
+      "step": 14946
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.010718481849654e-06,
+      "loss": 0.0424,
+      "step": 14947
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 8.003529844223346e-06,
+      "loss": 0.0415,
+      "step": 14948
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.996344299047042e-06,
+      "loss": 0.0459,
+      "step": 14949
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.98916184656231e-06,
+      "loss": 0.0434,
+      "step": 14950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.981982487010587e-06,
+      "loss": 0.0419,
+      "step": 14951
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.974806220633235e-06,
+      "loss": 0.0427,
+      "step": 14952
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.96763304767142e-06,
+      "loss": 0.0424,
+      "step": 14953
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.960462968366311e-06,
+      "loss": 0.0436,
+      "step": 14954
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.95329598295892e-06,
+      "loss": 0.0459,
+      "step": 14955
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.946132091690162e-06,
+      "loss": 0.0416,
+      "step": 14956
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.938971294800868e-06,
+      "loss": 0.0442,
+      "step": 14957
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.931813592531734e-06,
+      "loss": 0.0422,
+      "step": 14958
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.924658985123335e-06,
+      "loss": 0.0468,
+      "step": 14959
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.91750747281621e-06,
+      "loss": 0.0449,
+      "step": 14960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.910359055850746e-06,
+      "loss": 0.0413,
+      "step": 14961
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.903213734467252e-06,
+      "loss": 0.0423,
+      "step": 14962
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.896071508905888e-06,
+      "loss": 0.0426,
+      "step": 14963
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.888932379406754e-06,
+      "loss": 0.0417,
+      "step": 14964
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.881796346209825e-06,
+      "loss": 0.0414,
+      "step": 14965
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.874663409554994e-06,
+      "loss": 0.0417,
+      "step": 14966
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.867533569682018e-06,
+      "loss": 0.0417,
+      "step": 14967
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.860406826830557e-06,
+      "loss": 0.0453,
+      "step": 14968
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.8532831812402e-06,
+      "loss": 0.0433,
+      "step": 14969
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.84616263315039e-06,
+      "loss": 0.0427,
+      "step": 14970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.839045182800508e-06,
+      "loss": 0.0433,
+      "step": 14971
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.83193083042979e-06,
+      "loss": 0.0415,
+      "step": 14972
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.824819576277365e-06,
+      "loss": 0.0421,
+      "step": 14973
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.817711420582285e-06,
+      "loss": 0.0432,
+      "step": 14974
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.810606363583506e-06,
+      "loss": 0.0417,
+      "step": 14975
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.803504405519868e-06,
+      "loss": 0.0422,
+      "step": 14976
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.79640554663008e-06,
+      "loss": 0.0413,
+      "step": 14977
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.789309787152776e-06,
+      "loss": 0.0419,
+      "step": 14978
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.782217127326485e-06,
+      "loss": 0.0419,
+      "step": 14979
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.77512756738964e-06,
+      "loss": 0.0414,
+      "step": 14980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.768041107580514e-06,
+      "loss": 0.0417,
+      "step": 14981
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.760957748137354e-06,
+      "loss": 0.0423,
+      "step": 14982
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.753877489298244e-06,
+      "loss": 0.0417,
+      "step": 14983
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.74680033130123e-06,
+      "loss": 0.0418,
+      "step": 14984
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.73972627438415e-06,
+      "loss": 0.042,
+      "step": 14985
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.732655318784843e-06,
+      "loss": 0.0415,
+      "step": 14986
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.725587464740958e-06,
+      "loss": 0.0413,
+      "step": 14987
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.718522712490106e-06,
+      "loss": 0.0418,
+      "step": 14988
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.711461062269765e-06,
+      "loss": 0.0413,
+      "step": 14989
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.704402514317332e-06,
+      "loss": 0.0437,
+      "step": 14990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.697347068870032e-06,
+      "loss": 0.0428,
+      "step": 14991
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.69029472616506e-06,
+      "loss": 0.0462,
+      "step": 14992
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.683245486439472e-06,
+      "loss": 0.0431,
+      "step": 14993
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.676199349930258e-06,
+      "loss": 0.0414,
+      "step": 14994
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.669156316874237e-06,
+      "loss": 0.0417,
+      "step": 14995
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.662116387508156e-06,
+      "loss": 0.0421,
+      "step": 14996
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.6550795620687e-06,
+      "loss": 0.0424,
+      "step": 14997
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.648045840792373e-06,
+      "loss": 0.0441,
+      "step": 14998
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.641015223915626e-06,
+      "loss": 0.0465,
+      "step": 14999
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.633987711674817e-06,
+      "loss": 0.0421,
+      "step": 15000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.626963304306123e-06,
+      "loss": 0.0423,
+      "step": 15001
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.619942002045699e-06,
+      "loss": 0.0413,
+      "step": 15002
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.612923805129568e-06,
+      "loss": 0.0478,
+      "step": 15003
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.605908713793652e-06,
+      "loss": 0.0418,
+      "step": 15004
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.598896728273741e-06,
+      "loss": 0.0423,
+      "step": 15005
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.591887848805545e-06,
+      "loss": 0.0412,
+      "step": 15006
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.584882075624678e-06,
+      "loss": 0.0418,
+      "step": 15007
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.577879408966637e-06,
+      "loss": 0.0419,
+      "step": 15008
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.570879849066826e-06,
+      "loss": 0.0416,
+      "step": 15009
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.56388339616052e-06,
+      "loss": 0.0426,
+      "step": 15010
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.55689005048289e-06,
+      "loss": 0.0416,
+      "step": 15011
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.549899812269035e-06,
+      "loss": 0.042,
+      "step": 15012
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.542912681753922e-06,
+      "loss": 0.0412,
+      "step": 15013
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.535928659172453e-06,
+      "loss": 0.0483,
+      "step": 15014
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.52894774475934e-06,
+      "loss": 0.0433,
+      "step": 15015
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.521969938749285e-06,
+      "loss": 0.0424,
+      "step": 15016
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.514995241376843e-06,
+      "loss": 0.0434,
+      "step": 15017
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.508023652876472e-06,
+      "loss": 0.0417,
+      "step": 15018
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.501055173482485e-06,
+      "loss": 0.0414,
+      "step": 15019
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.494089803429161e-06,
+      "loss": 0.0437,
+      "step": 15020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4871275429506336e-06,
+      "loss": 0.0418,
+      "step": 15021
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.480168392280928e-06,
+      "loss": 0.0465,
+      "step": 15022
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.473212351654013e-06,
+      "loss": 0.0423,
+      "step": 15023
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.466259421303678e-06,
+      "loss": 0.0417,
+      "step": 15024
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.459309601463626e-06,
+      "loss": 0.0414,
+      "step": 15025
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4523628923675145e-06,
+      "loss": 0.0451,
+      "step": 15026
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.445419294248834e-06,
+      "loss": 0.0411,
+      "step": 15027
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4384788073410204e-06,
+      "loss": 0.0431,
+      "step": 15028
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.431541431877342e-06,
+      "loss": 0.0419,
+      "step": 15029
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.4246071680910135e-06,
+      "loss": 0.0421,
+      "step": 15030
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.417676016215125e-06,
+      "loss": 0.0412,
+      "step": 15031
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.410747976482702e-06,
+      "loss": 0.0419,
+      "step": 15032
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.403823049126568e-06,
+      "loss": 0.0416,
+      "step": 15033
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.396901234379539e-06,
+      "loss": 0.0415,
+      "step": 15034
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.389982532474282e-06,
+      "loss": 0.0416,
+      "step": 15035
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.383066943643391e-06,
+      "loss": 0.042,
+      "step": 15036
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.3761544681193006e-06,
+      "loss": 0.0414,
+      "step": 15037
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.369245106134392e-06,
+      "loss": 0.0426,
+      "step": 15038
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.362338857920914e-06,
+      "loss": 0.0414,
+      "step": 15039
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.355435723711013e-06,
+      "loss": 0.0455,
+      "step": 15040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.348535703736747e-06,
+      "loss": 0.0418,
+      "step": 15041
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.3416387982300775e-06,
+      "loss": 0.0443,
+      "step": 15042
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.3347450074228055e-06,
+      "loss": 0.0415,
+      "step": 15043
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.3278543315466905e-06,
+      "loss": 0.0431,
+      "step": 15044
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.3209667708333485e-06,
+      "loss": 0.0415,
+      "step": 15045
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.314082325514326e-06,
+      "loss": 0.0412,
+      "step": 15046
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.307200995821018e-06,
+      "loss": 0.0453,
+      "step": 15047
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.3003227819847365e-06,
+      "loss": 0.0477,
+      "step": 15048
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.293447684236731e-06,
+      "loss": 0.0426,
+      "step": 15049
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.286575702808052e-06,
+      "loss": 0.0418,
+      "step": 15050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.279706837929734e-06,
+      "loss": 0.0414,
+      "step": 15051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.272841089832694e-06,
+      "loss": 0.0432,
+      "step": 15052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.265978458747668e-06,
+      "loss": 0.0439,
+      "step": 15053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.259118944905375e-06,
+      "loss": 0.0419,
+      "step": 15054
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.252262548536393e-06,
+      "loss": 0.0462,
+      "step": 15055
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.245409269871206e-06,
+      "loss": 0.042,
+      "step": 15056
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.238559109140175e-06,
+      "loss": 0.0414,
+      "step": 15057
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.23171206657357e-06,
+      "loss": 0.0441,
+      "step": 15058
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.224868142401542e-06,
+      "loss": 0.0417,
+      "step": 15059
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.218027336854194e-06,
+      "loss": 0.0414,
+      "step": 15060
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.211189650161421e-06,
+      "loss": 0.0543,
+      "step": 15061
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.204355082553116e-06,
+      "loss": 0.0449,
+      "step": 15062
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.197523634258985e-06,
+      "loss": 0.0409,
+      "step": 15063
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.190695305508677e-06,
+      "loss": 0.0417,
+      "step": 15064
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.183870096531742e-06,
+      "loss": 0.0515,
+      "step": 15065
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 7.177048007557607e-06,
+      "loss": 0.0423,
+      "step": 15066
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.170229038815568e-06,
+      "loss": 0.0411,
+      "step": 15067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.163413190534873e-06,
+      "loss": 0.0424,
+      "step": 15068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.156600462944618e-06,
+      "loss": 0.0442,
+      "step": 15069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.149790856273841e-06,
+      "loss": 0.0416,
+      "step": 15070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.1429843707514155e-06,
+      "loss": 0.0414,
+      "step": 15071
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.136181006606135e-06,
+      "loss": 0.042,
+      "step": 15072
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.129380764066717e-06,
+      "loss": 0.0418,
+      "step": 15073
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.122583643361769e-06,
+      "loss": 0.0415,
+      "step": 15074
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.115789644719728e-06,
+      "loss": 0.0421,
+      "step": 15075
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.108998768368991e-06,
+      "loss": 0.0418,
+      "step": 15076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.10221101453784e-06,
+      "loss": 0.043,
+      "step": 15077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.0954263834544285e-06,
+      "loss": 0.0411,
+      "step": 15078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.088644875346839e-06,
+      "loss": 0.0419,
+      "step": 15079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.081866490443046e-06,
+      "loss": 0.0423,
+      "step": 15080
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.075091228970865e-06,
+      "loss": 0.0424,
+      "step": 15081
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.06831909115806e-06,
+      "loss": 0.0452,
+      "step": 15082
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.061550077232282e-06,
+      "loss": 0.0414,
+      "step": 15083
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.054784187421082e-06,
+      "loss": 0.0428,
+      "step": 15084
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.048021421951867e-06,
+      "loss": 0.0421,
+      "step": 15085
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.041261781051977e-06,
+      "loss": 0.0427,
+      "step": 15086
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.034505264948655e-06,
+      "loss": 0.0422,
+      "step": 15087
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.027751873868993e-06,
+      "loss": 0.0422,
+      "step": 15088
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.021001608040034e-06,
+      "loss": 0.045,
+      "step": 15089
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.014254467688652e-06,
+      "loss": 0.0429,
+      "step": 15090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.0075104530416655e-06,
+      "loss": 0.0415,
+      "step": 15091
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 7.000769564325782e-06,
+      "loss": 0.0416,
+      "step": 15092
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.994031801767597e-06,
+      "loss": 0.0436,
+      "step": 15093
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.987297165593598e-06,
+      "loss": 0.0435,
+      "step": 15094
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.980565656030158e-06,
+      "loss": 0.0411,
+      "step": 15095
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.973837273303563e-06,
+      "loss": 0.0427,
+      "step": 15096
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.967112017639987e-06,
+      "loss": 0.042,
+      "step": 15097
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.960389889265517e-06,
+      "loss": 0.0414,
+      "step": 15098
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.953670888406083e-06,
+      "loss": 0.0426,
+      "step": 15099
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.946955015287571e-06,
+      "loss": 0.0417,
+      "step": 15100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.94024227013571e-06,
+      "loss": 0.043,
+      "step": 15101
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.933532653176156e-06,
+      "loss": 0.0432,
+      "step": 15102
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.92682616463447e-06,
+      "loss": 0.0424,
+      "step": 15103
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.9201228047360625e-06,
+      "loss": 0.0426,
+      "step": 15104
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.913422573706274e-06,
+      "loss": 0.0418,
+      "step": 15105
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.906725471770348e-06,
+      "loss": 0.0413,
+      "step": 15106
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.900031499153392e-06,
+      "loss": 0.0426,
+      "step": 15107
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.893340656080438e-06,
+      "loss": 0.0417,
+      "step": 15108
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.886652942776373e-06,
+      "loss": 0.0415,
+      "step": 15109
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.879968359466016e-06,
+      "loss": 0.0414,
+      "step": 15110
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.8732869063740765e-06,
+      "loss": 0.0449,
+      "step": 15111
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.866608583725153e-06,
+      "loss": 0.0421,
+      "step": 15112
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.859933391743734e-06,
+      "loss": 0.0412,
+      "step": 15113
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.8532613306541816e-06,
+      "loss": 0.0448,
+      "step": 15114
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.846592400680796e-06,
+      "loss": 0.043,
+      "step": 15115
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.839926602047741e-06,
+      "loss": 0.0416,
+      "step": 15116
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.833263934979117e-06,
+      "loss": 0.042,
+      "step": 15117
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.826604399698855e-06,
+      "loss": 0.0425,
+      "step": 15118
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.819947996430831e-06,
+      "loss": 0.0415,
+      "step": 15119
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.813294725398789e-06,
+      "loss": 0.0422,
+      "step": 15120
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.806644586826383e-06,
+      "loss": 0.0423,
+      "step": 15121
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.799997580937168e-06,
+      "loss": 0.0417,
+      "step": 15122
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.793353707954564e-06,
+      "loss": 0.0437,
+      "step": 15123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.786712968101905e-06,
+      "loss": 0.0428,
+      "step": 15124
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.780075361602434e-06,
+      "loss": 0.0407,
+      "step": 15125
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.7734408886792835e-06,
+      "loss": 0.0423,
+      "step": 15126
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.766809549555442e-06,
+      "loss": 0.043,
+      "step": 15127
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.76018134445382e-06,
+      "loss": 0.0413,
+      "step": 15128
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.753556273597228e-06,
+      "loss": 0.0424,
+      "step": 15129
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.746934337208388e-06,
+      "loss": 0.0468,
+      "step": 15130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.740315535509867e-06,
+      "loss": 0.0472,
+      "step": 15131
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.733699868724197e-06,
+      "loss": 0.0419,
+      "step": 15132
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.727087337073712e-06,
+      "loss": 0.045,
+      "step": 15133
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.720477940780711e-06,
+      "loss": 0.0466,
+      "step": 15134
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.713871680067385e-06,
+      "loss": 0.0429,
+      "step": 15135
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.707268555155799e-06,
+      "loss": 0.0422,
+      "step": 15136
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.700668566267887e-06,
+      "loss": 0.0443,
+      "step": 15137
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.694071713625527e-06,
+      "loss": 0.0414,
+      "step": 15138
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.6874779974504865e-06,
+      "loss": 0.0413,
+      "step": 15139
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.680887417964388e-06,
+      "loss": 0.0415,
+      "step": 15140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.674299975388798e-06,
+      "loss": 0.0419,
+      "step": 15141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.667715669945107e-06,
+      "loss": 0.0422,
+      "step": 15142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.661134501854694e-06,
+      "loss": 0.0464,
+      "step": 15143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.654556471338746e-06,
+      "loss": 0.0431,
+      "step": 15144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.6479815786184116e-06,
+      "loss": 0.0422,
+      "step": 15145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.641409823914713e-06,
+      "loss": 0.0418,
+      "step": 15146
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.634841207448528e-06,
+      "loss": 0.042,
+      "step": 15147
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.62827572944067e-06,
+      "loss": 0.0423,
+      "step": 15148
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.621713390111839e-06,
+      "loss": 0.0425,
+      "step": 15149
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.615154189682638e-06,
+      "loss": 0.0422,
+      "step": 15150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.608598128373534e-06,
+      "loss": 0.0415,
+      "step": 15151
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.602045206404939e-06,
+      "loss": 0.0423,
+      "step": 15152
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.595495423997089e-06,
+      "loss": 0.0431,
+      "step": 15153
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.588948781370163e-06,
+      "loss": 0.0416,
+      "step": 15154
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.582405278744264e-06,
+      "loss": 0.0418,
+      "step": 15155
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.575864916339292e-06,
+      "loss": 0.043,
+      "step": 15156
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.569327694375149e-06,
+      "loss": 0.0416,
+      "step": 15157
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.562793613071549e-06,
+      "loss": 0.0424,
+      "step": 15158
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.556262672648161e-06,
+      "loss": 0.0421,
+      "step": 15159
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.54973487332452e-06,
+      "loss": 0.0418,
+      "step": 15160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.543210215320039e-06,
+      "loss": 0.0441,
+      "step": 15161
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.536688698854043e-06,
+      "loss": 0.0441,
+      "step": 15162
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.530170324145768e-06,
+      "loss": 0.0422,
+      "step": 15163
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.523655091414338e-06,
+      "loss": 0.0418,
+      "step": 15164
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.517143000878745e-06,
+      "loss": 0.0421,
+      "step": 15165
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5106340527578804e-06,
+      "loss": 0.0425,
+      "step": 15166
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.504128247270546e-06,
+      "loss": 0.0415,
+      "step": 15167
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.497625584635447e-06,
+      "loss": 0.0418,
+      "step": 15168
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.491126065071173e-06,
+      "loss": 0.042,
+      "step": 15169
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4846296887961956e-06,
+      "loss": 0.0419,
+      "step": 15170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.478136456028883e-06,
+      "loss": 0.0463,
+      "step": 15171
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.471646366987505e-06,
+      "loss": 0.0419,
+      "step": 15172
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.465159421890232e-06,
+      "loss": 0.0436,
+      "step": 15173
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4586756209551454e-06,
+      "loss": 0.0416,
+      "step": 15174
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.452194964400149e-06,
+      "loss": 0.0418,
+      "step": 15175
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.445717452443123e-06,
+      "loss": 0.0416,
+      "step": 15176
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.439243085301794e-06,
+      "loss": 0.0416,
+      "step": 15177
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.43277186319382e-06,
+      "loss": 0.0414,
+      "step": 15178
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.426303786336707e-06,
+      "loss": 0.0412,
+      "step": 15179
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.4198388549478685e-06,
+      "loss": 0.0463,
+      "step": 15180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.413377069244642e-06,
+      "loss": 0.0417,
+      "step": 15181
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.406918429444242e-06,
+      "loss": 0.0524,
+      "step": 15182
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.400462935763774e-06,
+      "loss": 0.0411,
+      "step": 15183
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.394010588420218e-06,
+      "loss": 0.0416,
+      "step": 15184
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.387561387630481e-06,
+      "loss": 0.0416,
+      "step": 15185
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.381115333611365e-06,
+      "loss": 0.0425,
+      "step": 15186
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.374672426579531e-06,
+      "loss": 0.0425,
+      "step": 15187
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.368232666751594e-06,
+      "loss": 0.0417,
+      "step": 15188
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.361796054343983e-06,
+      "loss": 0.0436,
+      "step": 15189
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.355362589573077e-06,
+      "loss": 0.0424,
+      "step": 15190
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.348932272655162e-06,
+      "loss": 0.0418,
+      "step": 15191
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.342505103806351e-06,
+      "loss": 0.0415,
+      "step": 15192
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.336081083242729e-06,
+      "loss": 0.0419,
+      "step": 15193
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.329660211180211e-06,
+      "loss": 0.0416,
+      "step": 15194
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.323242487834635e-06,
+      "loss": 0.0431,
+      "step": 15195
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.316827913421741e-06,
+      "loss": 0.0413,
+      "step": 15196
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.310416488157178e-06,
+      "loss": 0.042,
+      "step": 15197
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.304008212256429e-06,
+      "loss": 0.0436,
+      "step": 15198
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2976030859349115e-06,
+      "loss": 0.0421,
+      "step": 15199
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.291201109407951e-06,
+      "loss": 0.0415,
+      "step": 15200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.284802282890734e-06,
+      "loss": 0.0426,
+      "step": 15201
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.278406606598386e-06,
+      "loss": 0.043,
+      "step": 15202
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.272014080745847e-06,
+      "loss": 0.0432,
+      "step": 15203
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2656247055480565e-06,
+      "loss": 0.0425,
+      "step": 15204
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.259238481219732e-06,
+      "loss": 0.0414,
+      "step": 15205
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.252855407975588e-06,
+      "loss": 0.0415,
+      "step": 15206
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2464754860301896e-06,
+      "loss": 0.0415,
+      "step": 15207
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.240098715597975e-06,
+      "loss": 0.0419,
+      "step": 15208
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.233725096893306e-06,
+      "loss": 0.0432,
+      "step": 15209
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.227354630130445e-06,
+      "loss": 0.0419,
+      "step": 15210
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.220987315523541e-06,
+      "loss": 0.0417,
+      "step": 15211
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.214623153286592e-06,
+      "loss": 0.042,
+      "step": 15212
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.2082621436335475e-06,
+      "loss": 0.0432,
+      "step": 15213
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.201904286778248e-06,
+      "loss": 0.0432,
+      "step": 15214
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.19554958293439e-06,
+      "loss": 0.0427,
+      "step": 15215
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.1891980323156126e-06,
+      "loss": 0.042,
+      "step": 15216
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.182849635135413e-06,
+      "loss": 0.0405,
+      "step": 15217
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.176504391607163e-06,
+      "loss": 0.0423,
+      "step": 15218
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.170162301944183e-06,
+      "loss": 0.0456,
+      "step": 15219
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.163823366359656e-06,
+      "loss": 0.0428,
+      "step": 15220
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.157487585066679e-06,
+      "loss": 0.0417,
+      "step": 15221
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.1511549582781935e-06,
+      "loss": 0.0416,
+      "step": 15222
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.144825486207107e-06,
+      "loss": 0.0448,
+      "step": 15223
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.138499169066159e-06,
+      "loss": 0.041,
+      "step": 15224
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.132176007068036e-06,
+      "loss": 0.0408,
+      "step": 15225
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.125856000425256e-06,
+      "loss": 0.0423,
+      "step": 15226
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.119539149350284e-06,
+      "loss": 0.0415,
+      "step": 15227
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.113225454055449e-06,
+      "loss": 0.0407,
+      "step": 15228
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.1069149147530144e-06,
+      "loss": 0.0422,
+      "step": 15229
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.100607531655078e-06,
+      "loss": 0.0426,
+      "step": 15230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.094303304973681e-06,
+      "loss": 0.0412,
+      "step": 15231
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.088002234920709e-06,
+      "loss": 0.0426,
+      "step": 15232
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.081704321708004e-06,
+      "loss": 0.0448,
+      "step": 15233
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.075409565547252e-06,
+      "loss": 0.0421,
+      "step": 15234
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.0691179666500616e-06,
+      "loss": 0.0413,
+      "step": 15235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.062829525227909e-06,
+      "loss": 0.042,
+      "step": 15236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.05654424149219e-06,
+      "loss": 0.0415,
+      "step": 15237
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.0502621156541926e-06,
+      "loss": 0.0457,
+      "step": 15238
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.043983147925081e-06,
+      "loss": 0.0415,
+      "step": 15239
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.037707338515907e-06,
+      "loss": 0.0417,
+      "step": 15240
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.0314346876376495e-06,
+      "loss": 0.0413,
+      "step": 15241
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.02516519550117e-06,
+      "loss": 0.0426,
+      "step": 15242
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.0188988623171905e-06,
+      "loss": 0.0427,
+      "step": 15243
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.012635688296375e-06,
+      "loss": 0.0408,
+      "step": 15244
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.006375673649256e-06,
+      "loss": 0.0418,
+      "step": 15245
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.000118818586253e-06,
+      "loss": 0.0422,
+      "step": 15246
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.993865123317688e-06,
+      "loss": 0.0412,
+      "step": 15247
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.987614588053781e-06,
+      "loss": 0.0422,
+      "step": 15248
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.9813672130046735e-06,
+      "loss": 0.0413,
+      "step": 15249
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.975122998380322e-06,
+      "loss": 0.0413,
+      "step": 15250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.968881944390648e-06,
+      "loss": 0.0433,
+      "step": 15251
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.962644051245447e-06,
+      "loss": 0.043,
+      "step": 15252
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.956409319154399e-06,
+      "loss": 0.0411,
+      "step": 15253
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.950177748327091e-06,
+      "loss": 0.0415,
+      "step": 15254
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.943949338972999e-06,
+      "loss": 0.0421,
+      "step": 15255
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.937724091301478e-06,
+      "loss": 0.0438,
+      "step": 15256
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.931502005521772e-06,
+      "loss": 0.0497,
+      "step": 15257
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.9252830818430694e-06,
+      "loss": 0.0427,
+      "step": 15258
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.9190673204744255e-06,
+      "loss": 0.0445,
+      "step": 15259
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.912854721624739e-06,
+      "loss": 0.0419,
+      "step": 15260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.906645285502876e-06,
+      "loss": 0.0422,
+      "step": 15261
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.900439012317549e-06,
+      "loss": 0.043,
+      "step": 15262
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.894235902277412e-06,
+      "loss": 0.0414,
+      "step": 15263
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.888035955590954e-06,
+      "loss": 0.0425,
+      "step": 15264
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.881839172466586e-06,
+      "loss": 0.0414,
+      "step": 15265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.875645553112619e-06,
+      "loss": 0.0415,
+      "step": 15266
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.869455097737253e-06,
+      "loss": 0.0412,
+      "step": 15267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.863267806548589e-06,
+      "loss": 0.0445,
+      "step": 15268
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.857083679754605e-06,
+      "loss": 0.041,
+      "step": 15269
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.8509027175631555e-06,
+      "loss": 0.0414,
+      "step": 15270
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.844724920182032e-06,
+      "loss": 0.0413,
+      "step": 15271
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.8385502878189005e-06,
+      "loss": 0.0421,
+      "step": 15272
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.832378820681339e-06,
+      "loss": 0.0409,
+      "step": 15273
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.82621051897676e-06,
+      "loss": 0.0418,
+      "step": 15274
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.820045382912542e-06,
+      "loss": 0.0408,
+      "step": 15275
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.813883412695909e-06,
+      "loss": 0.0416,
+      "step": 15276
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.807724608534015e-06,
+      "loss": 0.0441,
+      "step": 15277
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.801568970633864e-06,
+      "loss": 0.0416,
+      "step": 15278
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.795416499202378e-06,
+      "loss": 0.0411,
+      "step": 15279
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.789267194446391e-06,
+      "loss": 0.0438,
+      "step": 15280
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.783121056572615e-06,
+      "loss": 0.0417,
+      "step": 15281
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.77697808578761e-06,
+      "loss": 0.0439,
+      "step": 15282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.77083828229793e-06,
+      "loss": 0.0438,
+      "step": 15283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.764701646309911e-06,
+      "loss": 0.0419,
+      "step": 15284
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.758568178029855e-06,
+      "loss": 0.0419,
+      "step": 15285
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.75243787766393e-06,
+      "loss": 0.0423,
+      "step": 15286
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.746310745418238e-06,
+      "loss": 0.0415,
+      "step": 15287
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.7401867814987134e-06,
+      "loss": 0.0415,
+      "step": 15288
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.7340659861112035e-06,
+      "loss": 0.0424,
+      "step": 15289
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.727948359461477e-06,
+      "loss": 0.0422,
+      "step": 15290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.72183390175518e-06,
+      "loss": 0.0423,
+      "step": 15291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.715722613197838e-06,
+      "loss": 0.0416,
+      "step": 15292
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.709614493994874e-06,
+      "loss": 0.0422,
+      "step": 15293
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.703509544351649e-06,
+      "loss": 0.0436,
+      "step": 15294
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6974077644733284e-06,
+      "loss": 0.0426,
+      "step": 15295
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.691309154565061e-06,
+      "loss": 0.0452,
+      "step": 15296
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.685213714831839e-06,
+      "loss": 0.0419,
+      "step": 15297
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.679121445478553e-06,
+      "loss": 0.0418,
+      "step": 15298
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.673032346709994e-06,
+      "loss": 0.0427,
+      "step": 15299
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.666946418730845e-06,
+      "loss": 0.0419,
+      "step": 15300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.660863661745719e-06,
+      "loss": 0.0414,
+      "step": 15301
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.65478407595903e-06,
+      "loss": 0.0431,
+      "step": 15302
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.648707661575181e-06,
+      "loss": 0.0414,
+      "step": 15303
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.64263441879842e-06,
+      "loss": 0.0423,
+      "step": 15304
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.636564347832907e-06,
+      "loss": 0.0417,
+      "step": 15305
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.630497448882666e-06,
+      "loss": 0.0426,
+      "step": 15306
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.624433722151667e-06,
+      "loss": 0.0451,
+      "step": 15307
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.618373167843693e-06,
+      "loss": 0.0426,
+      "step": 15308
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.612315786162514e-06,
+      "loss": 0.0497,
+      "step": 15309
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.606261577311722e-06,
+      "loss": 0.0444,
+      "step": 15310
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.600210541494855e-06,
+      "loss": 0.0424,
+      "step": 15311
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.594162678915283e-06,
+      "loss": 0.0439,
+      "step": 15312
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.58811798977632e-06,
+      "loss": 0.0408,
+      "step": 15313
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.582076474281173e-06,
+      "loss": 0.0437,
+      "step": 15314
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.57603813263291e-06,
+      "loss": 0.042,
+      "step": 15315
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.570002965034515e-06,
+      "loss": 0.0427,
+      "step": 15316
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.563970971688848e-06,
+      "loss": 0.0464,
+      "step": 15317
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.557942152798679e-06,
+      "loss": 0.043,
+      "step": 15318
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.55191650856669e-06,
+      "loss": 0.0424,
+      "step": 15319
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.5458940391954094e-06,
+      "loss": 0.0414,
+      "step": 15320
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.539874744887264e-06,
+      "loss": 0.0409,
+      "step": 15321
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.533858625844612e-06,
+      "loss": 0.0412,
+      "step": 15322
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.527845682269695e-06,
+      "loss": 0.0415,
+      "step": 15323
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.5218359143646145e-06,
+      "loss": 0.0415,
+      "step": 15324
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.515829322331423e-06,
+      "loss": 0.0407,
+      "step": 15325
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.5098259063719905e-06,
+      "loss": 0.0453,
+      "step": 15326
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.503825666688145e-06,
+      "loss": 0.0426,
+      "step": 15327
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.497828603481569e-06,
+      "loss": 0.0415,
+      "step": 15328
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.491834716953881e-06,
+      "loss": 0.0419,
+      "step": 15329
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.485844007306529e-06,
+      "loss": 0.0414,
+      "step": 15330
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4798564747409075e-06,
+      "loss": 0.0436,
+      "step": 15331
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.47387211945829e-06,
+      "loss": 0.0456,
+      "step": 15332
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.467890941659848e-06,
+      "loss": 0.0421,
+      "step": 15333
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4619129415466204e-06,
+      "loss": 0.0413,
+      "step": 15334
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.455938119319559e-06,
+      "loss": 0.0417,
+      "step": 15335
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.449966475179502e-06,
+      "loss": 0.0422,
+      "step": 15336
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4439980093272e-06,
+      "loss": 0.0464,
+      "step": 15337
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.438032721963271e-06,
+      "loss": 0.0414,
+      "step": 15338
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.432070613288254e-06,
+      "loss": 0.0419,
+      "step": 15339
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4261116835025325e-06,
+      "loss": 0.0413,
+      "step": 15340
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.420155932806447e-06,
+      "loss": 0.0453,
+      "step": 15341
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.414203361400183e-06,
+      "loss": 0.0421,
+      "step": 15342
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.408253969483856e-06,
+      "loss": 0.0414,
+      "step": 15343
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.40230775725743e-06,
+      "loss": 0.0429,
+      "step": 15344
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.396364724920788e-06,
+      "loss": 0.0422,
+      "step": 15345
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.390424872673727e-06,
+      "loss": 0.041,
+      "step": 15346
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.384488200715876e-06,
+      "loss": 0.0416,
+      "step": 15347
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.378554709246841e-06,
+      "loss": 0.0428,
+      "step": 15348
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.3726243984660306e-06,
+      "loss": 0.0429,
+      "step": 15349
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.3666972685728175e-06,
+      "loss": 0.0421,
+      "step": 15350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.3607733197664436e-06,
+      "loss": 0.0418,
+      "step": 15351
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.354852552246026e-06,
+      "loss": 0.0428,
+      "step": 15352
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.348934966210617e-06,
+      "loss": 0.0419,
+      "step": 15353
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.343020561859102e-06,
+      "loss": 0.041,
+      "step": 15354
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.337109339390311e-06,
+      "loss": 0.0415,
+      "step": 15355
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.331201299002952e-06,
+      "loss": 0.0414,
+      "step": 15356
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.325296440895622e-06,
+      "loss": 0.0422,
+      "step": 15357
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.319394765266816e-06,
+      "loss": 0.0448,
+      "step": 15358
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.3134962723149e-06,
+      "loss": 0.0448,
+      "step": 15359
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.307600962238157e-06,
+      "loss": 0.043,
+      "step": 15360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.301708835234753e-06,
+      "loss": 0.0414,
+      "step": 15361
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.295819891502785e-06,
+      "loss": 0.0443,
+      "step": 15362
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.2899341312401705e-06,
+      "loss": 0.0417,
+      "step": 15363
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.284051554644764e-06,
+      "loss": 0.0423,
+      "step": 15364
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.2781721619143185e-06,
+      "loss": 0.041,
+      "step": 15365
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.272295953246476e-06,
+      "loss": 0.0437,
+      "step": 15366
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.266422928838754e-06,
+      "loss": 0.0415,
+      "step": 15367
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.260553088888576e-06,
+      "loss": 0.0408,
+      "step": 15368
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.2546864335932365e-06,
+      "loss": 0.0416,
+      "step": 15369
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.248822963149969e-06,
+      "loss": 0.0428,
+      "step": 15370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.242962677755881e-06,
+      "loss": 0.0416,
+      "step": 15371
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.237105577607948e-06,
+      "loss": 0.0418,
+      "step": 15372
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.231251662903036e-06,
+      "loss": 0.0459,
+      "step": 15373
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.225400933837954e-06,
+      "loss": 0.041,
+      "step": 15374
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.219553390609355e-06,
+      "loss": 0.0411,
+      "step": 15375
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.213709033413816e-06,
+      "loss": 0.0418,
+      "step": 15376
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.207867862447802e-06,
+      "loss": 0.043,
+      "step": 15377
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.2020298779076434e-06,
+      "loss": 0.0424,
+      "step": 15378
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.196195079989585e-06,
+      "loss": 0.0414,
+      "step": 15379
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.1903634688897695e-06,
+      "loss": 0.0416,
+      "step": 15380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.18453504480424e-06,
+      "loss": 0.0408,
+      "step": 15381
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.178709807928894e-06,
+      "loss": 0.0441,
+      "step": 15382
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.172887758459555e-06,
+      "loss": 0.0413,
+      "step": 15383
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.167068896591942e-06,
+      "loss": 0.0416,
+      "step": 15384
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.161253222521633e-06,
+      "loss": 0.0414,
+      "step": 15385
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.15544073644415e-06,
+      "loss": 0.042,
+      "step": 15386
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.149631438554847e-06,
+      "loss": 0.0416,
+      "step": 15387
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.143825329049012e-06,
+      "loss": 0.0431,
+      "step": 15388
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.138022408121834e-06,
+      "loss": 0.0416,
+      "step": 15389
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.132222675968357e-06,
+      "loss": 0.0415,
+      "step": 15390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.126426132783557e-06,
+      "loss": 0.0421,
+      "step": 15391
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.1206327787622685e-06,
+      "loss": 0.0435,
+      "step": 15392
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.114842614099235e-06,
+      "loss": 0.0414,
+      "step": 15393
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.109055638989102e-06,
+      "loss": 0.0432,
+      "step": 15394
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.103271853626412e-06,
+      "loss": 0.0435,
+      "step": 15395
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0974912582055446e-06,
+      "loss": 0.0424,
+      "step": 15396
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.091713852920854e-06,
+      "loss": 0.0423,
+      "step": 15397
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.085939637966508e-06,
+      "loss": 0.0436,
+      "step": 15398
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.08016861353664e-06,
+      "loss": 0.0421,
+      "step": 15399
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.074400779825239e-06,
+      "loss": 0.042,
+      "step": 15400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.068636137026173e-06,
+      "loss": 0.0411,
+      "step": 15401
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.062874685333218e-06,
+      "loss": 0.0414,
+      "step": 15402
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.057116424940056e-06,
+      "loss": 0.0406,
+      "step": 15403
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0513613560402515e-06,
+      "loss": 0.0424,
+      "step": 15404
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.045609478827274e-06,
+      "loss": 0.0417,
+      "step": 15405
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.039860793494444e-06,
+      "loss": 0.0431,
+      "step": 15406
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.034115300235009e-06,
+      "loss": 0.0417,
+      "step": 15407
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.028372999242103e-06,
+      "loss": 0.0432,
+      "step": 15408
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.0226338907087835e-06,
+      "loss": 0.0423,
+      "step": 15409
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.016897974827939e-06,
+      "loss": 0.0414,
+      "step": 15410
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.011165251792371e-06,
+      "loss": 0.0413,
+      "step": 15411
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.005435721794805e-06,
+      "loss": 0.0422,
+      "step": 15412
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.99970938502784e-06,
+      "loss": 0.0414,
+      "step": 15413
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.993986241683968e-06,
+      "loss": 0.0416,
+      "step": 15414
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9882662919555455e-06,
+      "loss": 0.047,
+      "step": 15415
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.982549536034875e-06,
+      "loss": 0.0418,
+      "step": 15416
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.976835974114103e-06,
+      "loss": 0.0414,
+      "step": 15417
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.971125606385318e-06,
+      "loss": 0.0433,
+      "step": 15418
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.965418433040459e-06,
+      "loss": 0.0433,
+      "step": 15419
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.959714454271369e-06,
+      "loss": 0.0423,
+      "step": 15420
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.954013670269786e-06,
+      "loss": 0.0451,
+      "step": 15421
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.948316081227344e-06,
+      "loss": 0.0439,
+      "step": 15422
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.94262168733558e-06,
+      "loss": 0.0415,
+      "step": 15423
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9369304887858956e-06,
+      "loss": 0.041,
+      "step": 15424
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.931242485769583e-06,
+      "loss": 0.0427,
+      "step": 15425
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.925557678477865e-06,
+      "loss": 0.0419,
+      "step": 15426
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.919876067101836e-06,
+      "loss": 0.0409,
+      "step": 15427
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.914197651832497e-06,
+      "loss": 0.0416,
+      "step": 15428
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.908522432860685e-06,
+      "loss": 0.042,
+      "step": 15429
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.902850410377213e-06,
+      "loss": 0.0478,
+      "step": 15430
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.897181584572718e-06,
+      "loss": 0.0414,
+      "step": 15431
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.89151595563776e-06,
+      "loss": 0.042,
+      "step": 15432
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.885853523762818e-06,
+      "loss": 0.043,
+      "step": 15433
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.880194289138195e-06,
+      "loss": 0.0417,
+      "step": 15434
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.874538251954153e-06,
+      "loss": 0.0412,
+      "step": 15435
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.868885412400814e-06,
+      "loss": 0.042,
+      "step": 15436
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.863235770668173e-06,
+      "loss": 0.041,
+      "step": 15437
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.857589326946188e-06,
+      "loss": 0.0423,
+      "step": 15438
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.851946081424618e-06,
+      "loss": 0.0431,
+      "step": 15439
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8463060342931684e-06,
+      "loss": 0.0443,
+      "step": 15440
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.840669185741453e-06,
+      "loss": 0.0413,
+      "step": 15441
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.835035535958943e-06,
+      "loss": 0.0417,
+      "step": 15442
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.829405085134997e-06,
+      "loss": 0.0458,
+      "step": 15443
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.823777833458898e-06,
+      "loss": 0.042,
+      "step": 15444
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.818153781119805e-06,
+      "loss": 0.0417,
+      "step": 15445
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8125329283067565e-06,
+      "loss": 0.0416,
+      "step": 15446
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8069152752087345e-06,
+      "loss": 0.0425,
+      "step": 15447
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.801300822014521e-06,
+      "loss": 0.0413,
+      "step": 15448
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.795689568912886e-06,
+      "loss": 0.0425,
+      "step": 15449
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.790081516092437e-06,
+      "loss": 0.0413,
+      "step": 15450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7844766637416776e-06,
+      "loss": 0.0414,
+      "step": 15451
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7788750120490336e-06,
+      "loss": 0.0413,
+      "step": 15452
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.773276561202789e-06,
+      "loss": 0.0436,
+      "step": 15453
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.767681311391148e-06,
+      "loss": 0.0414,
+      "step": 15454
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7620892628021715e-06,
+      "loss": 0.0427,
+      "step": 15455
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.756500415623877e-06,
+      "loss": 0.0439,
+      "step": 15456
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.75091477004409e-06,
+      "loss": 0.042,
+      "step": 15457
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.745332326250595e-06,
+      "loss": 0.0423,
+      "step": 15458
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.739753084431031e-06,
+      "loss": 0.0417,
+      "step": 15459
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.734177044772947e-06,
+      "loss": 0.0411,
+      "step": 15460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.728604207463805e-06,
+      "loss": 0.0426,
+      "step": 15461
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7230345726909205e-06,
+      "loss": 0.0441,
+      "step": 15462
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.71746814064149e-06,
+      "loss": 0.0415,
+      "step": 15463
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.711904911502652e-06,
+      "loss": 0.0423,
+      "step": 15464
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.706344885461411e-06,
+      "loss": 0.0412,
+      "step": 15465
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.700788062704687e-06,
+      "loss": 0.0413,
+      "step": 15466
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.695234443419227e-06,
+      "loss": 0.0426,
+      "step": 15467
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.689684027791741e-06,
+      "loss": 0.0447,
+      "step": 15468
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.68413681600881e-06,
+      "loss": 0.042,
+      "step": 15469
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.67859280825691e-06,
+      "loss": 0.0412,
+      "step": 15470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.673052004722378e-06,
+      "loss": 0.0413,
+      "step": 15471
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.6675144055914664e-06,
+      "loss": 0.0456,
+      "step": 15472
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.6619800110503485e-06,
+      "loss": 0.0421,
+      "step": 15473
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.656448821285042e-06,
+      "loss": 0.0428,
+      "step": 15474
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.650920836481498e-06,
+      "loss": 0.0448,
+      "step": 15475
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.645396056825513e-06,
+      "loss": 0.0425,
+      "step": 15476
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.6398744825028155e-06,
+      "loss": 0.0425,
+      "step": 15477
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.63435611369899e-06,
+      "loss": 0.0425,
+      "step": 15478
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.6288409505995665e-06,
+      "loss": 0.0461,
+      "step": 15479
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.623328993389942e-06,
+      "loss": 0.0411,
+      "step": 15480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.6178202422553665e-06,
+      "loss": 0.0412,
+      "step": 15481
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.612314697381026e-06,
+      "loss": 0.0423,
+      "step": 15482
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.606812358952006e-06,
+      "loss": 0.0421,
+      "step": 15483
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.601313227153259e-06,
+      "loss": 0.0415,
+      "step": 15484
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.595817302169625e-06,
+      "loss": 0.0408,
+      "step": 15485
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.590324584185868e-06,
+      "loss": 0.0415,
+      "step": 15486
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.584835073386617e-06,
+      "loss": 0.0417,
+      "step": 15487
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.579348769956393e-06,
+      "loss": 0.0437,
+      "step": 15488
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.573865674079625e-06,
+      "loss": 0.0414,
+      "step": 15489
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.568385785940643e-06,
+      "loss": 0.042,
+      "step": 15490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.562909105723612e-06,
+      "loss": 0.0416,
+      "step": 15491
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.557435633612661e-06,
+      "loss": 0.041,
+      "step": 15492
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.5519653697917774e-06,
+      "loss": 0.0422,
+      "step": 15493
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.546498314444847e-06,
+      "loss": 0.0434,
+      "step": 15494
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.541034467755623e-06,
+      "loss": 0.0425,
+      "step": 15495
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.53557382990778e-06,
+      "loss": 0.0412,
+      "step": 15496
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.530116401084894e-06,
+      "loss": 0.0431,
+      "step": 15497
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.524662181470396e-06,
+      "loss": 0.0414,
+      "step": 15498
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.51921117124765e-06,
+      "loss": 0.0418,
+      "step": 15499
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.513763370599877e-06,
+      "loss": 0.0421,
+      "step": 15500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.508318779710197e-06,
+      "loss": 0.0413,
+      "step": 15501
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.502877398761629e-06,
+      "loss": 0.042,
+      "step": 15502
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.497439227937095e-06,
+      "loss": 0.0425,
+      "step": 15503
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.492004267419414e-06,
+      "loss": 0.0468,
+      "step": 15504
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.486572517391241e-06,
+      "loss": 0.0432,
+      "step": 15505
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.481143978035196e-06,
+      "loss": 0.0413,
+      "step": 15506
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.475718649533745e-06,
+      "loss": 0.0415,
+      "step": 15507
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.470296532069274e-06,
+      "loss": 0.0418,
+      "step": 15508
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.4648776258240265e-06,
+      "loss": 0.0451,
+      "step": 15509
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.459461930980158e-06,
+      "loss": 0.0427,
+      "step": 15510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.454049447719744e-06,
+      "loss": 0.0418,
+      "step": 15511
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.448640176224694e-06,
+      "loss": 0.0414,
+      "step": 15512
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.443234116676865e-06,
+      "loss": 0.0413,
+      "step": 15513
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.437831269257975e-06,
+      "loss": 0.0514,
+      "step": 15514
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.432431634149614e-06,
+      "loss": 0.0412,
+      "step": 15515
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.427035211533326e-06,
+      "loss": 0.0426,
+      "step": 15516
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.4216420015904755e-06,
+      "loss": 0.0416,
+      "step": 15517
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.416252004502397e-06,
+      "loss": 0.0415,
+      "step": 15518
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.410865220450245e-06,
+      "loss": 0.0413,
+      "step": 15519
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.405481649615095e-06,
+      "loss": 0.0454,
+      "step": 15520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.400101292177916e-06,
+      "loss": 0.0412,
+      "step": 15521
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.394724148319595e-06,
+      "loss": 0.0455,
+      "step": 15522
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.389350218220856e-06,
+      "loss": 0.0431,
+      "step": 15523
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.383979502062341e-06,
+      "loss": 0.0416,
+      "step": 15524
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3786120000245955e-06,
+      "loss": 0.042,
+      "step": 15525
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.373247712288064e-06,
+      "loss": 0.0423,
+      "step": 15526
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3678866390330365e-06,
+      "loss": 0.0419,
+      "step": 15527
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.362528780439757e-06,
+      "loss": 0.0417,
+      "step": 15528
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.357174136688291e-06,
+      "loss": 0.0438,
+      "step": 15529
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.351822707958652e-06,
+      "loss": 0.042,
+      "step": 15530
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3464744944307305e-06,
+      "loss": 0.0415,
+      "step": 15531
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3411294962843245e-06,
+      "loss": 0.0416,
+      "step": 15532
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.335787713699069e-06,
+      "loss": 0.0411,
+      "step": 15533
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.330449146854543e-06,
+      "loss": 0.045,
+      "step": 15534
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.325113795930203e-06,
+      "loss": 0.0411,
+      "step": 15535
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.319781661105404e-06,
+      "loss": 0.0412,
+      "step": 15536
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.314452742559372e-06,
+      "loss": 0.0408,
+      "step": 15537
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.309127040471239e-06,
+      "loss": 0.0421,
+      "step": 15538
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.3038045550200415e-06,
+      "loss": 0.0417,
+      "step": 15539
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.298485286384679e-06,
+      "loss": 0.0417,
+      "step": 15540
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.293169234743955e-06,
+      "loss": 0.0413,
+      "step": 15541
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.2878564002765904e-06,
+      "loss": 0.043,
+      "step": 15542
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.2825467831611565e-06,
+      "loss": 0.0411,
+      "step": 15543
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.277240383576131e-06,
+      "loss": 0.0437,
+      "step": 15544
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.2719372016999045e-06,
+      "loss": 0.0415,
+      "step": 15545
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.2666372377107464e-06,
+      "loss": 0.0416,
+      "step": 15546
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.261340491786792e-06,
+      "loss": 0.0416,
+      "step": 15547
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.256046964106109e-06,
+      "loss": 0.042,
+      "step": 15548
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.250756654846622e-06,
+      "loss": 0.0411,
+      "step": 15549
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.245469564186188e-06,
+      "loss": 0.0428,
+      "step": 15550
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.240185692302501e-06,
+      "loss": 0.0412,
+      "step": 15551
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.234905039373216e-06,
+      "loss": 0.0435,
+      "step": 15552
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.229627605575803e-06,
+      "loss": 0.0422,
+      "step": 15553
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.224353391087688e-06,
+      "loss": 0.041,
+      "step": 15554
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.219082396086149e-06,
+      "loss": 0.0416,
+      "step": 15555
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.213814620748379e-06,
+      "loss": 0.0416,
+      "step": 15556
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.208550065251449e-06,
+      "loss": 0.042,
+      "step": 15557
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.203288729772326e-06,
+      "loss": 0.0411,
+      "step": 15558
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.198030614487869e-06,
+      "loss": 0.0419,
+      "step": 15559
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.192775719574838e-06,
+      "loss": 0.0439,
+      "step": 15560
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.187524045209868e-06,
+      "loss": 0.0414,
+      "step": 15561
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.182275591569485e-06,
+      "loss": 0.0413,
+      "step": 15562
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1770303588301255e-06,
+      "loss": 0.0418,
+      "step": 15563
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1717883471681155e-06,
+      "loss": 0.0431,
+      "step": 15564
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.166549556759658e-06,
+      "loss": 0.0428,
+      "step": 15565
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.161313987780835e-06,
+      "loss": 0.0434,
+      "step": 15566
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.15608164040765e-06,
+      "loss": 0.0419,
+      "step": 15567
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.150852514816006e-06,
+      "loss": 0.0432,
+      "step": 15568
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.145626611181652e-06,
+      "loss": 0.0429,
+      "step": 15569
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.140403929680292e-06,
+      "loss": 0.0441,
+      "step": 15570
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.135184470487452e-06,
+      "loss": 0.0438,
+      "step": 15571
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1299682337786026e-06,
+      "loss": 0.041,
+      "step": 15572
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.12475521972907e-06,
+      "loss": 0.0429,
+      "step": 15573
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.119545428514115e-06,
+      "loss": 0.0411,
+      "step": 15574
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.114338860308841e-06,
+      "loss": 0.0415,
+      "step": 15575
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.109135515288265e-06,
+      "loss": 0.0418,
+      "step": 15576
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1039353936273115e-06,
+      "loss": 0.0415,
+      "step": 15577
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.098738495500787e-06,
+      "loss": 0.0418,
+      "step": 15578
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.093544821083372e-06,
+      "loss": 0.0413,
+      "step": 15579
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.0883543705496405e-06,
+      "loss": 0.0429,
+      "step": 15580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.083167144074073e-06,
+      "loss": 0.0417,
+      "step": 15581
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.077983141831054e-06,
+      "loss": 0.0419,
+      "step": 15582
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.072802363994832e-06,
+      "loss": 0.0416,
+      "step": 15583
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.067624810739579e-06,
+      "loss": 0.0412,
+      "step": 15584
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.0624504822393e-06,
+      "loss": 0.0422,
+      "step": 15585
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.057279378667955e-06,
+      "loss": 0.0459,
+      "step": 15586
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.052111500199362e-06,
+      "loss": 0.0459,
+      "step": 15587
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.046946847007249e-06,
+      "loss": 0.0425,
+      "step": 15588
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.041785419265209e-06,
+      "loss": 0.0427,
+      "step": 15589
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.036627217146749e-06,
+      "loss": 0.0417,
+      "step": 15590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.031472240825274e-06,
+      "loss": 0.0414,
+      "step": 15591
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.026320490474045e-06,
+      "loss": 0.0415,
+      "step": 15592
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.021171966266268e-06,
+      "loss": 0.042,
+      "step": 15593
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.016026668374973e-06,
+      "loss": 0.0422,
+      "step": 15594
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.01088459697313e-06,
+      "loss": 0.042,
+      "step": 15595
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.005745752233603e-06,
+      "loss": 0.0421,
+      "step": 15596
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.00061013432913e-06,
+      "loss": 0.042,
+      "step": 15597
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.995477743432352e-06,
+      "loss": 0.0454,
+      "step": 15598
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.990348579715764e-06,
+      "loss": 0.0418,
+      "step": 15599
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.985222643351805e-06,
+      "loss": 0.0439,
+      "step": 15600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.980099934512771e-06,
+      "loss": 0.0424,
+      "step": 15601
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.974980453370891e-06,
+      "loss": 0.0414,
+      "step": 15602
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.9698642000982165e-06,
+      "loss": 0.0416,
+      "step": 15603
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.964751174866765e-06,
+      "loss": 0.0415,
+      "step": 15604
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.959641377848367e-06,
+      "loss": 0.0421,
+      "step": 15605
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.954534809214827e-06,
+      "loss": 0.0417,
+      "step": 15606
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.949431469137787e-06,
+      "loss": 0.041,
+      "step": 15607
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.944331357788788e-06,
+      "loss": 0.0414,
+      "step": 15608
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.939234475339271e-06,
+      "loss": 0.0431,
+      "step": 15609
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.934140821960586e-06,
+      "loss": 0.0415,
+      "step": 15610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.92905039782393e-06,
+      "loss": 0.0412,
+      "step": 15611
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.923963203100445e-06,
+      "loss": 0.0417,
+      "step": 15612
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.918879237961115e-06,
+      "loss": 0.0433,
+      "step": 15613
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.9137985025768375e-06,
+      "loss": 0.0418,
+      "step": 15614
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.908720997118398e-06,
+      "loss": 0.0423,
+      "step": 15615
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.903646721756504e-06,
+      "loss": 0.0405,
+      "step": 15616
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.898575676661709e-06,
+      "loss": 0.0421,
+      "step": 15617
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.893507862004464e-06,
+      "loss": 0.046,
+      "step": 15618
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.888443277955123e-06,
+      "loss": 0.0418,
+      "step": 15619
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.883381924683949e-06,
+      "loss": 0.0427,
+      "step": 15620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.878323802361062e-06,
+      "loss": 0.0416,
+      "step": 15621
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.873268911156524e-06,
+      "loss": 0.0411,
+      "step": 15622
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.868217251240203e-06,
+      "loss": 0.0415,
+      "step": 15623
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.863168822781948e-06,
+      "loss": 0.0421,
+      "step": 15624
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.858123625951437e-06,
+      "loss": 0.0416,
+      "step": 15625
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.853081660918301e-06,
+      "loss": 0.0416,
+      "step": 15626
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.84804292785198e-06,
+      "loss": 0.0417,
+      "step": 15627
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.843007426921874e-06,
+      "loss": 0.0415,
+      "step": 15628
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.837975158297258e-06,
+      "loss": 0.0419,
+      "step": 15629
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.832946122147263e-06,
+      "loss": 0.0415,
+      "step": 15630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.8279203186409765e-06,
+      "loss": 0.0413,
+      "step": 15631
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.822897747947307e-06,
+      "loss": 0.0425,
+      "step": 15632
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.817878410235099e-06,
+      "loss": 0.044,
+      "step": 15633
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.8128623056730815e-06,
+      "loss": 0.0413,
+      "step": 15634
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.8078494344298664e-06,
+      "loss": 0.0417,
+      "step": 15635
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.8028397966739737e-06,
+      "loss": 0.0414,
+      "step": 15636
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7978333925737796e-06,
+      "loss": 0.0417,
+      "step": 15637
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.792830222297583e-06,
+      "loss": 0.0415,
+      "step": 15638
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7878302860135715e-06,
+      "loss": 0.0413,
+      "step": 15639
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.782833583889822e-06,
+      "loss": 0.0405,
+      "step": 15640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7778401160942666e-06,
+      "loss": 0.0415,
+      "step": 15641
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7728498827948045e-06,
+      "loss": 0.0412,
+      "step": 15642
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7678628841591456e-06,
+      "loss": 0.0416,
+      "step": 15643
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.762879120354934e-06,
+      "loss": 0.0417,
+      "step": 15644
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7578985915497245e-06,
+      "loss": 0.0414,
+      "step": 15645
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7529212979109053e-06,
+      "loss": 0.0467,
+      "step": 15646
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.747947239605787e-06,
+      "loss": 0.0429,
+      "step": 15647
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7429764168016023e-06,
+      "loss": 0.0429,
+      "step": 15648
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7380088296654183e-06,
+      "loss": 0.0414,
+      "step": 15649
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7330444783642338e-06,
+      "loss": 0.0436,
+      "step": 15650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.728083363064916e-06,
+      "loss": 0.0417,
+      "step": 15651
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7231254839342424e-06,
+      "loss": 0.043,
+      "step": 15652
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.718170841138846e-06,
+      "loss": 0.0414,
+      "step": 15653
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7132194348453277e-06,
+      "loss": 0.0442,
+      "step": 15654
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7082712652200867e-06,
+      "loss": 0.0415,
+      "step": 15655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7033263324294465e-06,
+      "loss": 0.041,
+      "step": 15656
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.698384636639651e-06,
+      "loss": 0.0416,
+      "step": 15657
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.693446178016824e-06,
+      "loss": 0.0438,
+      "step": 15658
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.688510956726965e-06,
+      "loss": 0.0416,
+      "step": 15659
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6835789729359415e-06,
+      "loss": 0.0415,
+      "step": 15660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.678650226809577e-06,
+      "loss": 0.0419,
+      "step": 15661
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.673724718513538e-06,
+      "loss": 0.0413,
+      "step": 15662
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6688024482133933e-06,
+      "loss": 0.0415,
+      "step": 15663
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6638834160746204e-06,
+      "loss": 0.0421,
+      "step": 15664
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6589676222625435e-06,
+      "loss": 0.0414,
+      "step": 15665
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.65405506694243e-06,
+      "loss": 0.0418,
+      "step": 15666
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6491457502793926e-06,
+      "loss": 0.0413,
+      "step": 15667
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6442396724384985e-06,
+      "loss": 0.0418,
+      "step": 15668
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6393368335846278e-06,
+      "loss": 0.0423,
+      "step": 15669
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.634437233882582e-06,
+      "loss": 0.0417,
+      "step": 15670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.629540873497084e-06,
+      "loss": 0.0413,
+      "step": 15671
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6246477525927134e-06,
+      "loss": 0.0425,
+      "step": 15672
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.619757871333973e-06,
+      "loss": 0.0421,
+      "step": 15673
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6148712298851973e-06,
+      "loss": 0.042,
+      "step": 15674
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6099878284106768e-06,
+      "loss": 0.0419,
+      "step": 15675
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.60510766707457e-06,
+      "loss": 0.0427,
+      "step": 15676
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6002307460409e-06,
+      "loss": 0.0417,
+      "step": 15677
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.595357065473637e-06,
+      "loss": 0.0412,
+      "step": 15678
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.590486625536571e-06,
+      "loss": 0.042,
+      "step": 15679
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5856194263934493e-06,
+      "loss": 0.0414,
+      "step": 15680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.580755468207875e-06,
+      "loss": 0.042,
+      "step": 15681
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5758947511433384e-06,
+      "loss": 0.0417,
+      "step": 15682
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.571037275363254e-06,
+      "loss": 0.0429,
+      "step": 15683
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5661830410308795e-06,
+      "loss": 0.0417,
+      "step": 15684
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.561332048309407e-06,
+      "loss": 0.0424,
+      "step": 15685
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.556484297361895e-06,
+      "loss": 0.0414,
+      "step": 15686
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.551639788351313e-06,
+      "loss": 0.042,
+      "step": 15687
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5467985214404755e-06,
+      "loss": 0.0502,
+      "step": 15688
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5419604967921626e-06,
+      "loss": 0.0417,
+      "step": 15689
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.537125714568967e-06,
+      "loss": 0.0413,
+      "step": 15690
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.532294174933437e-06,
+      "loss": 0.0412,
+      "step": 15691
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5274658780479863e-06,
+      "loss": 0.0414,
+      "step": 15692
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.5226408240748856e-06,
+      "loss": 0.0437,
+      "step": 15693
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.51781901317636e-06,
+      "loss": 0.0422,
+      "step": 15694
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.5130004455144805e-06,
+      "loss": 0.0416,
+      "step": 15695
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.5081851212512175e-06,
+      "loss": 0.0415,
+      "step": 15696
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.5033730405484634e-06,
+      "loss": 0.0413,
+      "step": 15697
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.498564203567933e-06,
+      "loss": 0.0419,
+      "step": 15698
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4937586104713094e-06,
+      "loss": 0.0414,
+      "step": 15699
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4889562614201178e-06,
+      "loss": 0.0409,
+      "step": 15700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4841571565757957e-06,
+      "loss": 0.0416,
+      "step": 15701
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4793612960996592e-06,
+      "loss": 0.043,
+      "step": 15702
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.474568680152912e-06,
+      "loss": 0.0421,
+      "step": 15703
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.469779308896681e-06,
+      "loss": 0.0421,
+      "step": 15704
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.464993182491938e-06,
+      "loss": 0.0439,
+      "step": 15705
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.460210301099587e-06,
+      "loss": 0.0444,
+      "step": 15706
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.455430664880399e-06,
+      "loss": 0.0418,
+      "step": 15707
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4506542739950133e-06,
+      "loss": 0.0421,
+      "step": 15708
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4458811286040226e-06,
+      "loss": 0.0421,
+      "step": 15709
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4411112288678548e-06,
+      "loss": 0.0427,
+      "step": 15710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4363445749468703e-06,
+      "loss": 0.0412,
+      "step": 15711
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4315811670012742e-06,
+      "loss": 0.0425,
+      "step": 15712
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4268210051912052e-06,
+      "loss": 0.0425,
+      "step": 15713
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4220640896766687e-06,
+      "loss": 0.0417,
+      "step": 15714
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.417310420617581e-06,
+      "loss": 0.0417,
+      "step": 15715
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4125599981737143e-06,
+      "loss": 0.0442,
+      "step": 15716
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4078128225047634e-06,
+      "loss": 0.0416,
+      "step": 15717
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4030688937703115e-06,
+      "loss": 0.0426,
+      "step": 15718
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3983282121298086e-06,
+      "loss": 0.0416,
+      "step": 15719
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.393590777742639e-06,
+      "loss": 0.0415,
+      "step": 15720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3888565907680413e-06,
+      "loss": 0.0517,
+      "step": 15721
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.384125651365122e-06,
+      "loss": 0.0412,
+      "step": 15722
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3793979596929425e-06,
+      "loss": 0.0458,
+      "step": 15723
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3746735159104205e-06,
+      "loss": 0.0415,
+      "step": 15724
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3699523201763728e-06,
+      "loss": 0.0419,
+      "step": 15725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.365234372649484e-06,
+      "loss": 0.0411,
+      "step": 15726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.360519673488349e-06,
+      "loss": 0.0439,
+      "step": 15727
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3558082228514643e-06,
+      "loss": 0.0416,
+      "step": 15728
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3511000208972134e-06,
+      "loss": 0.0419,
+      "step": 15729
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.346395067783836e-06,
+      "loss": 0.0409,
+      "step": 15730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.341693363669507e-06,
+      "loss": 0.0424,
+      "step": 15731
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3369949087122655e-06,
+      "loss": 0.0416,
+      "step": 15732
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3322997030700633e-06,
+      "loss": 0.0411,
+      "step": 15733
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.327607746900696e-06,
+      "loss": 0.0421,
+      "step": 15734
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.322919040361927e-06,
+      "loss": 0.0418,
+      "step": 15735
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3182335836113297e-06,
+      "loss": 0.0429,
+      "step": 15736
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.313551376806423e-06,
+      "loss": 0.0425,
+      "step": 15737
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3088724201045917e-06,
+      "loss": 0.0423,
+      "step": 15738
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3041967136631322e-06,
+      "loss": 0.0413,
+      "step": 15739
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.299524257639208e-06,
+      "loss": 0.042,
+      "step": 15740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2948550521898713e-06,
+      "loss": 0.0422,
+      "step": 15741
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.290189097472096e-06,
+      "loss": 0.0413,
+      "step": 15742
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2855263936427126e-06,
+      "loss": 0.0415,
+      "step": 15743
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2808669408584735e-06,
+      "loss": 0.0438,
+      "step": 15744
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.276210739275998e-06,
+      "loss": 0.0413,
+      "step": 15745
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2715577890518044e-06,
+      "loss": 0.0423,
+      "step": 15746
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.266908090342291e-06,
+      "loss": 0.0414,
+      "step": 15747
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2622616433037655e-06,
+      "loss": 0.0413,
+      "step": 15748
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2576184480924367e-06,
+      "loss": 0.0427,
+      "step": 15749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.252978504864346e-06,
+      "loss": 0.0412,
+      "step": 15750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2483418137754797e-06,
+      "loss": 0.0415,
+      "step": 15751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2437083749817134e-06,
+      "loss": 0.0412,
+      "step": 15752
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2390781886388e-06,
+      "loss": 0.0411,
+      "step": 15753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.23445125490236e-06,
+      "loss": 0.0438,
+      "step": 15754
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2298275739279347e-06,
+      "loss": 0.0415,
+      "step": 15755
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2252071458709566e-06,
+      "loss": 0.0423,
+      "step": 15756
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2205899708867448e-06,
+      "loss": 0.0457,
+      "step": 15757
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.215976049130498e-06,
+      "loss": 0.0414,
+      "step": 15758
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.211365380757325e-06,
+      "loss": 0.0415,
+      "step": 15759
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2067579659221913e-06,
+      "loss": 0.0426,
+      "step": 15760
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2021538047799727e-06,
+      "loss": 0.0421,
+      "step": 15761
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1975528974854564e-06,
+      "loss": 0.042,
+      "step": 15762
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1929552441933073e-06,
+      "loss": 0.042,
+      "step": 15763
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.188360845058047e-06,
+      "loss": 0.042,
+      "step": 15764
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1837697002341293e-06,
+      "loss": 0.0411,
+      "step": 15765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1791818098758862e-06,
+      "loss": 0.0446,
+      "step": 15766
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1745971741375503e-06,
+      "loss": 0.0425,
+      "step": 15767
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1700157931732086e-06,
+      "loss": 0.0426,
+      "step": 15768
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1654376671368837e-06,
+      "loss": 0.042,
+      "step": 15769
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.160862796182451e-06,
+      "loss": 0.0445,
+      "step": 15770
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.156291180463733e-06,
+      "loss": 0.0417,
+      "step": 15771
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.151722820134351e-06,
+      "loss": 0.042,
+      "step": 15772
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1471577153479147e-06,
+      "loss": 0.0422,
+      "step": 15773
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.142595866257847e-06,
+      "loss": 0.0416,
+      "step": 15774
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.138037273017502e-06,
+      "loss": 0.0425,
+      "step": 15775
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.133481935780125e-06,
+      "loss": 0.0437,
+      "step": 15776
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.128929854698859e-06,
+      "loss": 0.0414,
+      "step": 15777
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1243810299266818e-06,
+      "loss": 0.042,
+      "step": 15778
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.119835461616516e-06,
+      "loss": 0.0412,
+      "step": 15779
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1152931499211726e-06,
+      "loss": 0.0417,
+      "step": 15780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.110754094993351e-06,
+      "loss": 0.0426,
+      "step": 15781
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.106218296985597e-06,
+      "loss": 0.0414,
+      "step": 15782
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1016857560503988e-06,
+      "loss": 0.0412,
+      "step": 15783
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.097156472340135e-06,
+      "loss": 0.0445,
+      "step": 15784
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0926304460070168e-06,
+      "loss": 0.0422,
+      "step": 15785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.088107677203211e-06,
+      "loss": 0.0413,
+      "step": 15786
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0835881660807642e-06,
+      "loss": 0.0408,
+      "step": 15787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0790719127915646e-06,
+      "loss": 0.0424,
+      "step": 15788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.074558917487447e-06,
+      "loss": 0.0429,
+      "step": 15789
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0700491803201005e-06,
+      "loss": 0.0434,
+      "step": 15790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0655427014411486e-06,
+      "loss": 0.0415,
+      "step": 15791
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.061039481002048e-06,
+      "loss": 0.0422,
+      "step": 15792
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.056539519154178e-06,
+      "loss": 0.0414,
+      "step": 15793
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0520428160488167e-06,
+      "loss": 0.0409,
+      "step": 15794
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.047549371837122e-06,
+      "loss": 0.0446,
+      "step": 15795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0430591866701162e-06,
+      "loss": 0.043,
+      "step": 15796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0385722606987575e-06,
+      "loss": 0.0421,
+      "step": 15797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0340885940738695e-06,
+      "loss": 0.0418,
+      "step": 15798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0296081869461536e-06,
+      "loss": 0.0434,
+      "step": 15799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0251310394662447e-06,
+      "loss": 0.0414,
+      "step": 15800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.020657151784634e-06,
+      "loss": 0.0451,
+      "step": 15801
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0161865240517006e-06,
+      "loss": 0.042,
+      "step": 15802
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.011719156417725e-06,
+      "loss": 0.0423,
+      "step": 15803
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.0072550490328753e-06,
+      "loss": 0.0447,
+      "step": 15804
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.002794202047232e-06,
+      "loss": 0.0414,
+      "step": 15805
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9983366156107308e-06,
+      "loss": 0.0416,
+      "step": 15806
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9938822898732065e-06,
+      "loss": 0.0411,
+      "step": 15807
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9894312249843957e-06,
+      "loss": 0.0415,
+      "step": 15808
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.9849834210939454e-06,
+      "loss": 0.0414,
+      "step": 15809
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9805388783513356e-06,
+      "loss": 0.0413,
+      "step": 15810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.976097596905969e-06,
+      "loss": 0.0419,
+      "step": 15811
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9716595769071598e-06,
+      "loss": 0.0443,
+      "step": 15812
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.967224818504066e-06,
+      "loss": 0.0427,
+      "step": 15813
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9627933218457917e-06,
+      "loss": 0.0423,
+      "step": 15814
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.958365087081283e-06,
+      "loss": 0.0428,
+      "step": 15815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9539401143593883e-06,
+      "loss": 0.0411,
+      "step": 15816
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.949518403828866e-06,
+      "loss": 0.042,
+      "step": 15817
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9450999556383527e-06,
+      "loss": 0.0419,
+      "step": 15818
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9406847699363638e-06,
+      "loss": 0.0417,
+      "step": 15819
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.936272846871313e-06,
+      "loss": 0.041,
+      "step": 15820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9318641865915154e-06,
+      "loss": 0.0419,
+      "step": 15821
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9274587892451744e-06,
+      "loss": 0.0417,
+      "step": 15822
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9230566549803607e-06,
+      "loss": 0.0414,
+      "step": 15823
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.918657783945067e-06,
+      "loss": 0.0423,
+      "step": 15824
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9142621762871414e-06,
+      "loss": 0.0418,
+      "step": 15825
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.909869832154355e-06,
+      "loss": 0.0416,
+      "step": 15826
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9054807516943562e-06,
+      "loss": 0.0424,
+      "step": 15827
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.9010949350546713e-06,
+      "loss": 0.0412,
+      "step": 15828
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.896712382382749e-06,
+      "loss": 0.0417,
+      "step": 15829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8923330938258943e-06,
+      "loss": 0.0412,
+      "step": 15830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.887957069531311e-06,
+      "loss": 0.0423,
+      "step": 15831
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8835843096461146e-06,
+      "loss": 0.0412,
+      "step": 15832
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.879214814317299e-06,
+      "loss": 0.0419,
+      "step": 15833
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.874848583691714e-06,
+      "loss": 0.0435,
+      "step": 15834
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8704856179161522e-06,
+      "loss": 0.0429,
+      "step": 15835
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8661259171372857e-06,
+      "loss": 0.0418,
+      "step": 15836
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8617694815016417e-06,
+      "loss": 0.0416,
+      "step": 15837
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8574163111556693e-06,
+      "loss": 0.0426,
+      "step": 15838
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.853066406245697e-06,
+      "loss": 0.0411,
+      "step": 15839
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.848719766917951e-06,
+      "loss": 0.0411,
+      "step": 15840
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8443763933185374e-06,
+      "loss": 0.0416,
+      "step": 15841
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8400362855934725e-06,
+      "loss": 0.0411,
+      "step": 15842
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8356994438886396e-06,
+      "loss": 0.0416,
+      "step": 15843
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.831365868349811e-06,
+      "loss": 0.0433,
+      "step": 15844
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8270355591226704e-06,
+      "loss": 0.0408,
+      "step": 15845
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8227085163527677e-06,
+      "loss": 0.0433,
+      "step": 15846
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8183847401855867e-06,
+      "loss": 0.0422,
+      "step": 15847
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8140642307664445e-06,
+      "loss": 0.0426,
+      "step": 15848
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8097469882405804e-06,
+      "loss": 0.0418,
+      "step": 15849
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8054330127531116e-06,
+      "loss": 0.0459,
+      "step": 15850
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.8011223044490666e-06,
+      "loss": 0.0411,
+      "step": 15851
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7968148634733405e-06,
+      "loss": 0.0408,
+      "step": 15852
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7925106899707286e-06,
+      "loss": 0.0439,
+      "step": 15853
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7882097840859046e-06,
+      "loss": 0.0414,
+      "step": 15854
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.783912145963452e-06,
+      "loss": 0.0427,
+      "step": 15855
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7796177757478448e-06,
+      "loss": 0.0422,
+      "step": 15856
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7753266735834338e-06,
+      "loss": 0.0412,
+      "step": 15857
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.771038839614448e-06,
+      "loss": 0.0411,
+      "step": 15858
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7667542739850396e-06,
+      "loss": 0.0418,
+      "step": 15859
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7624729768392255e-06,
+      "loss": 0.0407,
+      "step": 15860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7581949483209357e-06,
+      "loss": 0.0416,
+      "step": 15861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.753920188573955e-06,
+      "loss": 0.0414,
+      "step": 15862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.749648697741991e-06,
+      "loss": 0.0439,
+      "step": 15863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7453804759686063e-06,
+      "loss": 0.0411,
+      "step": 15864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7411155233973084e-06,
+      "loss": 0.0431,
+      "step": 15865
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7368538401714495e-06,
+      "loss": 0.0416,
+      "step": 15866
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7325954264343035e-06,
+      "loss": 0.0413,
+      "step": 15867
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7283402823289783e-06,
+      "loss": 0.0416,
+      "step": 15868
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7240884079985374e-06,
+      "loss": 0.0411,
+      "step": 15869
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7198398035858997e-06,
+      "loss": 0.0411,
+      "step": 15870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.715594469233884e-06,
+      "loss": 0.0421,
+      "step": 15871
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7113524050851993e-06,
+      "loss": 0.0421,
+      "step": 15872
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7071136112824302e-06,
+      "loss": 0.0416,
+      "step": 15873
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.702878087968075e-06,
+      "loss": 0.0414,
+      "step": 15874
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.698645835284519e-06,
+      "loss": 0.0483,
+      "step": 15875
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6944168533740045e-06,
+      "loss": 0.0422,
+      "step": 15876
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6901911423787063e-06,
+      "loss": 0.0421,
+      "step": 15877
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6859687024406553e-06,
+      "loss": 0.0428,
+      "step": 15878
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.681749533701794e-06,
+      "loss": 0.0433,
+      "step": 15879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.677533636303964e-06,
+      "loss": 0.0407,
+      "step": 15880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6733210103888737e-06,
+      "loss": 0.0411,
+      "step": 15881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6691116560981223e-06,
+      "loss": 0.0429,
+      "step": 15882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6649055735732066e-06,
+      "loss": 0.0409,
+      "step": 15883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6607027629555246e-06,
+      "loss": 0.041,
+      "step": 15884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.656503224386353e-06,
+      "loss": 0.0411,
+      "step": 15885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.652306958006845e-06,
+      "loss": 0.0422,
+      "step": 15886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6481139639580875e-06,
+      "loss": 0.0407,
+      "step": 15887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6439242423809796e-06,
+      "loss": 0.0411,
+      "step": 15888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.639737793416397e-06,
+      "loss": 0.0417,
+      "step": 15889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6355546172050606e-06,
+      "loss": 0.042,
+      "step": 15890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.631374713887569e-06,
+      "loss": 0.0438,
+      "step": 15891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.627198083604432e-06,
+      "loss": 0.0416,
+      "step": 15892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6230247264960706e-06,
+      "loss": 0.0415,
+      "step": 15893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6188546427027507e-06,
+      "loss": 0.0413,
+      "step": 15894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6146878323646706e-06,
+      "loss": 0.0431,
+      "step": 15895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6105242956218636e-06,
+      "loss": 0.0413,
+      "step": 15896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6063640326143167e-06,
+      "loss": 0.0408,
+      "step": 15897
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6022070434818526e-06,
+      "loss": 0.0422,
+      "step": 15898
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5980533283642363e-06,
+      "loss": 0.0432,
+      "step": 15899
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5939028874010784e-06,
+      "loss": 0.041,
+      "step": 15900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.589755720731879e-06,
+      "loss": 0.0433,
+      "step": 15901
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5856118284960708e-06,
+      "loss": 0.0415,
+      "step": 15902
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.581471210832931e-06,
+      "loss": 0.0469,
+      "step": 15903
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5773338678816707e-06,
+      "loss": 0.0437,
+      "step": 15904
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.573199799781334e-06,
+      "loss": 0.0414,
+      "step": 15905
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.56906900667091e-06,
+      "loss": 0.0421,
+      "step": 15906
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5649414886892544e-06,
+      "loss": 0.0427,
+      "step": 15907
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5608172459751e-06,
+      "loss": 0.0417,
+      "step": 15908
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.556696278667092e-06,
+      "loss": 0.0419,
+      "step": 15909
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.552578586903753e-06,
+      "loss": 0.0414,
+      "step": 15910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.548464170823506e-06,
+      "loss": 0.0429,
+      "step": 15911
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.544353030564639e-06,
+      "loss": 0.0424,
+      "step": 15912
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.540245166265376e-06,
+      "loss": 0.0412,
+      "step": 15913
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.536140578063784e-06,
+      "loss": 0.0414,
+      "step": 15914
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5320392660978298e-06,
+      "loss": 0.0433,
+      "step": 15915
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.527941230505393e-06,
+      "loss": 0.0416,
+      "step": 15916
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5238464714242182e-06,
+      "loss": 0.0416,
+      "step": 15917
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5197549889919736e-06,
+      "loss": 0.0406,
+      "step": 15918
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.51566678334616e-06,
+      "loss": 0.0412,
+      "step": 15919
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.511581854624212e-06,
+      "loss": 0.0416,
+      "step": 15920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.507500202963464e-06,
+      "loss": 0.0413,
+      "step": 15921
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.503421828501096e-06,
+      "loss": 0.0422,
+      "step": 15922
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.4993467313742193e-06,
+      "loss": 0.041,
+      "step": 15923
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.495274911719814e-06,
+      "loss": 0.0422,
+      "step": 15924
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.491206369674737e-06,
+      "loss": 0.0414,
+      "step": 15925
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4871411053757898e-06,
+      "loss": 0.0426,
+      "step": 15926
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.483079118959586e-06,
+      "loss": 0.0419,
+      "step": 15927
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4790204105626823e-06,
+      "loss": 0.0417,
+      "step": 15928
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4749649803215147e-06,
+      "loss": 0.0422,
+      "step": 15929
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.470912828372407e-06,
+      "loss": 0.0413,
+      "step": 15930
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4668639548515726e-06,
+      "loss": 0.041,
+      "step": 15931
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4628183598951137e-06,
+      "loss": 0.0411,
+      "step": 15932
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4587760436390106e-06,
+      "loss": 0.0419,
+      "step": 15933
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4547370062191543e-06,
+      "loss": 0.0417,
+      "step": 15934
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.450701247771314e-06,
+      "loss": 0.0418,
+      "step": 15935
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4466687684311597e-06,
+      "loss": 0.0459,
+      "step": 15936
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4426395683342486e-06,
+      "loss": 0.0412,
+      "step": 15937
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.438613647615995e-06,
+      "loss": 0.0421,
+      "step": 15938
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4345910064117573e-06,
+      "loss": 0.0422,
+      "step": 15939
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4305716448567274e-06,
+      "loss": 0.048,
+      "step": 15940
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4265555630860416e-06,
+      "loss": 0.0415,
+      "step": 15941
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4225427612346918e-06,
+      "loss": 0.0419,
+      "step": 15942
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4185332394375702e-06,
+      "loss": 0.0423,
+      "step": 15943
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.414526997829436e-06,
+      "loss": 0.0414,
+      "step": 15944
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.410524036544981e-06,
+      "loss": 0.0448,
+      "step": 15945
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4065243557187757e-06,
+      "loss": 0.0433,
+      "step": 15946
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4025279554852344e-06,
+      "loss": 0.0424,
+      "step": 15947
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3985348359787163e-06,
+      "loss": 0.0406,
+      "step": 15948
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.394544997333437e-06,
+      "loss": 0.0422,
+      "step": 15949
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3905584396835324e-06,
+      "loss": 0.0411,
+      "step": 15950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.386575163163007e-06,
+      "loss": 0.0417,
+      "step": 15951
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3825951679057546e-06,
+      "loss": 0.0422,
+      "step": 15952
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3786184540455448e-06,
+      "loss": 0.0424,
+      "step": 15953
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.374645021716082e-06,
+      "loss": 0.0415,
+      "step": 15954
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.370674871050904e-06,
+      "loss": 0.0425,
+      "step": 15955
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3667080021835043e-06,
+      "loss": 0.0419,
+      "step": 15956
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.362744415247198e-06,
+      "loss": 0.0426,
+      "step": 15957
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.358784110375223e-06,
+      "loss": 0.0411,
+      "step": 15958
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3548270877007173e-06,
+      "loss": 0.0417,
+      "step": 15959
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.350873347356697e-06,
+      "loss": 0.0426,
+      "step": 15960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3469228894760442e-06,
+      "loss": 0.0416,
+      "step": 15961
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.342975714191575e-06,
+      "loss": 0.0533,
+      "step": 15962
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.339031821635973e-06,
+      "loss": 0.0468,
+      "step": 15963
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3350912119417976e-06,
+      "loss": 0.0415,
+      "step": 15964
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.331153885241533e-06,
+      "loss": 0.0409,
+      "step": 15965
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.327219841667505e-06,
+      "loss": 0.0414,
+      "step": 15966
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.323289081351976e-06,
+      "loss": 0.0412,
+      "step": 15967
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3193616044270616e-06,
+      "loss": 0.0434,
+      "step": 15968
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3154374110248013e-06,
+      "loss": 0.0428,
+      "step": 15969
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3115165012771e-06,
+      "loss": 0.0409,
+      "step": 15970
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.307598875315753e-06,
+      "loss": 0.0428,
+      "step": 15971
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.3036845332724543e-06,
+      "loss": 0.0436,
+      "step": 15972
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.299773475278788e-06,
+      "loss": 0.0415,
+      "step": 15973
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2958657014662267e-06,
+      "loss": 0.0415,
+      "step": 15974
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2919612119661205e-06,
+      "loss": 0.0418,
+      "step": 15975
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2880600069097093e-06,
+      "loss": 0.0421,
+      "step": 15976
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2841620864281544e-06,
+      "loss": 0.0409,
+      "step": 15977
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2802674506524736e-06,
+      "loss": 0.044,
+      "step": 15978
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.276376099713573e-06,
+      "loss": 0.0415,
+      "step": 15979
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2724880337422926e-06,
+      "loss": 0.0457,
+      "step": 15980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.268603252869295e-06,
+      "loss": 0.0421,
+      "step": 15981
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2647217572251745e-06,
+      "loss": 0.0415,
+      "step": 15982
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2608435469404054e-06,
+      "loss": 0.0422,
+      "step": 15983
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2569686221453835e-06,
+      "loss": 0.0427,
+      "step": 15984
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.253096982970326e-06,
+      "loss": 0.0447,
+      "step": 15985
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.249228629545386e-06,
+      "loss": 0.0476,
+      "step": 15986
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.245363562000613e-06,
+      "loss": 0.0431,
+      "step": 15987
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2415017804659157e-06,
+      "loss": 0.0419,
+      "step": 15988
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2376432850711226e-06,
+      "loss": 0.0414,
+      "step": 15989
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2337880759459195e-06,
+      "loss": 0.0414,
+      "step": 15990
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2299361532199137e-06,
+      "loss": 0.0423,
+      "step": 15991
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.226087517022568e-06,
+      "loss": 0.042,
+      "step": 15992
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.222242167483268e-06,
+      "loss": 0.0431,
+      "step": 15993
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2184001047312886e-06,
+      "loss": 0.042,
+      "step": 15994
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2145613288957478e-06,
+      "loss": 0.0435,
+      "step": 15995
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2107258401056987e-06,
+      "loss": 0.0414,
+      "step": 15996
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.206893638490082e-06,
+      "loss": 0.0417,
+      "step": 15997
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.203064724177706e-06,
+      "loss": 0.0417,
+      "step": 15998
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.199239097297279e-06,
+      "loss": 0.0416,
+      "step": 15999
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1954167579774086e-06,
+      "loss": 0.0427,
+      "step": 16000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.19159770634656e-06,
+      "loss": 0.0422,
+      "step": 16001
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1877819425331403e-06,
+      "loss": 0.0422,
+      "step": 16002
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.183969466665403e-06,
+      "loss": 0.0419,
+      "step": 16003
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1801602788715016e-06,
+      "loss": 0.0433,
+      "step": 16004
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.176354379279477e-06,
+      "loss": 0.0404,
+      "step": 16005
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1725517680172614e-06,
+      "loss": 0.0414,
+      "step": 16006
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.168752445212685e-06,
+      "loss": 0.0416,
+      "step": 16007
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1649564109934795e-06,
+      "loss": 0.0416,
+      "step": 16008
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.16116366548722e-06,
+      "loss": 0.0436,
+      "step": 16009
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1573742088214165e-06,
+      "loss": 0.0419,
+      "step": 16010
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.153588041123444e-06,
+      "loss": 0.0427,
+      "step": 16011
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1498051625205797e-06,
+      "loss": 0.0412,
+      "step": 16012
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1460255731399758e-06,
+      "loss": 0.0421,
+      "step": 16013
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1422492731086874e-06,
+      "loss": 0.041,
+      "step": 16014
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1384762625536457e-06,
+      "loss": 0.0421,
+      "step": 16015
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1347065416017055e-06,
+      "loss": 0.0418,
+      "step": 16016
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1309401103795646e-06,
+      "loss": 0.0416,
+      "step": 16017
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1271769690138332e-06,
+      "loss": 0.0442,
+      "step": 16018
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1234171176310103e-06,
+      "loss": 0.0438,
+      "step": 16019
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1196605563574833e-06,
+      "loss": 0.0414,
+      "step": 16020
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.115907285319518e-06,
+      "loss": 0.0419,
+      "step": 16021
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.112157304643314e-06,
+      "loss": 0.0419,
+      "step": 16022
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.108410614454881e-06,
+      "loss": 0.0419,
+      "step": 16023
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.104667214880196e-06,
+      "loss": 0.0424,
+      "step": 16024
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1009271060450807e-06,
+      "loss": 0.0416,
+      "step": 16025
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.097190288075268e-06,
+      "loss": 0.0423,
+      "step": 16026
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0934567610963573e-06,
+      "loss": 0.0418,
+      "step": 16027
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.089726525233848e-06,
+      "loss": 0.0413,
+      "step": 16028
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0859995806131627e-06,
+      "loss": 0.0409,
+      "step": 16029
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0822759273595337e-06,
+      "loss": 0.0425,
+      "step": 16030
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0785555655981725e-06,
+      "loss": 0.0435,
+      "step": 16031
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.074838495454123e-06,
+      "loss": 0.0413,
+      "step": 16032
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0711247170523197e-06,
+      "loss": 0.0415,
+      "step": 16033
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0674142305176282e-06,
+      "loss": 0.0428,
+      "step": 16034
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.06370703597476e-06,
+      "loss": 0.0416,
+      "step": 16035
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.0600031335483382e-06,
+      "loss": 0.0417,
+      "step": 16036
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.056302523362852e-06,
+      "loss": 0.0423,
+      "step": 16037
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0526052055427234e-06,
+      "loss": 0.0417,
+      "step": 16038
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0489111802122206e-06,
+      "loss": 0.0427,
+      "step": 16039
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0452204474955216e-06,
+      "loss": 0.0427,
+      "step": 16040
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0415330075166937e-06,
+      "loss": 0.0416,
+      "step": 16041
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0378488603996937e-06,
+      "loss": 0.0437,
+      "step": 16042
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0341680062683444e-06,
+      "loss": 0.0417,
+      "step": 16043
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.03049044524638e-06,
+      "loss": 0.0409,
+      "step": 16044
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.026816177457447e-06,
+      "loss": 0.0418,
+      "step": 16045
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.023145203025034e-06,
+      "loss": 0.052,
+      "step": 16046
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0194775220725436e-06,
+      "loss": 0.0407,
+      "step": 16047
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0158131347232656e-06,
+      "loss": 0.0419,
+      "step": 16048
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0121520411003793e-06,
+      "loss": 0.0419,
+      "step": 16049
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0084942413269527e-06,
+      "loss": 0.0413,
+      "step": 16050
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0048397355259317e-06,
+      "loss": 0.0419,
+      "step": 16051
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.0011885238201744e-06,
+      "loss": 0.0412,
+      "step": 16052
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.997540606332415e-06,
+      "loss": 0.0415,
+      "step": 16053
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9938959831852897e-06,
+      "loss": 0.0418,
+      "step": 16054
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.990254654501289e-06,
+      "loss": 0.0412,
+      "step": 16055
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.986616620402815e-06,
+      "loss": 0.0418,
+      "step": 16056
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.982981881012169e-06,
+      "loss": 0.0429,
+      "step": 16057
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9793504364515323e-06,
+      "loss": 0.0411,
+      "step": 16058
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9757222868429737e-06,
+      "loss": 0.0417,
+      "step": 16059
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.972097432308473e-06,
+      "loss": 0.0428,
+      "step": 16060
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.968475872969844e-06,
+      "loss": 0.0435,
+      "step": 16061
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9648576089488446e-06,
+      "loss": 0.0416,
+      "step": 16062
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9612426403671003e-06,
+      "loss": 0.042,
+      "step": 16063
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9576309673461357e-06,
+      "loss": 0.0417,
+      "step": 16064
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9540225900073318e-06,
+      "loss": 0.0423,
+      "step": 16065
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9504175084720133e-06,
+      "loss": 0.0413,
+      "step": 16066
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.946815722861339e-06,
+      "loss": 0.0482,
+      "step": 16067
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.943217233296413e-06,
+      "loss": 0.0431,
+      "step": 16068
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.93962203989817e-06,
+      "loss": 0.0416,
+      "step": 16069
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9360301427874705e-06,
+      "loss": 0.0423,
+      "step": 16070
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9324415420850616e-06,
+      "loss": 0.0429,
+      "step": 16071
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.928856237911558e-06,
+      "loss": 0.0415,
+      "step": 16072
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.925274230387497e-06,
+      "loss": 0.0414,
+      "step": 16073
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.921695519633293e-06,
+      "loss": 0.0423,
+      "step": 16074
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9181201057692168e-06,
+      "loss": 0.0414,
+      "step": 16075
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9145479889154716e-06,
+      "loss": 0.0426,
+      "step": 16076
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.91097916919214e-06,
+      "loss": 0.0444,
+      "step": 16077
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9074136467191804e-06,
+      "loss": 0.0412,
+      "step": 16078
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9038514216164316e-06,
+      "loss": 0.0417,
+      "step": 16079
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9002924940036638e-06,
+      "loss": 0.0411,
+      "step": 16080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8967368640005034e-06,
+      "loss": 0.0434,
+      "step": 16081
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8931845317264663e-06,
+      "loss": 0.0414,
+      "step": 16082
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8896354973009677e-06,
+      "loss": 0.0416,
+      "step": 16083
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8860897608433015e-06,
+      "loss": 0.0413,
+      "step": 16084
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8825473224726498e-06,
+      "loss": 0.0413,
+      "step": 16085
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8790081823081173e-06,
+      "loss": 0.0426,
+      "step": 16086
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8754723404686425e-06,
+      "loss": 0.0447,
+      "step": 16087
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.871939797073119e-06,
+      "loss": 0.0419,
+      "step": 16088
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8684105522402517e-06,
+      "loss": 0.0417,
+      "step": 16089
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8648846060887015e-06,
+      "loss": 0.0415,
+      "step": 16090
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8613619587369847e-06,
+      "loss": 0.0409,
+      "step": 16091
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8578426103035285e-06,
+      "loss": 0.0414,
+      "step": 16092
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8543265609066052e-06,
+      "loss": 0.0422,
+      "step": 16093
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8508138106644423e-06,
+      "loss": 0.0411,
+      "step": 16094
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8473043596950902e-06,
+      "loss": 0.0411,
+      "step": 16095
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8437982081165207e-06,
+      "loss": 0.0412,
+      "step": 16096
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8402953560466174e-06,
+      "loss": 0.0419,
+      "step": 16097
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8367958036031085e-06,
+      "loss": 0.041,
+      "step": 16098
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8332995509036333e-06,
+      "loss": 0.042,
+      "step": 16099
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8298065980657198e-06,
+      "loss": 0.0415,
+      "step": 16100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8263169452067852e-06,
+      "loss": 0.0411,
+      "step": 16101
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.822830592444147e-06,
+      "loss": 0.0425,
+      "step": 16102
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8193475398949666e-06,
+      "loss": 0.0415,
+      "step": 16103
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8158677876763397e-06,
+      "loss": 0.041,
+      "step": 16104
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8123913359052502e-06,
+      "loss": 0.0419,
+      "step": 16105
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8089181846985604e-06,
+      "loss": 0.0408,
+      "step": 16106
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.80544833417301e-06,
+      "loss": 0.0432,
+      "step": 16107
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8019817844452169e-06,
+      "loss": 0.0426,
+      "step": 16108
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7985185356317435e-06,
+      "loss": 0.041,
+      "step": 16109
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7950585878489856e-06,
+      "loss": 0.0412,
+      "step": 16110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7916019412132502e-06,
+      "loss": 0.0417,
+      "step": 16111
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.788148595840755e-06,
+      "loss": 0.041,
+      "step": 16112
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7846985518475524e-06,
+      "loss": 0.041,
+      "step": 16113
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7812518093496266e-06,
+      "loss": 0.0412,
+      "step": 16114
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.777808368462841e-06,
+      "loss": 0.0435,
+      "step": 16115
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7743682293029584e-06,
+      "loss": 0.0406,
+      "step": 16116
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7709313919856085e-06,
+      "loss": 0.0421,
+      "step": 16117
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.76749785662631e-06,
+      "loss": 0.0414,
+      "step": 16118
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7640676233404818e-06,
+      "loss": 0.0413,
+      "step": 16119
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7606406922434647e-06,
+      "loss": 0.0408,
+      "step": 16120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7572170634504115e-06,
+      "loss": 0.0418,
+      "step": 16121
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7537967370764297e-06,
+      "loss": 0.0418,
+      "step": 16122
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.750379713236472e-06,
+      "loss": 0.0415,
+      "step": 16123
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7469659920454351e-06,
+      "loss": 0.0415,
+      "step": 16124
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7435555736180387e-06,
+      "loss": 0.0418,
+      "step": 16125
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7401484580689465e-06,
+      "loss": 0.0428,
+      "step": 16126
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.736744645512678e-06,
+      "loss": 0.0426,
+      "step": 16127
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7333441360636526e-06,
+      "loss": 0.0473,
+      "step": 16128
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7299469298361793e-06,
+      "loss": 0.0424,
+      "step": 16129
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7265530269444553e-06,
+      "loss": 0.0418,
+      "step": 16130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.723162427502556e-06,
+      "loss": 0.0425,
+      "step": 16131
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7197751316244682e-06,
+      "loss": 0.0434,
+      "step": 16132
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7163911394240672e-06,
+      "loss": 0.0418,
+      "step": 16133
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7130104510150734e-06,
+      "loss": 0.0414,
+      "step": 16134
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7096330665111626e-06,
+      "loss": 0.042,
+      "step": 16135
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7062589860258328e-06,
+      "loss": 0.0414,
+      "step": 16136
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.702888209672515e-06,
+      "loss": 0.0422,
+      "step": 16137
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6995207375645307e-06,
+      "loss": 0.0417,
+      "step": 16138
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.696156569815055e-06,
+      "loss": 0.0417,
+      "step": 16139
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.692795706537209e-06,
+      "loss": 0.0414,
+      "step": 16140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6894381478439248e-06,
+      "loss": 0.0416,
+      "step": 16141
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6860838938480894e-06,
+      "loss": 0.043,
+      "step": 16142
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6827329446624573e-06,
+      "loss": 0.0413,
+      "step": 16143
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6793853003996606e-06,
+      "loss": 0.0415,
+      "step": 16144
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6760409611722428e-06,
+      "loss": 0.0417,
+      "step": 16145
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6726999270926137e-06,
+      "loss": 0.0415,
+      "step": 16146
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6693621982730834e-06,
+      "loss": 0.0411,
+      "step": 16147
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6660277748258402e-06,
+      "loss": 0.0416,
+      "step": 16148
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6626966568629944e-06,
+      "loss": 0.042,
+      "step": 16149
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6593688444964894e-06,
+      "loss": 0.0413,
+      "step": 16150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.656044337838214e-06,
+      "loss": 0.0415,
+      "step": 16151
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.652723136999912e-06,
+      "loss": 0.0427,
+      "step": 16152
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6494052420932272e-06,
+      "loss": 0.0413,
+      "step": 16153
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6460906532296928e-06,
+      "loss": 0.0416,
+      "step": 16154
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6427793705207305e-06,
+      "loss": 0.0416,
+      "step": 16155
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6394713940776296e-06,
+      "loss": 0.0414,
+      "step": 16156
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6361667240116009e-06,
+      "loss": 0.0416,
+      "step": 16157
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.632865360433744e-06,
+      "loss": 0.0412,
+      "step": 16158
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6295673034550152e-06,
+      "loss": 0.0422,
+      "step": 16159
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.626272553186281e-06,
+      "loss": 0.0413,
+      "step": 16160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6229811097382863e-06,
+      "loss": 0.0413,
+      "step": 16161
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6196929732216981e-06,
+      "loss": 0.041,
+      "step": 16162
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6164081437470279e-06,
+      "loss": 0.0414,
+      "step": 16163
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6131266214246877e-06,
+      "loss": 0.0409,
+      "step": 16164
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.609848406365e-06,
+      "loss": 0.0416,
+      "step": 16165
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6065734986781545e-06,
+      "loss": 0.0425,
+      "step": 16166
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6033018984742521e-06,
+      "loss": 0.0426,
+      "step": 16167
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6000336058632491e-06,
+      "loss": 0.042,
+      "step": 16168
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5967686209550138e-06,
+      "loss": 0.0411,
+      "step": 16169
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.593506943859291e-06,
+      "loss": 0.0414,
+      "step": 16170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5902485746857488e-06,
+      "loss": 0.0412,
+      "step": 16171
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5869935135438774e-06,
+      "loss": 0.0429,
+      "step": 16172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5837417605431337e-06,
+      "loss": 0.0415,
+      "step": 16173
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.580493315792797e-06,
+      "loss": 0.0413,
+      "step": 16174
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5772481794020688e-06,
+      "loss": 0.0416,
+      "step": 16175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5740063514800396e-06,
+      "loss": 0.0447,
+      "step": 16176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5707678321356888e-06,
+      "loss": 0.0415,
+      "step": 16177
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.567532621477874e-06,
+      "loss": 0.0416,
+      "step": 16178
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5643007196153302e-06,
+      "loss": 0.0413,
+      "step": 16179
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5610721266567263e-06,
+      "loss": 0.0411,
+      "step": 16180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.557846842710564e-06,
+      "loss": 0.0419,
+      "step": 16181
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5546248678852792e-06,
+      "loss": 0.043,
+      "step": 16182
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.551406202289174e-06,
+      "loss": 0.042,
+      "step": 16183
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5481908460304395e-06,
+      "loss": 0.0412,
+      "step": 16184
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.544978799217167e-06,
+      "loss": 0.0419,
+      "step": 16185
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.541770061957315e-06,
+      "loss": 0.0418,
+      "step": 16186
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5385646343587523e-06,
+      "loss": 0.0413,
+      "step": 16187
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5353625165292373e-06,
+      "loss": 0.0417,
+      "step": 16188
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.532163708576395e-06,
+      "loss": 0.0418,
+      "step": 16189
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5289682106077618e-06,
+      "loss": 0.0411,
+      "step": 16190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5257760227307515e-06,
+      "loss": 0.0416,
+      "step": 16191
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5225871450526673e-06,
+      "loss": 0.0423,
+      "step": 16192
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5194015776807124e-06,
+      "loss": 0.0414,
+      "step": 16193
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5162193207219454e-06,
+      "loss": 0.0467,
+      "step": 16194
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5130403742833699e-06,
+      "loss": 0.0417,
+      "step": 16195
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5098647384718223e-06,
+      "loss": 0.043,
+      "step": 16196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.506692413394073e-06,
+      "loss": 0.0425,
+      "step": 16197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5035233991567254e-06,
+      "loss": 0.0413,
+      "step": 16198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.500357695866328e-06,
+      "loss": 0.0411,
+      "step": 16199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4971953036292951e-06,
+      "loss": 0.0421,
+      "step": 16200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.494036222551931e-06,
+      "loss": 0.0412,
+      "step": 16201
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4908804527404286e-06,
+      "loss": 0.0431,
+      "step": 16202
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4877279943008471e-06,
+      "loss": 0.0412,
+      "step": 16203
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.484578847339191e-06,
+      "loss": 0.0414,
+      "step": 16204
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4814330119612974e-06,
+      "loss": 0.041,
+      "step": 16205
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4782904882729044e-06,
+      "loss": 0.0483,
+      "step": 16206
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4751512763796714e-06,
+      "loss": 0.0416,
+      "step": 16207
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4720153763871037e-06,
+      "loss": 0.0417,
+      "step": 16208
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4688827884006274e-06,
+      "loss": 0.0492,
+      "step": 16209
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4657535125255362e-06,
+      "loss": 0.0424,
+      "step": 16210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.462627548867035e-06,
+      "loss": 0.0411,
+      "step": 16211
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4595048975301728e-06,
+      "loss": 0.0414,
+      "step": 16212
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4563855586199327e-06,
+      "loss": 0.0423,
+      "step": 16213
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.453269532241186e-06,
+      "loss": 0.0407,
+      "step": 16214
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4501568184986603e-06,
+      "loss": 0.0415,
+      "step": 16215
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4470474174969828e-06,
+      "loss": 0.0422,
+      "step": 16216
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4439413293406923e-06,
+      "loss": 0.0447,
+      "step": 16217
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.440838554134194e-06,
+      "loss": 0.041,
+      "step": 16218
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4377390919817824e-06,
+      "loss": 0.0417,
+      "step": 16219
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4346429429876518e-06,
+      "loss": 0.0414,
+      "step": 16220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4315501072558746e-06,
+      "loss": 0.0406,
+      "step": 16221
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4284605848904232e-06,
+      "loss": 0.0408,
+      "step": 16222
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4253743759951477e-06,
+      "loss": 0.0424,
+      "step": 16223
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4222914806737875e-06,
+      "loss": 0.0412,
+      "step": 16224
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4192118990299707e-06,
+      "loss": 0.0427,
+      "step": 16225
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4161356311672258e-06,
+      "loss": 0.0423,
+      "step": 16226
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4130626771889477e-06,
+      "loss": 0.0416,
+      "step": 16227
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4099930371984538e-06,
+      "loss": 0.0427,
+      "step": 16228
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4069267112989171e-06,
+      "loss": 0.0451,
+      "step": 16229
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4038636995934107e-06,
+      "loss": 0.0414,
+      "step": 16230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4008040021848968e-06,
+      "loss": 0.0413,
+      "step": 16231
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3977476191762373e-06,
+      "loss": 0.0419,
+      "step": 16232
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3946945506701503e-06,
+      "loss": 0.0421,
+      "step": 16233
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.391644796769298e-06,
+      "loss": 0.0414,
+      "step": 16234
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3885983575761652e-06,
+      "loss": 0.0419,
+      "step": 16235
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.385555233193181e-06,
+      "loss": 0.043,
+      "step": 16236
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3825154237226301e-06,
+      "loss": 0.0419,
+      "step": 16237
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3794789292666866e-06,
+      "loss": 0.0416,
+      "step": 16238
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3764457499274353e-06,
+      "loss": 0.0423,
+      "step": 16239
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.373415885806828e-06,
+      "loss": 0.041,
+      "step": 16240
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3703893370067055e-06,
+      "loss": 0.0422,
+      "step": 16241
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.367366103628831e-06,
+      "loss": 0.0413,
+      "step": 16242
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3643461857748119e-06,
+      "loss": 0.0431,
+      "step": 16243
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3613295835461559e-06,
+      "loss": 0.0409,
+      "step": 16244
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.358316297044282e-06,
+      "loss": 0.042,
+      "step": 16245
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3553063263704645e-06,
+      "loss": 0.0416,
+      "step": 16246
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3522996716259007e-06,
+      "loss": 0.0416,
+      "step": 16247
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3492963329116537e-06,
+      "loss": 0.0419,
+      "step": 16248
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3462963103286874e-06,
+      "loss": 0.0428,
+      "step": 16249
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3432996039778212e-06,
+      "loss": 0.0411,
+      "step": 16250
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3403062139598076e-06,
+      "loss": 0.0417,
+      "step": 16251
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3373161403752665e-06,
+      "loss": 0.0421,
+      "step": 16252
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3343293833247173e-06,
+      "loss": 0.0421,
+      "step": 16253
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.331345942908535e-06,
+      "loss": 0.0411,
+      "step": 16254
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3283658192270398e-06,
+      "loss": 0.0414,
+      "step": 16255
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3253890123803847e-06,
+      "loss": 0.0408,
+      "step": 16256
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3224155224686451e-06,
+      "loss": 0.0436,
+      "step": 16257
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3194453495917636e-06,
+      "loss": 0.042,
+      "step": 16258
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3164784938496044e-06,
+      "loss": 0.0419,
+      "step": 16259
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3135149553418658e-06,
+      "loss": 0.0412,
+      "step": 16260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3105547341682012e-06,
+      "loss": 0.0427,
+      "step": 16261
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3075978304280978e-06,
+      "loss": 0.0409,
+      "step": 16262
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.304644244220965e-06,
+      "loss": 0.0418,
+      "step": 16263
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3016939756460566e-06,
+      "loss": 0.0426,
+      "step": 16264
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.2987470248025824e-06,
+      "loss": 0.0419,
+      "step": 16265
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.295803391789574e-06,
+      "loss": 0.0422,
+      "step": 16266
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.292863076706008e-06,
+      "loss": 0.0425,
+      "step": 16267
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.289926079650705e-06,
+      "loss": 0.0418,
+      "step": 16268
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2869924007223976e-06,
+      "loss": 0.0413,
+      "step": 16269
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2840620400196957e-06,
+      "loss": 0.0415,
+      "step": 16270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2811349976411202e-06,
+      "loss": 0.0422,
+      "step": 16271
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.278211273685037e-06,
+      "loss": 0.042,
+      "step": 16272
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2752908682497456e-06,
+      "loss": 0.0419,
+      "step": 16273
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2723737814334113e-06,
+      "loss": 0.0499,
+      "step": 16274
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2694600133340895e-06,
+      "loss": 0.0427,
+      "step": 16275
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2665495640497239e-06,
+      "loss": 0.0467,
+      "step": 16276
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2636424336781582e-06,
+      "loss": 0.0416,
+      "step": 16277
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2607386223171037e-06,
+      "loss": 0.0416,
+      "step": 16278
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2578381300641817e-06,
+      "loss": 0.0431,
+      "step": 16279
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2549409570168812e-06,
+      "loss": 0.0412,
+      "step": 16280
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2520471032726022e-06,
+      "loss": 0.0439,
+      "step": 16281
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2491565689286e-06,
+      "loss": 0.0413,
+      "step": 16282
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2462693540820636e-06,
+      "loss": 0.0417,
+      "step": 16283
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2433854588300374e-06,
+      "loss": 0.0416,
+      "step": 16284
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2405048832694777e-06,
+      "loss": 0.0458,
+      "step": 16285
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2376276274971844e-06,
+      "loss": 0.0425,
+      "step": 16286
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2347536916099023e-06,
+      "loss": 0.0417,
+      "step": 16287
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.23188307570421e-06,
+      "loss": 0.042,
+      "step": 16288
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.22901577987663e-06,
+      "loss": 0.0414,
+      "step": 16289
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.226151804223541e-06,
+      "loss": 0.042,
+      "step": 16290
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2232911488412103e-06,
+      "loss": 0.042,
+      "step": 16291
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2204338138258054e-06,
+      "loss": 0.0414,
+      "step": 16292
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2175797992733607e-06,
+      "loss": 0.042,
+      "step": 16293
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2147291052798216e-06,
+      "loss": 0.0417,
+      "step": 16294
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2118817319410114e-06,
+      "loss": 0.0419,
+      "step": 16295
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2090376793526537e-06,
+      "loss": 0.0426,
+      "step": 16296
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2061969476103385e-06,
+      "loss": 0.0411,
+      "step": 16297
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2033595368095674e-06,
+      "loss": 0.0418,
+      "step": 16298
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2005254470457195e-06,
+      "loss": 0.0421,
+      "step": 16299
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1976946784140631e-06,
+      "loss": 0.041,
+      "step": 16300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.194867231009733e-06,
+      "loss": 0.0437,
+      "step": 16301
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.192043104927787e-06,
+      "loss": 0.0418,
+      "step": 16302
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1892223002631708e-06,
+      "loss": 0.0414,
+      "step": 16303
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1864048171106867e-06,
+      "loss": 0.0412,
+      "step": 16304
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.183590655565059e-06,
+      "loss": 0.0415,
+      "step": 16305
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1807798157208783e-06,
+      "loss": 0.0458,
+      "step": 16306
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1779722976726248e-06,
+      "loss": 0.0425,
+      "step": 16307
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1751681015146786e-06,
+      "loss": 0.0417,
+      "step": 16308
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.172367227341309e-06,
+      "loss": 0.0405,
+      "step": 16309
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1695696752466623e-06,
+      "loss": 0.0417,
+      "step": 16310
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1667754453247636e-06,
+      "loss": 0.0421,
+      "step": 16311
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1639845376695602e-06,
+      "loss": 0.0419,
+      "step": 16312
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1611969523748656e-06,
+      "loss": 0.0416,
+      "step": 16313
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1584126895343827e-06,
+      "loss": 0.0415,
+      "step": 16314
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1556317492416925e-06,
+      "loss": 0.0422,
+      "step": 16315
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1528541315902752e-06,
+      "loss": 0.0417,
+      "step": 16316
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1500798366735233e-06,
+      "loss": 0.0418,
+      "step": 16317
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.147308864584673e-06,
+      "loss": 0.0431,
+      "step": 16318
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1445412154168833e-06,
+      "loss": 0.0411,
+      "step": 16319
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1417768892631796e-06,
+      "loss": 0.0417,
+      "step": 16320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.139015886216488e-06,
+      "loss": 0.0428,
+      "step": 16321
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1362582063696226e-06,
+      "loss": 0.0418,
+      "step": 16322
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1335038498152761e-06,
+      "loss": 0.0465,
+      "step": 16323
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1307528166460413e-06,
+      "loss": 0.0416,
+      "step": 16324
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1280051069543884e-06,
+      "loss": 0.0422,
+      "step": 16325
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.125260720832688e-06,
+      "loss": 0.0421,
+      "step": 16326
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.122519658373178e-06,
+      "loss": 0.042,
+      "step": 16327
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.119781919668017e-06,
+      "loss": 0.0407,
+      "step": 16328
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.117047504809221e-06,
+      "loss": 0.041,
+      "step": 16329
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1143164138887163e-06,
+      "loss": 0.0436,
+      "step": 16330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.111588646998296e-06,
+      "loss": 0.0413,
+      "step": 16331
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1088642042296648e-06,
+      "loss": 0.0444,
+      "step": 16332
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1061430856744159e-06,
+      "loss": 0.0429,
+      "step": 16333
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1034252914239873e-06,
+      "loss": 0.0418,
+      "step": 16334
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1007108215697504e-06,
+      "loss": 0.0413,
+      "step": 16335
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0979996762029654e-06,
+      "loss": 0.0419,
+      "step": 16336
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0952918554147595e-06,
+      "loss": 0.0421,
+      "step": 16337
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0925873592961489e-06,
+      "loss": 0.0422,
+      "step": 16338
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0898861879380607e-06,
+      "loss": 0.0419,
+      "step": 16339
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0871883414312777e-06,
+      "loss": 0.0409,
+      "step": 16340
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0844938198664834e-06,
+      "loss": 0.0415,
+      "step": 16341
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0818026233342827e-06,
+      "loss": 0.0416,
+      "step": 16342
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0791147519251033e-06,
+      "loss": 0.0431,
+      "step": 16343
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0764302057293174e-06,
+      "loss": 0.0453,
+      "step": 16344
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0737489848371752e-06,
+      "loss": 0.0418,
+      "step": 16345
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.071071089338782e-06,
+      "loss": 0.0431,
+      "step": 16346
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0683965193241775e-06,
+      "loss": 0.0427,
+      "step": 16347
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.065725274883256e-06,
+      "loss": 0.0429,
+      "step": 16348
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0630573561058011e-06,
+      "loss": 0.0413,
+      "step": 16349
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.060392763081508e-06,
+      "loss": 0.0414,
+      "step": 16350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0577314958999607e-06,
+      "loss": 0.0431,
+      "step": 16351
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0550735546505874e-06,
+      "loss": 0.0428,
+      "step": 16352
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0524189394227502e-06,
+      "loss": 0.0412,
+      "step": 16353
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0497676503056775e-06,
+      "loss": 0.0414,
+      "step": 16354
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.047119687388498e-06,
+      "loss": 0.0412,
+      "step": 16355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0444750507602187e-06,
+      "loss": 0.042,
+      "step": 16356
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0418337405097456e-06,
+      "loss": 0.0409,
+      "step": 16357
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0391957567258526e-06,
+      "loss": 0.0412,
+      "step": 16358
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0365610994972242e-06,
+      "loss": 0.0416,
+      "step": 16359
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0339297689124228e-06,
+      "loss": 0.0416,
+      "step": 16360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.031301765059911e-06,
+      "loss": 0.0416,
+      "step": 16361
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0286770880280072e-06,
+      "loss": 0.0428,
+      "step": 16362
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0260557379049519e-06,
+      "loss": 0.0418,
+      "step": 16363
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0234377147788631e-06,
+      "loss": 0.043,
+      "step": 16364
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0208230187377489e-06,
+      "loss": 0.0425,
+      "step": 16365
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0182116498694938e-06,
+      "loss": 0.0431,
+      "step": 16366
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0156036082618836e-06,
+      "loss": 0.0416,
+      "step": 16367
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0129988940025702e-06,
+      "loss": 0.042,
+      "step": 16368
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0103975071791283e-06,
+      "loss": 0.0425,
+      "step": 16369
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0077994478790098e-06,
+      "loss": 0.0415,
+      "step": 16370
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0052047161895339e-06,
+      "loss": 0.0424,
+      "step": 16371
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0026133121979198e-06,
+      "loss": 0.0406,
+      "step": 16372
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0000252359912976e-06,
+      "loss": 0.0411,
+      "step": 16373
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.974404876566311e-07,
+      "loss": 0.041,
+      "step": 16374
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.948590672808512e-07,
+      "loss": 0.046,
+      "step": 16375
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.922809749506878e-07,
+      "loss": 0.0427,
+      "step": 16376
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.897062107528277e-07,
+      "loss": 0.0412,
+      "step": 16377
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.871347747738234e-07,
+      "loss": 0.0412,
+      "step": 16378
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.845666671001063e-07,
+      "loss": 0.0413,
+      "step": 16379
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.820018878179959e-07,
+      "loss": 0.0409,
+      "step": 16380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.794404370137123e-07,
+      "loss": 0.0417,
+      "step": 16381
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.76882314773364e-07,
+      "loss": 0.0413,
+      "step": 16382
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.743275211829383e-07,
+      "loss": 0.0434,
+      "step": 16383
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.717760563283107e-07,
+      "loss": 0.0415,
+      "step": 16384
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.69227920295268e-07,
+      "loss": 0.0413,
+      "step": 16385
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.66683113169431e-07,
+      "loss": 0.0425,
+      "step": 16386
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.641416350363641e-07,
+      "loss": 0.0413,
+      "step": 16387
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.616034859814993e-07,
+      "loss": 0.0446,
+      "step": 16388
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.59068666090157e-07,
+      "loss": 0.0414,
+      "step": 16389
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.565371754475361e-07,
+      "loss": 0.0417,
+      "step": 16390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.540090141387349e-07,
+      "loss": 0.0415,
+      "step": 16391
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.514841822487408e-07,
+      "loss": 0.041,
+      "step": 16392
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.489626798624307e-07,
+      "loss": 0.0406,
+      "step": 16393
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.464445070645478e-07,
+      "loss": 0.0412,
+      "step": 16394
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.439296639397576e-07,
+      "loss": 0.0416,
+      "step": 16395
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.414181505725817e-07,
+      "loss": 0.0412,
+      "step": 16396
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.389099670474632e-07,
+      "loss": 0.0423,
+      "step": 16397
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.364051134486907e-07,
+      "loss": 0.0408,
+      "step": 16398
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.339035898604853e-07,
+      "loss": 0.0431,
+      "step": 16399
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.314053963669245e-07,
+      "loss": 0.0423,
+      "step": 16400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.289105330519854e-07,
+      "loss": 0.0417,
+      "step": 16401
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.264189999995343e-07,
+      "loss": 0.0423,
+      "step": 16402
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.239307972933264e-07,
+      "loss": 0.0422,
+      "step": 16403
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.21445925017006e-07,
+      "loss": 0.0408,
+      "step": 16404
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.189643832540839e-07,
+      "loss": 0.0427,
+      "step": 16405
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.164861720879935e-07,
+      "loss": 0.0413,
+      "step": 16406
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.14011291602035e-07,
+      "loss": 0.0444,
+      "step": 16407
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.115397418794081e-07,
+      "loss": 0.0417,
+      "step": 16408
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.090715230031688e-07,
+      "loss": 0.0422,
+      "step": 16409
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.066066350563174e-07,
+      "loss": 0.0417,
+      "step": 16410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.041450781216876e-07,
+      "loss": 0.0413,
+      "step": 16411
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.016868522820354e-07,
+      "loss": 0.0419,
+      "step": 16412
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.992319576199948e-07,
+      "loss": 0.0433,
+      "step": 16413
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.967803942180775e-07,
+      "loss": 0.041,
+      "step": 16414
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.943321621587064e-07,
+      "loss": 0.0427,
+      "step": 16415
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.918872615241713e-07,
+      "loss": 0.0416,
+      "step": 16416
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.89445692396651e-07,
+      "loss": 0.0419,
+      "step": 16417
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.870074548582353e-07,
+      "loss": 0.0414,
+      "step": 16418
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.845725489908696e-07,
+      "loss": 0.043,
+      "step": 16419
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.821409748763998e-07,
+      "loss": 0.0415,
+      "step": 16420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.797127325965715e-07,
+      "loss": 0.0415,
+      "step": 16421
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.772878222330194e-07,
+      "loss": 0.0421,
+      "step": 16422
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.74866243867245e-07,
+      "loss": 0.0421,
+      "step": 16423
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.724479975806388e-07,
+      "loss": 0.0419,
+      "step": 16424
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.700330834545134e-07,
+      "loss": 0.0422,
+      "step": 16425
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.676215015700262e-07,
+      "loss": 0.0418,
+      "step": 16426
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.652132520082568e-07,
+      "loss": 0.0595,
+      "step": 16427
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.628083348501515e-07,
+      "loss": 0.0421,
+      "step": 16428
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.604067501765567e-07,
+      "loss": 0.0417,
+      "step": 16429
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.580084980681857e-07,
+      "loss": 0.0411,
+      "step": 16430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.556135786056629e-07,
+      "loss": 0.0427,
+      "step": 16431
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.532219918695128e-07,
+      "loss": 0.0416,
+      "step": 16432
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.508337379400932e-07,
+      "loss": 0.0415,
+      "step": 16433
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.484488168977067e-07,
+      "loss": 0.0426,
+      "step": 16434
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.460672288225224e-07,
+      "loss": 0.041,
+      "step": 16435
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.436889737946096e-07,
+      "loss": 0.042,
+      "step": 16436
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.413140518938712e-07,
+      "loss": 0.0415,
+      "step": 16437
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.389424632001763e-07,
+      "loss": 0.0413,
+      "step": 16438
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.365742077932393e-07,
+      "loss": 0.0414,
+      "step": 16439
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.342092857526518e-07,
+      "loss": 0.0436,
+      "step": 16440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.318476971579392e-07,
+      "loss": 0.0415,
+      "step": 16441
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.294894420884713e-07,
+      "loss": 0.0419,
+      "step": 16442
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27134520623507e-07,
+      "loss": 0.0419,
+      "step": 16443
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.247829328422163e-07,
+      "loss": 0.0418,
+      "step": 16444
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.224346788236581e-07,
+      "loss": 0.0415,
+      "step": 16445
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.200897586467582e-07,
+      "loss": 0.0425,
+      "step": 16446
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.177481723903424e-07,
+      "loss": 0.0418,
+      "step": 16447
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.154099201331256e-07,
+      "loss": 0.041,
+      "step": 16448
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.130750019537114e-07,
+      "loss": 0.0432,
+      "step": 16449
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.107434179305706e-07,
+      "loss": 0.0414,
+      "step": 16450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.08415168142096e-07,
+      "loss": 0.0421,
+      "step": 16451
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.060902526665471e-07,
+      "loss": 0.0409,
+      "step": 16452
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.037686715820725e-07,
+      "loss": 0.0408,
+      "step": 16453
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.01450424966721e-07,
+      "loss": 0.0412,
+      "step": 16454
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.991355128984079e-07,
+      "loss": 0.0414,
+      "step": 16455
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.968239354549601e-07,
+      "loss": 0.0419,
+      "step": 16456
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.945156927140706e-07,
+      "loss": 0.0415,
+      "step": 16457
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.922107847533333e-07,
+      "loss": 0.0422,
+      "step": 16458
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.899092116502305e-07,
+      "loss": 0.0439,
+      "step": 16459
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.876109734821225e-07,
+      "loss": 0.0417,
+      "step": 16460
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.853160703262808e-07,
+      "loss": 0.0415,
+      "step": 16461
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.830245022598215e-07,
+      "loss": 0.0415,
+      "step": 16462
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.807362693598053e-07,
+      "loss": 0.0416,
+      "step": 16463
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.784513717031372e-07,
+      "loss": 0.0432,
+      "step": 16464
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.761698093666115e-07,
+      "loss": 0.0422,
+      "step": 16465
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.738915824269444e-07,
+      "loss": 0.0412,
+      "step": 16466
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.716166909607081e-07,
+      "loss": 0.0413,
+      "step": 16467
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.693451350443748e-07,
+      "loss": 0.0442,
+      "step": 16468
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.670769147542944e-07,
+      "loss": 0.0408,
+      "step": 16469
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.648120301667172e-07,
+      "loss": 0.0416,
+      "step": 16470
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.625504813577822e-07,
+      "loss": 0.0412,
+      "step": 16471
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.602922684035063e-07,
+      "loss": 0.0417,
+      "step": 16472
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.580373913798067e-07,
+      "loss": 0.0415,
+      "step": 16473
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.557858503624782e-07,
+      "loss": 0.0413,
+      "step": 16474
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.535376454271936e-07,
+      "loss": 0.0417,
+      "step": 16475
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.512927766495259e-07,
+      "loss": 0.0411,
+      "step": 16476
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.49051244104959e-07,
+      "loss": 0.0416,
+      "step": 16477
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.468130478688218e-07,
+      "loss": 0.0417,
+      "step": 16478
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.445781880163649e-07,
+      "loss": 0.0416,
+      "step": 16479
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.423466646227062e-07,
+      "loss": 0.0439,
+      "step": 16480
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.401184777628523e-07,
+      "loss": 0.0419,
+      "step": 16481
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.378936275116988e-07,
+      "loss": 0.0414,
+      "step": 16482
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.356721139440526e-07,
+      "loss": 0.042,
+      "step": 16483
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.334539371345872e-07,
+      "loss": 0.0414,
+      "step": 16484
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.312390971578431e-07,
+      "loss": 0.0424,
+      "step": 16485
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.29027594088294e-07,
+      "loss": 0.0418,
+      "step": 16486
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.268194280002805e-07,
+      "loss": 0.0423,
+      "step": 16487
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.246145989680208e-07,
+      "loss": 0.0413,
+      "step": 16488
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.224131070656226e-07,
+      "loss": 0.0419,
+      "step": 16489
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.202149523671042e-07,
+      "loss": 0.0413,
+      "step": 16490
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.180201349463511e-07,
+      "loss": 0.0413,
+      "step": 16491
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.158286548771376e-07,
+      "loss": 0.0413,
+      "step": 16492
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.13640512233138e-07,
+      "loss": 0.0411,
+      "step": 16493
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.114557070879047e-07,
+      "loss": 0.0509,
+      "step": 16494
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 7.092742395148677e-07,
+      "loss": 0.0419,
+      "step": 16495
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.070961095873685e-07,
+      "loss": 0.0438,
+      "step": 16496
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.049213173786151e-07,
+      "loss": 0.0411,
+      "step": 16497
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.027498629617379e-07,
+      "loss": 0.0413,
+      "step": 16498
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.005817464096898e-07,
+      "loss": 0.042,
+      "step": 16499
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.9841696779539e-07,
+      "loss": 0.0412,
+      "step": 16500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.962555271915805e-07,
+      "loss": 0.0419,
+      "step": 16501
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.940974246709364e-07,
+      "loss": 0.0415,
+      "step": 16502
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.919426603059887e-07,
+      "loss": 0.0413,
+      "step": 16503
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.897912341691793e-07,
+      "loss": 0.0416,
+      "step": 16504
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.87643146332817e-07,
+      "loss": 0.0416,
+      "step": 16505
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.85498396869122e-07,
+      "loss": 0.0418,
+      "step": 16506
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.83356985850181e-07,
+      "loss": 0.0414,
+      "step": 16507
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.812189133479918e-07,
+      "loss": 0.0427,
+      "step": 16508
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.790841794343971e-07,
+      "loss": 0.0421,
+      "step": 16509
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.76952784181184e-07,
+      "loss": 0.0423,
+      "step": 16510
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.748247276599839e-07,
+      "loss": 0.0411,
+      "step": 16511
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.727000099423397e-07,
+      "loss": 0.0418,
+      "step": 16512
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.705786310996609e-07,
+      "loss": 0.0415,
+      "step": 16513
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.684605912032571e-07,
+      "loss": 0.0411,
+      "step": 16514
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.663458903243491e-07,
+      "loss": 0.0419,
+      "step": 16515
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.642345285340023e-07,
+      "loss": 0.0413,
+      "step": 16516
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.621265059031933e-07,
+      "loss": 0.0423,
+      "step": 16517
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.600218225027876e-07,
+      "loss": 0.0414,
+      "step": 16518
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.579204784035287e-07,
+      "loss": 0.0417,
+      "step": 16519
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.558224736760599e-07,
+      "loss": 0.0414,
+      "step": 16520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.537278083908916e-07,
+      "loss": 0.0416,
+      "step": 16521
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.516364826184563e-07,
+      "loss": 0.042,
+      "step": 16522
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.495484964290421e-07,
+      "loss": 0.041,
+      "step": 16523
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.474638498928265e-07,
+      "loss": 0.0447,
+      "step": 16524
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.453825430798976e-07,
+      "loss": 0.0409,
+      "step": 16525
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.433045760602107e-07,
+      "loss": 0.041,
+      "step": 16526
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.412299489036322e-07,
+      "loss": 0.0421,
+      "step": 16527
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.391586616798839e-07,
+      "loss": 0.0416,
+      "step": 16528
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.370907144585881e-07,
+      "loss": 0.0414,
+      "step": 16529
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.350261073092778e-07,
+      "loss": 0.0432,
+      "step": 16530
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.329648403013422e-07,
+      "loss": 0.0415,
+      "step": 16531
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.309069135040702e-07,
+      "loss": 0.0409,
+      "step": 16532
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.288523269866509e-07,
+      "loss": 0.0421,
+      "step": 16533
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.268010808181291e-07,
+      "loss": 0.0419,
+      "step": 16534
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.247531750674718e-07,
+      "loss": 0.0412,
+      "step": 16535
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.227086098035129e-07,
+      "loss": 0.0414,
+      "step": 16536
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.206673850949863e-07,
+      "loss": 0.041,
+      "step": 16537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.186295010104926e-07,
+      "loss": 0.0418,
+      "step": 16538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.165949576185548e-07,
+      "loss": 0.0424,
+      "step": 16539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.145637549875516e-07,
+      "loss": 0.0425,
+      "step": 16540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.125358931857617e-07,
+      "loss": 0.0418,
+      "step": 16541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.105113722813638e-07,
+      "loss": 0.0415,
+      "step": 16542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.084901923423925e-07,
+      "loss": 0.0411,
+      "step": 16543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.064723534368044e-07,
+      "loss": 0.0409,
+      "step": 16544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.044578556324232e-07,
+      "loss": 0.0419,
+      "step": 16545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.024466989969723e-07,
+      "loss": 0.0438,
+      "step": 16546
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.004388835980423e-07,
+      "loss": 0.0449,
+      "step": 16547
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.984344095031347e-07,
+      "loss": 0.0413,
+      "step": 16548
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.964332767796399e-07,
+      "loss": 0.0424,
+      "step": 16549
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.944354854948042e-07,
+      "loss": 0.0414,
+      "step": 16550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.924410357157961e-07,
+      "loss": 0.0415,
+      "step": 16551
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.90449927509662e-07,
+      "loss": 0.0415,
+      "step": 16552
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.884621609433261e-07,
+      "loss": 0.0421,
+      "step": 16553
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.864777360836016e-07,
+      "loss": 0.0428,
+      "step": 16554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.844966529972018e-07,
+      "loss": 0.0414,
+      "step": 16555
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.82518911750729e-07,
+      "loss": 0.0416,
+      "step": 16556
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.805445124106524e-07,
+      "loss": 0.0432,
+      "step": 16557
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.785734550433408e-07,
+      "loss": 0.0419,
+      "step": 16558
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.766057397150637e-07,
+      "loss": 0.0413,
+      "step": 16559
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.746413664919459e-07,
+      "loss": 0.042,
+      "step": 16560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.726803354400345e-07,
+      "loss": 0.0416,
+      "step": 16561
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.707226466252546e-07,
+      "loss": 0.0426,
+      "step": 16562
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.68768300113387e-07,
+      "loss": 0.0431,
+      "step": 16563
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.668172959701568e-07,
+      "loss": 0.0415,
+      "step": 16564
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.648696342611337e-07,
+      "loss": 0.0418,
+      "step": 16565
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.629253150517877e-07,
+      "loss": 0.042,
+      "step": 16566
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.609843384074664e-07,
+      "loss": 0.0434,
+      "step": 16567
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.5904670439344e-07,
+      "loss": 0.0416,
+      "step": 16568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.571124130748339e-07,
+      "loss": 0.042,
+      "step": 16569
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.55181464516652e-07,
+      "loss": 0.0421,
+      "step": 16570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.532538587838198e-07,
+      "loss": 0.0412,
+      "step": 16571
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.51329595941119e-07,
+      "loss": 0.0425,
+      "step": 16572
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.494086760532646e-07,
+      "loss": 0.0413,
+      "step": 16573
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.474910991847826e-07,
+      "loss": 0.0416,
+      "step": 16574
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.455768654001658e-07,
+      "loss": 0.0426,
+      "step": 16575
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.436659747637518e-07,
+      "loss": 0.0414,
+      "step": 16576
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.41758427339778e-07,
+      "loss": 0.0425,
+      "step": 16577
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.398542231923598e-07,
+      "loss": 0.0414,
+      "step": 16578
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.379533623855015e-07,
+      "loss": 0.0421,
+      "step": 16579
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.360558449831188e-07,
+      "loss": 0.042,
+      "step": 16580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.341616710489827e-07,
+      "loss": 0.0414,
+      "step": 16581
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.322708406467647e-07,
+      "loss": 0.0414,
+      "step": 16582
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.30383353840036e-07,
+      "loss": 0.0418,
+      "step": 16583
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.284992106922348e-07,
+      "loss": 0.0441,
+      "step": 16584
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.266184112666994e-07,
+      "loss": 0.0416,
+      "step": 16585
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.247409556266458e-07,
+      "loss": 0.0411,
+      "step": 16586
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.228668438352014e-07,
+      "loss": 0.0415,
+      "step": 16587
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.209960759553489e-07,
+      "loss": 0.042,
+      "step": 16588
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 17145,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 572,
+  "total_flos": 3.836646565638413e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}