diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,85165 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.99996476515979,
+  "global_step": 14190,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253522e-07,
+      "loss": 5.0371,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 5.3616,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225352112676057e-07,
+      "loss": 5.4968,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 5.8604,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.042253521126762e-07,
+      "loss": 5.2722,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 5.9387,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.859154929577465e-07,
+      "loss": 4.6641,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 1.4037,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.267605633802817e-06,
+      "loss": 4.8589,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 5.1802,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5492957746478873e-06,
+      "loss": 4.2783,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 4.7039,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8309859154929579e-06,
+      "loss": 4.1821,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 3.9695,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1126760563380285e-06,
+      "loss": 3.5927,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 3.5366,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3943661971830984e-06,
+      "loss": 2.7949,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 3.9124,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.676056338028169e-06,
+      "loss": 3.2366,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 2.9004,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9577464788732396e-06,
+      "loss": 2.8367,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 2.7297,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2394366197183104e-06,
+      "loss": 2.5774,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 2.3713,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5211267605633804e-06,
+      "loss": 2.5123,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 2.304,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8028169014084508e-06,
+      "loss": 2.136,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 2.0724,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0845070422535216e-06,
+      "loss": 2.1901,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 2.1709,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3661971830985915e-06,
+      "loss": 2.1266,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 2.1449,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.647887323943662e-06,
+      "loss": 1.939,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 2.0326,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.929577464788733e-06,
+      "loss": 2.0133,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 1.9166,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.211267605633803e-06,
+      "loss": 1.9117,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 1.9699,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.492957746478874e-06,
+      "loss": 1.8115,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 1.8557,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.774647887323944e-06,
+      "loss": 1.8884,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 1.7433,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.056338028169015e-06,
+      "loss": 1.7396,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 1.7341,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.3380281690140855e-06,
+      "loss": 1.6259,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 1.6741,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.619718309859155e-06,
+      "loss": 1.0562,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 1.7396,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.901408450704225e-06,
+      "loss": 1.7028,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 1.6832,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.183098591549297e-06,
+      "loss": 1.6491,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 1.6183,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.464788732394367e-06,
+      "loss": 1.584,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 1.7493,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.746478873239436e-06,
+      "loss": 1.597,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 1.6552,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.028169014084509e-06,
+      "loss": 1.5762,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 1.638,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.309859154929578e-06,
+      "loss": 1.6046,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 1.6345,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.591549295774648e-06,
+      "loss": 1.5895,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 1.5041,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.87323943661972e-06,
+      "loss": 1.4894,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 1.6545,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.15492957746479e-06,
+      "loss": 1.6372,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 1.5523,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.43661971830986e-06,
+      "loss": 1.5197,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 1.6736,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.71830985915493e-06,
+      "loss": 1.5159,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 1.45,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 1.5231,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 1.4529,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0281690140845072e-05,
+      "loss": 1.5339,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 1.4646,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.056338028169014e-05,
+      "loss": 1.3995,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 1.3774,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0845070422535213e-05,
+      "loss": 1.4755,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 1.4822,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1126760563380284e-05,
+      "loss": 1.5201,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 1.5568,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1408450704225353e-05,
+      "loss": 1.4735,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 1.4301,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1690140845070424e-05,
+      "loss": 1.4363,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 1.5016,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1971830985915493e-05,
+      "loss": 1.5097,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 1.509,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2253521126760564e-05,
+      "loss": 1.514,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 1.4034,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2535211267605636e-05,
+      "loss": 1.3438,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 0.9382,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2816901408450705e-05,
+      "loss": 1.3337,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 1.3311,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3098591549295776e-05,
+      "loss": 1.4547,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 1.542,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3380281690140845e-05,
+      "loss": 1.4493,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 1.4072,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3661971830985916e-05,
+      "loss": 1.4177,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 1.4432,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3943661971830987e-05,
+      "loss": 1.3868,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 0.9232,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4225352112676058e-05,
+      "loss": 1.4075,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 0.9558,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4507042253521129e-05,
+      "loss": 1.3023,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 1.3247,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.47887323943662e-05,
+      "loss": 1.3325,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 1.3033,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5070422535211269e-05,
+      "loss": 1.3592,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 1.3703,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.535211267605634e-05,
+      "loss": 1.3104,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 1.3585,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.563380281690141e-05,
+      "loss": 1.4681,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 0.8796,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.591549295774648e-05,
+      "loss": 1.3325,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 1.3504,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.619718309859155e-05,
+      "loss": 1.2983,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 1.314,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6478873239436623e-05,
+      "loss": 1.2737,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 1.3381,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.676056338028169e-05,
+      "loss": 1.2546,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 1.3272,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.704225352112676e-05,
+      "loss": 1.4229,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 1.345,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.732394366197183e-05,
+      "loss": 1.2849,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 1.4736,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7605633802816902e-05,
+      "loss": 1.4219,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 1.4481,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7887323943661975e-05,
+      "loss": 1.2918,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 1.2253,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8169014084507044e-05,
+      "loss": 1.4067,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 0.8477,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8450704225352113e-05,
+      "loss": 0.8864,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 0.894,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8732394366197186e-05,
+      "loss": 1.3339,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 1.325,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9014084507042255e-05,
+      "loss": 1.2922,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 1.294,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9295774647887327e-05,
+      "loss": 1.2751,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 1.277,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9577464788732396e-05,
+      "loss": 0.845,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 1.3497,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9859154929577465e-05,
+      "loss": 1.203,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.3991,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999999749942e-05,
+      "loss": 1.3217,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999899976802e-05,
+      "loss": 1.3824,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999999774947809e-05,
+      "loss": 1.331,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999995999072276e-05,
+      "loss": 1.2451,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999993748550662e-05,
+      "loss": 1.3347,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999990997913365e-05,
+      "loss": 1.2969,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999987747160525e-05,
+      "loss": 1.3514,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999983996292297e-05,
+      "loss": 1.3089,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999979745308872e-05,
+      "loss": 1.2889,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999974994210465e-05,
+      "loss": 1.2267,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999969742997308e-05,
+      "loss": 1.4247,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999996399166967e-05,
+      "loss": 1.3365,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999957740227836e-05,
+      "loss": 1.2883,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999950988672118e-05,
+      "loss": 1.1403,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999994373700285e-05,
+      "loss": 1.2384,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999935985220403e-05,
+      "loss": 1.2683,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999927733325163e-05,
+      "loss": 1.2209,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999918981317533e-05,
+      "loss": 1.3217,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999909729197965e-05,
+      "loss": 1.2523,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999989997696691e-05,
+      "loss": 1.2339,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999988972462486e-05,
+      "loss": 1.2449,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999987897217233e-05,
+      "loss": 1.3018,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999867719609852e-05,
+      "loss": 1.2521,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999855966937994e-05,
+      "loss": 1.2355,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999843714157346e-05,
+      "loss": 1.2472,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999830961268513e-05,
+      "loss": 1.2939,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999817708272136e-05,
+      "loss": 1.3657,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999803955168883e-05,
+      "loss": 1.3818,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999789701959437e-05,
+      "loss": 1.3711,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999774948644507e-05,
+      "loss": 1.1742,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999975969522484e-05,
+      "loss": 1.2164,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999974394170119e-05,
+      "loss": 1.2466,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999727688074348e-05,
+      "loss": 1.1839,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999710934345132e-05,
+      "loss": 1.2563,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999693680514373e-05,
+      "loss": 1.3188,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999675926582937e-05,
+      "loss": 1.3164,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999657672551712e-05,
+      "loss": 0.8187,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999963891842161e-05,
+      "loss": 1.2756,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999619664193567e-05,
+      "loss": 1.2175,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999959990986855e-05,
+      "loss": 1.3435,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999579655447542e-05,
+      "loss": 1.3417,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999955890093156e-05,
+      "loss": 1.2903,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999537646321645e-05,
+      "loss": 1.2686,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999515891618856e-05,
+      "loss": 1.3678,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999493636824277e-05,
+      "loss": 1.3113,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999470881939028e-05,
+      "loss": 1.1826,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999447626964242e-05,
+      "loss": 1.2009,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999423871901082e-05,
+      "loss": 1.2231,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999939961675074e-05,
+      "loss": 1.2407,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999374861514428e-05,
+      "loss": 1.2554,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999934960619338e-05,
+      "loss": 1.233,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999323850788864e-05,
+      "loss": 0.8094,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999929759530217e-05,
+      "loss": 1.1398,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999270839734604e-05,
+      "loss": 1.3124,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999243584087507e-05,
+      "loss": 1.2882,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999215828362245e-05,
+      "loss": 1.2507,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99991875725602e-05,
+      "loss": 1.3073,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999915881668279e-05,
+      "loss": 1.3008,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999912956073145e-05,
+      "loss": 1.1382,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999909980470765e-05,
+      "loss": 1.2704,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999069548612867e-05,
+      "loss": 1.1934,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999038792448623e-05,
+      "loss": 1.1536,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999007536216456e-05,
+      "loss": 0.8169,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998975779917923e-05,
+      "loss": 1.2675,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998943523554616e-05,
+      "loss": 0.7754,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998910767128152e-05,
+      "loss": 1.2249,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999887751064016e-05,
+      "loss": 1.3841,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999884375409231e-05,
+      "loss": 1.2598,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999880949748629e-05,
+      "loss": 1.2066,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998774740823813e-05,
+      "loss": 1.2344,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998739484106613e-05,
+      "loss": 1.2281,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998703727336455e-05,
+      "loss": 1.1683,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998667470515132e-05,
+      "loss": 1.2437,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998630713644452e-05,
+      "loss": 1.2592,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998593456726256e-05,
+      "loss": 1.2293,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998555699762407e-05,
+      "loss": 1.3344,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998517442754792e-05,
+      "loss": 1.0955,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998478685705326e-05,
+      "loss": 1.2182,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998439428615946e-05,
+      "loss": 0.768,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998399671488614e-05,
+      "loss": 0.7598,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999835941432532e-05,
+      "loss": 1.1879,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998318657128076e-05,
+      "loss": 1.2317,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998277399898923e-05,
+      "loss": 1.2797,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998235642639926e-05,
+      "loss": 1.1558,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998193385353166e-05,
+      "loss": 1.2262,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998150628040762e-05,
+      "loss": 1.2687,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999810737070485e-05,
+      "loss": 1.162,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998063613347595e-05,
+      "loss": 1.2267,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999801935597119e-05,
+      "loss": 1.2942,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997974598577834e-05,
+      "loss": 1.2805,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999792934116978e-05,
+      "loss": 0.785,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997883583749283e-05,
+      "loss": 1.1939,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997837326318638e-05,
+      "loss": 1.2498,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999779056888015e-05,
+      "loss": 1.1612,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997743311436167e-05,
+      "loss": 1.2287,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997695553989044e-05,
+      "loss": 1.1798,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999764729654117e-05,
+      "loss": 1.1565,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997598539094967e-05,
+      "loss": 1.2304,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997549281652865e-05,
+      "loss": 1.2687,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999749952421733e-05,
+      "loss": 1.214,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997449266790855e-05,
+      "loss": 1.1943,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999739850937595e-05,
+      "loss": 1.1861,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997347251975142e-05,
+      "loss": 1.189,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997295494591014e-05,
+      "loss": 1.2864,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997243237226142e-05,
+      "loss": 1.2323,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997190479883147e-05,
+      "loss": 1.2303,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999713722256466e-05,
+      "loss": 1.1677,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9997083465273347e-05,
+      "loss": 1.1629,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99970292080119e-05,
+      "loss": 1.2206,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999697445078303e-05,
+      "loss": 1.1772,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996919193589475e-05,
+      "loss": 1.1541,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996863436433998e-05,
+      "loss": 1.2593,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999680717931939e-05,
+      "loss": 1.1852,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999675042224846e-05,
+      "loss": 1.2604,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999669316522405e-05,
+      "loss": 0.7856,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996635408249027e-05,
+      "loss": 1.1661,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996577151326273e-05,
+      "loss": 0.7684,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996518394458703e-05,
+      "loss": 0.8033,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999645913764926e-05,
+      "loss": 1.2557,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99963993809009e-05,
+      "loss": 1.1522,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996339124216616e-05,
+      "loss": 1.2866,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999627836759942e-05,
+      "loss": 1.1664,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996217111052355e-05,
+      "loss": 1.1187,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996155354578485e-05,
+      "loss": 1.2268,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996093098180887e-05,
+      "loss": 1.1512,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996030341862686e-05,
+      "loss": 0.8257,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995967085627017e-05,
+      "loss": 1.2311,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995903329477044e-05,
+      "loss": 1.2283,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995839073415953e-05,
+      "loss": 1.1917,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999577431744696e-05,
+      "loss": 1.1662,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995709061573302e-05,
+      "loss": 0.7655,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995643305798243e-05,
+      "loss": 1.1705,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995577050125075e-05,
+      "loss": 1.3135,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999551029455711e-05,
+      "loss": 0.778,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999544303909768e-05,
+      "loss": 1.2401,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999537528375016e-05,
+      "loss": 1.1561,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995307028517926e-05,
+      "loss": 1.1819,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995238273404402e-05,
+      "loss": 1.237,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995169018413022e-05,
+      "loss": 1.1888,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999509926354725e-05,
+      "loss": 0.8051,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9995029008810573e-05,
+      "loss": 1.2271,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994958254206506e-05,
+      "loss": 1.131,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999488699973859e-05,
+      "loss": 1.1501,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994815245410383e-05,
+      "loss": 1.1741,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999474299122548e-05,
+      "loss": 1.1524,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999467023718749e-05,
+      "loss": 1.2161,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994596983300053e-05,
+      "loss": 1.2067,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994523229566833e-05,
+      "loss": 1.2249,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994448975991515e-05,
+      "loss": 1.2535,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999437422257782e-05,
+      "loss": 1.1487,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999429896932948e-05,
+      "loss": 1.1546,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9994223216250258e-05,
+      "loss": 1.0728,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999414696334395e-05,
+      "loss": 1.0884,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999407021061436e-05,
+      "loss": 1.2126,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993992958065333e-05,
+      "loss": 1.112,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999391520570073e-05,
+      "loss": 1.1727,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999383695352444e-05,
+      "loss": 1.1302,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999375820154038e-05,
+      "loss": 1.2501,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999367894975248e-05,
+      "loss": 1.1022,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993599198164715e-05,
+      "loss": 1.317,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993518946781062e-05,
+      "loss": 1.1695,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993438195605543e-05,
+      "loss": 1.1563,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993356944642192e-05,
+      "loss": 0.7909,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993275193895073e-05,
+      "loss": 1.242,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999319294336828e-05,
+      "loss": 1.1567,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993110193065914e-05,
+      "loss": 1.1976,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9993026942992125e-05,
+      "loss": 1.1813,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999294319315107e-05,
+      "loss": 1.1771,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992858943546946e-05,
+      "loss": 1.2723,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992774194183955e-05,
+      "loss": 1.2075,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992688945066345e-05,
+      "loss": 1.2171,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999260319619837e-05,
+      "loss": 1.2598,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999251694758433e-05,
+      "loss": 1.183,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999243019922853e-05,
+      "loss": 1.2249,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992342951135312e-05,
+      "loss": 1.1618,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992255203309034e-05,
+      "loss": 1.2327,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999216695575409e-05,
+      "loss": 1.0135,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9992078208474894e-05,
+      "loss": 1.2318,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991988961475882e-05,
+      "loss": 1.2092,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999189921476152e-05,
+      "loss": 1.1655,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991808968336287e-05,
+      "loss": 1.2029,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999171822220471e-05,
+      "loss": 1.1397,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999162697637132e-05,
+      "loss": 1.2082,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999153523084068e-05,
+      "loss": 1.2012,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991442985617378e-05,
+      "loss": 1.2292,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991350240706028e-05,
+      "loss": 1.2182,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9991256996111274e-05,
+      "loss": 1.0609,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999116325183777e-05,
+      "loss": 1.1529,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999106900789021e-05,
+      "loss": 1.2099,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990974264273308e-05,
+      "loss": 1.2399,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990879020991802e-05,
+      "loss": 1.105,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999078327805045e-05,
+      "loss": 1.1355,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990687035454044e-05,
+      "loss": 1.2289,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990590293207398e-05,
+      "loss": 1.18,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999049305131535e-05,
+      "loss": 1.2177,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990395309782758e-05,
+      "loss": 1.1924,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990297068614523e-05,
+      "loss": 1.1823,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990198327815545e-05,
+      "loss": 1.0619,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9990099087390765e-05,
+      "loss": 1.1831,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989999347345153e-05,
+      "loss": 1.1042,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989899107683692e-05,
+      "loss": 1.2272,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989798368411394e-05,
+      "loss": 1.1678,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99896971295333e-05,
+      "loss": 1.1419,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989595391054473e-05,
+      "loss": 1.0829,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989493152979995e-05,
+      "loss": 1.1932,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989390415314987e-05,
+      "loss": 1.2082,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989287178064587e-05,
+      "loss": 1.2609,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989183441233952e-05,
+      "loss": 1.1035,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9989079204828275e-05,
+      "loss": 1.1806,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988974468852767e-05,
+      "loss": 1.2324,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988869233312663e-05,
+      "loss": 1.0581,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988763498213236e-05,
+      "loss": 1.1828,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988657263559763e-05,
+      "loss": 1.3044,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998855052935756e-05,
+      "loss": 1.2343,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998844329561197e-05,
+      "loss": 1.1448,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998833556232835e-05,
+      "loss": 1.1234,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998822732951209e-05,
+      "loss": 0.7707,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988118597168604e-05,
+      "loss": 1.1784,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9988009365303328e-05,
+      "loss": 1.118,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987899633921727e-05,
+      "loss": 1.224,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987789403029286e-05,
+      "loss": 1.1539,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998767867263152e-05,
+      "loss": 1.1476,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987567442733965e-05,
+      "loss": 1.1522,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987455713342188e-05,
+      "loss": 1.1979,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987343484461773e-05,
+      "loss": 1.0825,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998723075609833e-05,
+      "loss": 1.0644,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9987117528257502e-05,
+      "loss": 1.2965,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998700380094495e-05,
+      "loss": 1.1808,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986889574166365e-05,
+      "loss": 1.2332,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998677484792745e-05,
+      "loss": 1.1573,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998665962223395e-05,
+      "loss": 1.1108,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986543897091632e-05,
+      "loss": 1.2229,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986427672506273e-05,
+      "loss": 1.1086,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9986310948483692e-05,
+      "loss": 1.1518,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998619372502973e-05,
+      "loss": 1.0901,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998607600215024e-05,
+      "loss": 1.1192,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985957779851116e-05,
+      "loss": 1.1431,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985839058138266e-05,
+      "loss": 1.1329,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985719837017632e-05,
+      "loss": 1.1989,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985600116495173e-05,
+      "loss": 1.229,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985479896576883e-05,
+      "loss": 0.7863,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985359177268764e-05,
+      "loss": 1.1998,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998523795857686e-05,
+      "loss": 1.1963,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9985116240507233e-05,
+      "loss": 1.2091,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9984994023065974e-05,
+      "loss": 1.0945,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9984871306259185e-05,
+      "loss": 1.1833,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998474809009301e-05,
+      "loss": 1.0735,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998462437457361e-05,
+      "loss": 1.1303,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9984500159707173e-05,
+      "loss": 1.1663,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9984375445499912e-05,
+      "loss": 1.0461,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9984250231958066e-05,
+      "loss": 1.0906,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998412451908789e-05,
+      "loss": 1.0832,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998399830689567e-05,
+      "loss": 0.779,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998387159538773e-05,
+      "loss": 1.1625,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983744384570397e-05,
+      "loss": 1.208,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998361667445004e-05,
+      "loss": 1.1939,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998348846503304e-05,
+      "loss": 1.0826,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983359756325812e-05,
+      "loss": 1.0778,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983230548334792e-05,
+      "loss": 1.1259,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9983100841066442e-05,
+      "loss": 1.2406,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982970634527244e-05,
+      "loss": 1.1387,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982839928723723e-05,
+      "loss": 1.1485,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99827087236624e-05,
+      "loss": 1.0612,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998257701934985e-05,
+      "loss": 1.1207,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998244481579265e-05,
+      "loss": 1.1968,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998231211299742e-05,
+      "loss": 1.1907,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998217891097079e-05,
+      "loss": 1.0919,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9982045209719424e-05,
+      "loss": 1.1287,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998191100925001e-05,
+      "loss": 1.1692,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981776309569255e-05,
+      "loss": 1.136,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99816411106839e-05,
+      "loss": 1.1911,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981505412600705e-05,
+      "loss": 1.0973,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981369215326458e-05,
+      "loss": 1.1964,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998123251886797e-05,
+      "loss": 1.0754,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9981095323232076e-05,
+      "loss": 1.1407,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980957628425636e-05,
+      "loss": 1.1263,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980819434455544e-05,
+      "loss": 1.1363,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99806807413287e-05,
+      "loss": 1.0979,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980541549052047e-05,
+      "loss": 1.1859,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980401857632547e-05,
+      "loss": 1.1453,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998026166707718e-05,
+      "loss": 1.0863,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980120977392968e-05,
+      "loss": 1.1584,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979979788586936e-05,
+      "loss": 1.1581,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997983810066615e-05,
+      "loss": 1.1179,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979695913637697e-05,
+      "loss": 1.0666,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979553227508688e-05,
+      "loss": 1.1661,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979410042286254e-05,
+      "loss": 1.201,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979266357977564e-05,
+      "loss": 1.0857,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9979122174589798e-05,
+      "loss": 1.1486,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978977492130168e-05,
+      "loss": 1.1772,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978832310605908e-05,
+      "loss": 1.1013,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978686630024283e-05,
+      "loss": 1.2047,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978540450392576e-05,
+      "loss": 1.1808,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978393771718097e-05,
+      "loss": 1.0353,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978246594008187e-05,
+      "loss": 1.2056,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9978098917270202e-05,
+      "loss": 1.1331,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977950741511525e-05,
+      "loss": 1.2986,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977802066739568e-05,
+      "loss": 1.1059,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977652892961773e-05,
+      "loss": 1.1046,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977503220185593e-05,
+      "loss": 1.1062,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977353048418515e-05,
+      "loss": 1.1519,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977202377668052e-05,
+      "loss": 1.1755,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9977051207941738e-05,
+      "loss": 1.1371,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997689953924713e-05,
+      "loss": 1.154,
+      "step": 446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997674737159182e-05,
+      "loss": 1.1664,
+      "step": 447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997659470498341e-05,
+      "loss": 0.9559,
+      "step": 448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976441539429537e-05,
+      "loss": 1.1578,
+      "step": 449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997628787493787e-05,
+      "loss": 1.0441,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9976133711516082e-05,
+      "loss": 1.2232,
+      "step": 451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997597904917189e-05,
+      "loss": 1.2142,
+      "step": 452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997582388791303e-05,
+      "loss": 1.146,
+      "step": 453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975668227747253e-05,
+      "loss": 1.2556,
+      "step": 454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9975512068682356e-05,
+      "loss": 1.2208,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997535541072614e-05,
+      "loss": 1.1161,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997519825388644e-05,
+      "loss": 1.1176,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997504059817112e-05,
+      "loss": 1.0175,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997488244358806e-05,
+      "loss": 1.2078,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974723790145174e-05,
+      "loss": 1.1821,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974564637850397e-05,
+      "loss": 1.0804,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974404986711683e-05,
+      "loss": 1.1679,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997424483673702e-05,
+      "loss": 1.1729,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9974084187934417e-05,
+      "loss": 1.0888,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973923040311906e-05,
+      "loss": 1.1427,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973761393877553e-05,
+      "loss": 1.1102,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973599248639436e-05,
+      "loss": 1.2065,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973436604605663e-05,
+      "loss": 1.1833,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973273461784372e-05,
+      "loss": 1.1779,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973109820183722e-05,
+      "loss": 1.0461,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972945679811896e-05,
+      "loss": 1.0924,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99727810406771e-05,
+      "loss": 0.7405,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972615902787573e-05,
+      "loss": 1.0835,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9972450266151574e-05,
+      "loss": 0.7467,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997228413077738e-05,
+      "loss": 1.1446,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99721174966733e-05,
+      "loss": 1.0502,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997195036384768e-05,
+      "loss": 1.032,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971782732308865e-05,
+      "loss": 1.1429,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971614602065248e-05,
+      "loss": 1.1919,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971445973125226e-05,
+      "loss": 1.0525,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971276845497243e-05,
+      "loss": 1.1368,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9971107219189757e-05,
+      "loss": 1.0296,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970937094211246e-05,
+      "loss": 0.7734,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997076647057022e-05,
+      "loss": 1.1024,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970595348275212e-05,
+      "loss": 1.0911,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970423727334782e-05,
+      "loss": 0.7413,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997025160775751e-05,
+      "loss": 1.1676,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9970078989552007e-05,
+      "loss": 1.1815,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969905872726903e-05,
+      "loss": 1.2523,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969732257290856e-05,
+      "loss": 1.1515,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996955814325255e-05,
+      "loss": 1.1393,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969383530620696e-05,
+      "loss": 1.0968,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969208419404024e-05,
+      "loss": 1.1529,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969032809611288e-05,
+      "loss": 1.2585,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968856701251276e-05,
+      "loss": 1.058,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996868009433279e-05,
+      "loss": 1.1319,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996850298886467e-05,
+      "loss": 1.0896,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9968325384855767e-05,
+      "loss": 1.1438,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9968147282314965e-05,
+      "loss": 1.1421,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996796868125117e-05,
+      "loss": 0.7228,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9967789581673316e-05,
+      "loss": 1.1685,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9967609983590362e-05,
+      "loss": 1.1748,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9967429887011283e-05,
+      "loss": 1.1155,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9967249291945092e-05,
+      "loss": 1.1732,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996706819840082e-05,
+      "loss": 1.1697,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996688660638752e-05,
+      "loss": 1.1679,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996670451591428e-05,
+      "loss": 1.1046,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9966521926990205e-05,
+      "loss": 1.0392,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996633883962442e-05,
+      "loss": 1.0905,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996615525382609e-05,
+      "loss": 1.1427,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996597116960439e-05,
+      "loss": 1.2072,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9965786586968533e-05,
+      "loss": 1.2312,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9965601505927745e-05,
+      "loss": 1.1312,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9965415926491283e-05,
+      "loss": 1.1581,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9965229848668428e-05,
+      "loss": 1.1384,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9965043272468487e-05,
+      "loss": 1.063,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9964856197900794e-05,
+      "loss": 1.1607,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9964668624974696e-05,
+      "loss": 1.1083,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996448055369958e-05,
+      "loss": 1.0654,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9964291984084857e-05,
+      "loss": 1.0446,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9964102916139946e-05,
+      "loss": 1.1493,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9963913349874314e-05,
+      "loss": 1.1034,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996372328529743e-05,
+      "loss": 1.1119,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996353272241881e-05,
+      "loss": 0.7961,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9963341661247978e-05,
+      "loss": 0.7341,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996315010179449e-05,
+      "loss": 0.7347,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996295804406793e-05,
+      "loss": 1.1514,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9962765488077896e-05,
+      "loss": 1.1707,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9962572433834027e-05,
+      "loss": 1.1565,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996237888134597e-05,
+      "loss": 1.1105,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996218483062341e-05,
+      "loss": 1.2839,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9961990281676052e-05,
+      "loss": 1.2289,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996179523451362e-05,
+      "loss": 1.1231,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9961599689145873e-05,
+      "loss": 0.7941,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996140364558259e-05,
+      "loss": 0.7352,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9961207103833575e-05,
+      "loss": 1.0565,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996101006390866e-05,
+      "loss": 1.1675,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9960812525817698e-05,
+      "loss": 1.0273,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.996061448957056e-05,
+      "loss": 1.2393,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9960415955177164e-05,
+      "loss": 1.1363,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9960216922647432e-05,
+      "loss": 1.1492,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9960017391991315e-05,
+      "loss": 1.1272,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959817363218795e-05,
+      "loss": 1.1551,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995961683633988e-05,
+      "loss": 1.0148,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995941581136459e-05,
+      "loss": 1.1028,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959214288302987e-05,
+      "loss": 1.0457,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9959012267165145e-05,
+      "loss": 1.0519,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958809747961166e-05,
+      "loss": 1.1384,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995860673070118e-05,
+      "loss": 1.1724,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958403215395342e-05,
+      "loss": 1.1963,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9958199202053826e-05,
+      "loss": 1.1295,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957994690686845e-05,
+      "loss": 1.0633,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957789681304612e-05,
+      "loss": 1.1639,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957584173917392e-05,
+      "loss": 1.1263,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957378168535456e-05,
+      "loss": 1.1314,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9957171665169105e-05,
+      "loss": 1.1414,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956964663828676e-05,
+      "loss": 0.7559,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956757164524516e-05,
+      "loss": 1.1013,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956549167267e-05,
+      "loss": 1.073,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9956340672066532e-05,
+      "loss": 1.061,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995613167893354e-05,
+      "loss": 1.1356,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955922187878477e-05,
+      "loss": 0.7697,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995571219891182e-05,
+      "loss": 1.1778,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9955501712044066e-05,
+      "loss": 1.2485,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995529072728575e-05,
+      "loss": 1.0988,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995507924464742e-05,
+      "loss": 1.0736,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954867264139643e-05,
+      "loss": 1.1057,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9954654785773036e-05,
+      "loss": 1.0979,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995444180955822e-05,
+      "loss": 0.79,
+      "step": 569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995422833550584e-05,
+      "loss": 1.1344,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995401436362658e-05,
+      "loss": 0.7542,
+      "step": 571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953799893931135e-05,
+      "loss": 0.7309,
+      "step": 572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995358492643024e-05,
+      "loss": 1.1895,
+      "step": 573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9953369461134636e-05,
+      "loss": 1.1067,
+      "step": 574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99531534980551e-05,
+      "loss": 1.141,
+      "step": 575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952937037202444e-05,
+      "loss": 1.0866,
+      "step": 576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952720078587474e-05,
+      "loss": 1.1545,
+      "step": 577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995250262222106e-05,
+      "loss": 1.2453,
+      "step": 578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952284668114064e-05,
+      "loss": 1.1935,
+      "step": 579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9952066216277393e-05,
+      "loss": 1.0734,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.995184726672197e-05,
+      "loss": 1.1528,
+      "step": 581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951627819458746e-05,
+      "loss": 1.1352,
+      "step": 582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951407874498695e-05,
+      "loss": 1.1303,
+      "step": 583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9951187431852813e-05,
+      "loss": 1.1043,
+      "step": 584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950966491532135e-05,
+      "loss": 1.1459,
+      "step": 585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950745053547697e-05,
+      "loss": 1.0784,
+      "step": 586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950523117910586e-05,
+      "loss": 1.0675,
+      "step": 587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950300684631894e-05,
+      "loss": 1.1383,
+      "step": 588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9950077753722745e-05,
+      "loss": 1.1374,
+      "step": 589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949854325194293e-05,
+      "loss": 1.0821,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949630399057708e-05,
+      "loss": 1.1217,
+      "step": 591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994940597532419e-05,
+      "loss": 0.9554,
+      "step": 592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9949181054004967e-05,
+      "loss": 1.1024,
+      "step": 593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994895563511128e-05,
+      "loss": 1.1505,
+      "step": 594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948729718654408e-05,
+      "loss": 1.1125,
+      "step": 595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948503304645644e-05,
+      "loss": 1.1168,
+      "step": 596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948276393096316e-05,
+      "loss": 0.9964,
+      "step": 597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9948048984017773e-05,
+      "loss": 1.0673,
+      "step": 598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947821077421386e-05,
+      "loss": 1.1956,
+      "step": 599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947592673318555e-05,
+      "loss": 1.1257,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9947363771720697e-05,
+      "loss": 1.1005,
+      "step": 601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994713437263927e-05,
+      "loss": 0.7447,
+      "step": 602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946904476085735e-05,
+      "loss": 1.0868,
+      "step": 603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946674082071596e-05,
+      "loss": 0.7351,
+      "step": 604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9946443190608377e-05,
+      "loss": 1.0672,
+      "step": 605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994621180170762e-05,
+      "loss": 1.1429,
+      "step": 606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99459799153809e-05,
+      "loss": 1.1343,
+      "step": 607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945747531639816e-05,
+      "loss": 1.0723,
+      "step": 608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945514650495986e-05,
+      "loss": 1.0864,
+      "step": 609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945281271961058e-05,
+      "loss": 1.2095,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9945047396046702e-05,
+      "loss": 1.0832,
+      "step": 611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994481302276462e-05,
+      "loss": 1.1123,
+      "step": 612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944578152126526e-05,
+      "loss": 1.0957,
+      "step": 613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9944342784144176e-05,
+      "loss": 1.1339,
+      "step": 614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994410691882933e-05,
+      "loss": 1.1452,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943870556193792e-05,
+      "loss": 1.0991,
+      "step": 616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994363369624938e-05,
+      "loss": 1.0901,
+      "step": 617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943396339007937e-05,
+      "loss": 1.1795,
+      "step": 618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9943158484481342e-05,
+      "loss": 0.9853,
+      "step": 619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942920132681484e-05,
+      "loss": 1.192,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994268128362028e-05,
+      "loss": 1.2251,
+      "step": 621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942441937309684e-05,
+      "loss": 1.0986,
+      "step": 622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9942202093761657e-05,
+      "loss": 0.724,
+      "step": 623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941961752988204e-05,
+      "loss": 1.1408,
+      "step": 624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941720915001338e-05,
+      "loss": 0.7681,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9941479579813102e-05,
+      "loss": 0.7963,
+      "step": 626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994123774743557e-05,
+      "loss": 1.2082,
+      "step": 627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994099541788084e-05,
+      "loss": 1.1668,
+      "step": 628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.994075259116102e-05,
+      "loss": 1.1698,
+      "step": 629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940509267288265e-05,
+      "loss": 1.1217,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940265446274735e-05,
+      "loss": 1.0592,
+      "step": 631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9940021128132632e-05,
+      "loss": 1.0742,
+      "step": 632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993977631287417e-05,
+      "loss": 1.161,
+      "step": 633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939531000511594e-05,
+      "loss": 1.17,
+      "step": 634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993928519105717e-05,
+      "loss": 1.1824,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99390388845232e-05,
+      "loss": 1.0237,
+      "step": 636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938792080921988e-05,
+      "loss": 1.1147,
+      "step": 637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993854478026589e-05,
+      "loss": 1.1421,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9938296982567266e-05,
+      "loss": 0.7706,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9938048687838515e-05,
+      "loss": 1.1447,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993779989609205e-05,
+      "loss": 1.0999,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9937550607340313e-05,
+      "loss": 1.0195,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993730082159577e-05,
+      "loss": 1.1124,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9937050538870922e-05,
+      "loss": 1.0747,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993679975917828e-05,
+      "loss": 1.2224,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9936548482530383e-05,
+      "loss": 1.0089,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99362967089398e-05,
+      "loss": 1.0602,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9936044438419128e-05,
+      "loss": 1.0571,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9935791670980973e-05,
+      "loss": 1.0377,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9935538406637987e-05,
+      "loss": 1.0966,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9935284645402827e-05,
+      "loss": 1.0885,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9935030387288194e-05,
+      "loss": 1.0149,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9934775632306794e-05,
+      "loss": 1.074,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9934520380471372e-05,
+      "loss": 1.0567,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9934264631794694e-05,
+      "loss": 1.1739,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993400838628955e-05,
+      "loss": 1.0208,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9933751643968756e-05,
+      "loss": 1.0811,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9933494404845152e-05,
+      "loss": 1.1361,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9933236668931598e-05,
+      "loss": 1.1486,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993297843624099e-05,
+      "loss": 1.1494,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9932719706786238e-05,
+      "loss": 1.173,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993246048058029e-05,
+      "loss": 1.1208,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99322007576361e-05,
+      "loss": 1.1408,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993194053796666e-05,
+      "loss": 1.1271,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993167982158499e-05,
+      "loss": 1.182,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9931418608504122e-05,
+      "loss": 1.1596,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9931156898737122e-05,
+      "loss": 1.1815,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993089469229708e-05,
+      "loss": 0.7168,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9930631989197104e-05,
+      "loss": 1.2165,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.993036878945034e-05,
+      "loss": 1.0922,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9930105093069945e-05,
+      "loss": 1.1713,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9929840900069112e-05,
+      "loss": 1.1787,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9929576210461047e-05,
+      "loss": 1.0304,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9929311024258996e-05,
+      "loss": 1.0915,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992904534147621e-05,
+      "loss": 1.131,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992877916212599e-05,
+      "loss": 1.0798,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9928512486221635e-05,
+      "loss": 1.0773,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9928245313776487e-05,
+      "loss": 0.7303,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9927977644803916e-05,
+      "loss": 1.0783,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9927709479317293e-05,
+      "loss": 1.1487,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992744081733004e-05,
+      "loss": 0.732,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9927171658855595e-05,
+      "loss": 1.1141,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992690200390741e-05,
+      "loss": 1.0786,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926631852498978e-05,
+      "loss": 1.0465,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926361204643807e-05,
+      "loss": 1.1396,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9926090060355432e-05,
+      "loss": 1.1655,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925818419647414e-05,
+      "loss": 1.07,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925546282533342e-05,
+      "loss": 1.0847,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992527364902682e-05,
+      "loss": 1.1948,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9925000519141486e-05,
+      "loss": 1.1573,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924726892890996e-05,
+      "loss": 0.7263,
+      "step": 691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992445277028904e-05,
+      "loss": 1.0946,
+      "step": 692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9924178151349322e-05,
+      "loss": 1.0936,
+      "step": 693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923903036085578e-05,
+      "loss": 1.1589,
+      "step": 694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923627424511572e-05,
+      "loss": 1.167,
+      "step": 695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992335131664108e-05,
+      "loss": 1.1339,
+      "step": 696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9923074712487918e-05,
+      "loss": 1.1002,
+      "step": 697
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922797612065913e-05,
+      "loss": 1.1772,
+      "step": 698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922520015388928e-05,
+      "loss": 1.0432,
+      "step": 699
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9922241922470843e-05,
+      "loss": 1.0959,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921963333325566e-05,
+      "loss": 1.084,
+      "step": 701
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921684247967027e-05,
+      "loss": 1.2151,
+      "step": 702
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921404666409195e-05,
+      "loss": 1.147,
+      "step": 703
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9921124588666035e-05,
+      "loss": 1.0686,
+      "step": 704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.992084401475157e-05,
+      "loss": 1.0887,
+      "step": 705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920562944679823e-05,
+      "loss": 1.1431,
+      "step": 706
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9920281378464856e-05,
+      "loss": 0.7778,
+      "step": 707
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9919999316120742e-05,
+      "loss": 0.9524,
+      "step": 708
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99197167576616e-05,
+      "loss": 1.1428,
+      "step": 709
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991943370310155e-05,
+      "loss": 1.1669,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991915015245475e-05,
+      "loss": 1.183,
+      "step": 711
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918866105735387e-05,
+      "loss": 1.0261,
+      "step": 712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918581562957664e-05,
+      "loss": 1.0961,
+      "step": 713
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918296524135807e-05,
+      "loss": 1.1635,
+      "step": 714
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9918010989284076e-05,
+      "loss": 1.2178,
+      "step": 715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917724958416746e-05,
+      "loss": 0.7215,
+      "step": 716
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917438431548132e-05,
+      "loss": 1.072,
+      "step": 717
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9917151408692552e-05,
+      "loss": 1.1514,
+      "step": 718
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916863889864365e-05,
+      "loss": 1.2158,
+      "step": 719
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9916575875077954e-05,
+      "loss": 1.1313,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991628736434772e-05,
+      "loss": 1.1321,
+      "step": 721
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991599835768809e-05,
+      "loss": 1.1221,
+      "step": 722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915708855113518e-05,
+      "loss": 1.1433,
+      "step": 723
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9915418856638486e-05,
+      "loss": 1.1334,
+      "step": 724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991512836227749e-05,
+      "loss": 1.1375,
+      "step": 725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991483737204507e-05,
+      "loss": 1.0185,
+      "step": 726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9914545885955768e-05,
+      "loss": 1.1196,
+      "step": 727
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991425390402417e-05,
+      "loss": 1.0616,
+      "step": 728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991396142626487e-05,
+      "loss": 1.1366,
+      "step": 729
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913668452692504e-05,
+      "loss": 1.1332,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9913374983321718e-05,
+      "loss": 1.1515,
+      "step": 731
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991308101816719e-05,
+      "loss": 1.0919,
+      "step": 732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991278655724362e-05,
+      "loss": 1.0881,
+      "step": 733
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991249160056574e-05,
+      "loss": 1.1025,
+      "step": 734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9912196148148297e-05,
+      "loss": 1.1967,
+      "step": 735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911900200006066e-05,
+      "loss": 1.1865,
+      "step": 736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911603756153853e-05,
+      "loss": 1.1519,
+      "step": 737
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991130681660648e-05,
+      "loss": 0.7056,
+      "step": 738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9911009381378798e-05,
+      "loss": 1.1375,
+      "step": 739
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991071145048568e-05,
+      "loss": 1.1044,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9910413023942027e-05,
+      "loss": 0.9772,
+      "step": 741
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.991011410176277e-05,
+      "loss": 1.106,
+      "step": 742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990981468396285e-05,
+      "loss": 1.0578,
+      "step": 743
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909514770557244e-05,
+      "loss": 1.157,
+      "step": 744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9909214361560954e-05,
+      "loss": 1.0813,
+      "step": 745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908913456989004e-05,
+      "loss": 1.0591,
+      "step": 746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9908612056856437e-05,
+      "loss": 1.1054,
+      "step": 747
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990831016117833e-05,
+      "loss": 1.0812,
+      "step": 748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990800776996978e-05,
+      "loss": 1.1224,
+      "step": 749
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907704883245917e-05,
+      "loss": 1.0565,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990740150102188e-05,
+      "loss": 1.0256,
+      "step": 751
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9907097623312843e-05,
+      "loss": 1.0015,
+      "step": 752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906793250134008e-05,
+      "loss": 0.7171,
+      "step": 753
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906488381500594e-05,
+      "loss": 1.021,
+      "step": 754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9906183017427848e-05,
+      "loss": 1.1589,
+      "step": 755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905877157931044e-05,
+      "loss": 1.1728,
+      "step": 756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905570803025475e-05,
+      "loss": 1.1279,
+      "step": 757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9905263952726463e-05,
+      "loss": 1.028,
+      "step": 758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990495660704936e-05,
+      "loss": 1.1401,
+      "step": 759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904648766009527e-05,
+      "loss": 1.055,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904340429622366e-05,
+      "loss": 1.0418,
+      "step": 761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9904031597903298e-05,
+      "loss": 1.0639,
+      "step": 762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903722270867764e-05,
+      "loss": 1.1164,
+      "step": 763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9903412448531236e-05,
+      "loss": 1.0882,
+      "step": 764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990310213090921e-05,
+      "loss": 1.0757,
+      "step": 765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902791318017206e-05,
+      "loss": 1.0634,
+      "step": 766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902480009870766e-05,
+      "loss": 1.0391,
+      "step": 767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9902168206485456e-05,
+      "loss": 1.1374,
+      "step": 768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901855907876876e-05,
+      "loss": 1.1643,
+      "step": 769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990154311406064e-05,
+      "loss": 1.0636,
+      "step": 770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9901229825052396e-05,
+      "loss": 1.0154,
+      "step": 771
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900916040867807e-05,
+      "loss": 1.0375,
+      "step": 772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900601761522572e-05,
+      "loss": 1.0312,
+      "step": 773
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99002869870324e-05,
+      "loss": 1.0431,
+      "step": 774
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899971717413044e-05,
+      "loss": 1.0738,
+      "step": 775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899655952680262e-05,
+      "loss": 1.1363,
+      "step": 776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899339692849848e-05,
+      "loss": 1.1652,
+      "step": 777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989902293793762e-05,
+      "loss": 1.0536,
+      "step": 778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989870568795942e-05,
+      "loss": 1.0973,
+      "step": 779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898387942931118e-05,
+      "loss": 1.1404,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989806970286859e-05,
+      "loss": 1.1379,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989775096778777e-05,
+      "loss": 0.7346,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989743173770459e-05,
+      "loss": 0.992,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9897112012635013e-05,
+      "loss": 1.17,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9896791792595035e-05,
+      "loss": 1.1398,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9896471077600665e-05,
+      "loss": 0.991,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9896149867667945e-05,
+      "loss": 1.1765,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9895828162812943e-05,
+      "loss": 1.0998,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989550596305174e-05,
+      "loss": 1.0858,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9895183268400454e-05,
+      "loss": 1.1602,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9894860078875226e-05,
+      "loss": 1.1562,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9894536394492215e-05,
+      "loss": 0.7363,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989421221526761e-05,
+      "loss": 0.9999,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9893887541217625e-05,
+      "loss": 1.1311,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9893562372358494e-05,
+      "loss": 1.0825,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9893236708706483e-05,
+      "loss": 1.2136,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989291055027788e-05,
+      "loss": 1.2256,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9892583897088996e-05,
+      "loss": 1.152,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989225674915616e-05,
+      "loss": 1.0575,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9891929106495746e-05,
+      "loss": 1.0414,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989160096912413e-05,
+      "loss": 1.0529,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9891272337057724e-05,
+      "loss": 1.0717,
+      "step": 802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989094321031297e-05,
+      "loss": 0.749,
+      "step": 803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989061358890632e-05,
+      "loss": 1.0306,
+      "step": 804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.989028347285426e-05,
+      "loss": 1.0912,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9889952862173305e-05,
+      "loss": 1.0917,
+      "step": 806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9889621756879986e-05,
+      "loss": 1.0629,
+      "step": 807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9889290156990864e-05,
+      "loss": 1.0774,
+      "step": 808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9888958062522518e-05,
+      "loss": 1.1429,
+      "step": 809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9888625473491563e-05,
+      "loss": 1.1595,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9888292389914628e-05,
+      "loss": 1.0745,
+      "step": 811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988795881180837e-05,
+      "loss": 1.114,
+      "step": 812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9887624739189476e-05,
+      "loss": 1.0911,
+      "step": 813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9887290172074653e-05,
+      "loss": 1.1447,
+      "step": 814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988695511048063e-05,
+      "loss": 1.0879,
+      "step": 815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9886619554424164e-05,
+      "loss": 1.1939,
+      "step": 816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988628350392204e-05,
+      "loss": 1.1407,
+      "step": 817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9885946958991063e-05,
+      "loss": 1.0488,
+      "step": 818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9885609919648065e-05,
+      "loss": 1.1079,
+      "step": 819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9885272385909896e-05,
+      "loss": 1.0855,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988493435779345e-05,
+      "loss": 1.0925,
+      "step": 821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9884595835315618e-05,
+      "loss": 1.1708,
+      "step": 822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9884256818493336e-05,
+      "loss": 1.0517,
+      "step": 823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883917307343558e-05,
+      "loss": 1.0624,
+      "step": 824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9883577301883265e-05,
+      "loss": 1.0982,
+      "step": 825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988323680212946e-05,
+      "loss": 0.9789,
+      "step": 826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988289580809917e-05,
+      "loss": 1.1046,
+      "step": 827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9882554319809455e-05,
+      "loss": 1.1139,
+      "step": 828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988221233727739e-05,
+      "loss": 1.1093,
+      "step": 829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881869860520072e-05,
+      "loss": 1.0708,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988152688955464e-05,
+      "loss": 1.11,
+      "step": 831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9881183424398237e-05,
+      "loss": 1.1497,
+      "step": 832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880839465068042e-05,
+      "loss": 0.7388,
+      "step": 833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9880495011581265e-05,
+      "loss": 0.9879,
+      "step": 834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.988015006395512e-05,
+      "loss": 1.0549,
+      "step": 835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879804622206872e-05,
+      "loss": 1.1373,
+      "step": 836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879458686353782e-05,
+      "loss": 1.0503,
+      "step": 837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9879112256413164e-05,
+      "loss": 1.0714,
+      "step": 838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987876533240234e-05,
+      "loss": 1.0999,
+      "step": 839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878417914338652e-05,
+      "loss": 1.1187,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9878070002239487e-05,
+      "loss": 1.1355,
+      "step": 841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877721596122237e-05,
+      "loss": 1.0999,
+      "step": 842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987737269600433e-05,
+      "loss": 1.0612,
+      "step": 843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9877023301903213e-05,
+      "loss": 1.0623,
+      "step": 844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987667341383636e-05,
+      "loss": 1.1117,
+      "step": 845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9876323031821267e-05,
+      "loss": 0.6982,
+      "step": 846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875972155875463e-05,
+      "loss": 1.1137,
+      "step": 847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875620786016493e-05,
+      "loss": 1.0294,
+      "step": 848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875268922261926e-05,
+      "loss": 1.1141,
+      "step": 849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874916564629366e-05,
+      "loss": 1.1086,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987456371313643e-05,
+      "loss": 0.984,
+      "step": 851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9874210367800768e-05,
+      "loss": 1.0922,
+      "step": 852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987385652864005e-05,
+      "loss": 1.2066,
+      "step": 853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987350219567197e-05,
+      "loss": 1.0462,
+      "step": 854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987314736891425e-05,
+      "loss": 1.1908,
+      "step": 855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9872792048384638e-05,
+      "loss": 1.1739,
+      "step": 856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98724362341009e-05,
+      "loss": 1.1024,
+      "step": 857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9872079926080835e-05,
+      "loss": 1.0302,
+      "step": 858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987172312434226e-05,
+      "loss": 1.251,
+      "step": 859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9871365828903015e-05,
+      "loss": 1.0406,
+      "step": 860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987100803978098e-05,
+      "loss": 1.069,
+      "step": 861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.987064975699404e-05,
+      "loss": 1.0859,
+      "step": 862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9870290980560113e-05,
+      "loss": 0.9888,
+      "step": 863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869931710497144e-05,
+      "loss": 1.1448,
+      "step": 864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869571946823106e-05,
+      "loss": 0.7069,
+      "step": 865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9869211689555983e-05,
+      "loss": 1.1429,
+      "step": 866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9868850938713797e-05,
+      "loss": 0.9801,
+      "step": 867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986848969431459e-05,
+      "loss": 1.0796,
+      "step": 868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986812795637642e-05,
+      "loss": 1.0594,
+      "step": 869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867765724917394e-05,
+      "loss": 1.0408,
+      "step": 870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867402999955613e-05,
+      "loss": 1.0833,
+      "step": 871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9867039781509224e-05,
+      "loss": 0.7296,
+      "step": 872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986667606959639e-05,
+      "loss": 1.0288,
+      "step": 873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9866311864235302e-05,
+      "loss": 1.0884,
+      "step": 874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865947165444174e-05,
+      "loss": 1.0436,
+      "step": 875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986558197324125e-05,
+      "loss": 1.1847,
+      "step": 876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9865216287644787e-05,
+      "loss": 1.0932,
+      "step": 877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864850108673074e-05,
+      "loss": 1.0418,
+      "step": 878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864483436344425e-05,
+      "loss": 1.0253,
+      "step": 879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9864116270677183e-05,
+      "loss": 1.1927,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9863748611689707e-05,
+      "loss": 1.2694,
+      "step": 881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986338045940038e-05,
+      "loss": 0.9648,
+      "step": 882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986301181382762e-05,
+      "loss": 1.0403,
+      "step": 883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.986264267498986e-05,
+      "loss": 1.0607,
+      "step": 884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9862273042905564e-05,
+      "loss": 0.7278,
+      "step": 885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861902917593213e-05,
+      "loss": 1.0587,
+      "step": 886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861532299071325e-05,
+      "loss": 1.0477,
+      "step": 887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9861161187358432e-05,
+      "loss": 1.0916,
+      "step": 888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860789582473087e-05,
+      "loss": 1.1261,
+      "step": 889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860417484433884e-05,
+      "loss": 1.1155,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9860044893259428e-05,
+      "loss": 1.0756,
+      "step": 891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859671808968357e-05,
+      "loss": 1.076,
+      "step": 892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9859298231579322e-05,
+      "loss": 1.0152,
+      "step": 893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858924161111016e-05,
+      "loss": 1.0653,
+      "step": 894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9858549597582134e-05,
+      "loss": 1.0431,
+      "step": 895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985817454101142e-05,
+      "loss": 1.0591,
+      "step": 896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857798991417627e-05,
+      "loss": 1.0901,
+      "step": 897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857422948819535e-05,
+      "loss": 1.016,
+      "step": 898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9857046413235956e-05,
+      "loss": 1.1192,
+      "step": 899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9856669384685713e-05,
+      "loss": 1.1388,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985629186318767e-05,
+      "loss": 1.1415,
+      "step": 901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98559138487607e-05,
+      "loss": 0.7335,
+      "step": 902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9855535341423713e-05,
+      "loss": 1.0714,
+      "step": 903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985515634119564e-05,
+      "loss": 1.0317,
+      "step": 904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985477684809543e-05,
+      "loss": 1.1566,
+      "step": 905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854396862142067e-05,
+      "loss": 0.7199,
+      "step": 906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9854016383354553e-05,
+      "loss": 1.0762,
+      "step": 907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853635411751916e-05,
+      "loss": 1.1226,
+      "step": 908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9853253947353205e-05,
+      "loss": 1.0898,
+      "step": 909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852871990177505e-05,
+      "loss": 1.0951,
+      "step": 910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852489540243916e-05,
+      "loss": 1.129,
+      "step": 911
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9852106597571562e-05,
+      "loss": 1.0888,
+      "step": 912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98517231621796e-05,
+      "loss": 1.1145,
+      "step": 913
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98513392340872e-05,
+      "loss": 0.9704,
+      "step": 914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850954813313566e-05,
+      "loss": 1.0597,
+      "step": 915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985056989987792e-05,
+      "loss": 1.0572,
+      "step": 916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850184493799518e-05,
+      "loss": 0.7368,
+      "step": 917
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849798595097633e-05,
+      "loss": 1.1142,
+      "step": 918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849412203791566e-05,
+      "loss": 1.0612,
+      "step": 919
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984902531990063e-05,
+      "loss": 1.0436,
+      "step": 920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848637943444192e-05,
+      "loss": 1.0172,
+      "step": 921
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984825007444161e-05,
+      "loss": 1.077,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9847861712912287e-05,
+      "loss": 1.1281,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9847472858875646e-05,
+      "loss": 1.0323,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9847083512351136e-05,
+      "loss": 0.9928,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9846693673358225e-05,
+      "loss": 1.0763,
+      "step": 926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9846303341916414e-05,
+      "loss": 1.1068,
+      "step": 927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9845912518045224e-05,
+      "loss": 1.014,
+      "step": 928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9845521201764192e-05,
+      "loss": 1.0587,
+      "step": 929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98451293930929e-05,
+      "loss": 1.1044,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9844737092050934e-05,
+      "loss": 1.0597,
+      "step": 931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9844344298657918e-05,
+      "loss": 1.0609,
+      "step": 932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9843951012933496e-05,
+      "loss": 1.1747,
+      "step": 933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.984355723489734e-05,
+      "loss": 1.0528,
+      "step": 934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9843162964569136e-05,
+      "loss": 1.0218,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9842768201968604e-05,
+      "loss": 1.0939,
+      "step": 936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9842372947115492e-05,
+      "loss": 1.0865,
+      "step": 937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9841977200029563e-05,
+      "loss": 1.0145,
+      "step": 938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9841580960730612e-05,
+      "loss": 1.0743,
+      "step": 939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.984118422923845e-05,
+      "loss": 1.0708,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.984078700557292e-05,
+      "loss": 0.7082,
+      "step": 941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9840389289753895e-05,
+      "loss": 1.0908,
+      "step": 942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9839991081801257e-05,
+      "loss": 1.0659,
+      "step": 943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9839592381734924e-05,
+      "loss": 1.1034,
+      "step": 944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9839193189574836e-05,
+      "loss": 1.1184,
+      "step": 945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9838793505340955e-05,
+      "loss": 1.074,
+      "step": 946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9838393329053276e-05,
+      "loss": 1.1251,
+      "step": 947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9837992660731805e-05,
+      "loss": 1.1286,
+      "step": 948
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9837591500396582e-05,
+      "loss": 1.038,
+      "step": 949
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983718984806767e-05,
+      "loss": 0.9476,
+      "step": 950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983678770376516e-05,
+      "loss": 1.1284,
+      "step": 951
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983638506750916e-05,
+      "loss": 1.103,
+      "step": 952
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983598193931981e-05,
+      "loss": 0.9993,
+      "step": 953
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9835578319217265e-05,
+      "loss": 1.0226,
+      "step": 954
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9835174207221713e-05,
+      "loss": 1.0201,
+      "step": 955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9834769603353366e-05,
+      "loss": 1.0764,
+      "step": 956
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9834364507632462e-05,
+      "loss": 1.1186,
+      "step": 957
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9833958920079255e-05,
+      "loss": 1.1128,
+      "step": 958
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983355284071403e-05,
+      "loss": 1.0606,
+      "step": 959
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9833146269557096e-05,
+      "loss": 1.0739,
+      "step": 960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983273920662879e-05,
+      "loss": 0.7235,
+      "step": 961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9832331651949467e-05,
+      "loss": 1.093,
+      "step": 962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9831923605539506e-05,
+      "loss": 1.0754,
+      "step": 963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9831515067419316e-05,
+      "loss": 1.0127,
+      "step": 964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9831106037609334e-05,
+      "loss": 1.0284,
+      "step": 965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983069651613001e-05,
+      "loss": 1.0041,
+      "step": 966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.983028650300183e-05,
+      "loss": 1.0926,
+      "step": 967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982987599824529e-05,
+      "loss": 1.1758,
+      "step": 968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982946500188093e-05,
+      "loss": 1.0923,
+      "step": 969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98290535139293e-05,
+      "loss": 1.0258,
+      "step": 970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9828641534410983e-05,
+      "loss": 1.0744,
+      "step": 971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982822906334658e-05,
+      "loss": 1.095,
+      "step": 972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9827816100756718e-05,
+      "loss": 1.183,
+      "step": 973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982740264666205e-05,
+      "loss": 1.0659,
+      "step": 974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982698870108325e-05,
+      "loss": 0.7645,
+      "step": 975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9826574264041034e-05,
+      "loss": 1.027,
+      "step": 976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982615933555611e-05,
+      "loss": 1.104,
+      "step": 977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825743915649246e-05,
+      "loss": 1.0771,
+      "step": 978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9825328004341208e-05,
+      "loss": 1.097,
+      "step": 979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98249116016528e-05,
+      "loss": 0.7299,
+      "step": 980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9824494707604846e-05,
+      "loss": 1.0406,
+      "step": 981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982407732221819e-05,
+      "loss": 1.0504,
+      "step": 982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9823659445513715e-05,
+      "loss": 1.1239,
+      "step": 983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982324107751232e-05,
+      "loss": 1.0226,
+      "step": 984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982282221823492e-05,
+      "loss": 1.1778,
+      "step": 985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982240286770247e-05,
+      "loss": 1.0564,
+      "step": 986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821983025935937e-05,
+      "loss": 1.0865,
+      "step": 987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982156269295632e-05,
+      "loss": 1.084,
+      "step": 988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9821141868784643e-05,
+      "loss": 1.0468,
+      "step": 989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982072055344195e-05,
+      "loss": 1.0841,
+      "step": 990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.982029874694931e-05,
+      "loss": 1.1653,
+      "step": 991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981987644932782e-05,
+      "loss": 1.0835,
+      "step": 992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9819453660598604e-05,
+      "loss": 1.0791,
+      "step": 993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98190303807828e-05,
+      "loss": 1.0576,
+      "step": 994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9818606609901578e-05,
+      "loss": 1.0802,
+      "step": 995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981818234797613e-05,
+      "loss": 1.0934,
+      "step": 996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9817757595027683e-05,
+      "loss": 1.0124,
+      "step": 997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981733235107747e-05,
+      "loss": 1.0504,
+      "step": 998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981690661614676e-05,
+      "loss": 1.0347,
+      "step": 999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981648039025685e-05,
+      "loss": 1.1719,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9816053673429045e-05,
+      "loss": 1.1581,
+      "step": 1001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98156264656847e-05,
+      "loss": 1.0912,
+      "step": 1002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9815198767045167e-05,
+      "loss": 1.1101,
+      "step": 1003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814770577531844e-05,
+      "loss": 1.0873,
+      "step": 1004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9814341897166147e-05,
+      "loss": 1.1378,
+      "step": 1005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981391272596951e-05,
+      "loss": 1.0149,
+      "step": 1006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9813483063963398e-05,
+      "loss": 1.1743,
+      "step": 1007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9813052911169297e-05,
+      "loss": 0.7331,
+      "step": 1008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9812622267608724e-05,
+      "loss": 0.9804,
+      "step": 1009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9812191133303215e-05,
+      "loss": 1.0012,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811759508274328e-05,
+      "loss": 1.1079,
+      "step": 1011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9811327392543655e-05,
+      "loss": 1.1478,
+      "step": 1012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98108947861328e-05,
+      "loss": 1.1678,
+      "step": 1013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9810461689063406e-05,
+      "loss": 1.0074,
+      "step": 1014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.981002810135713e-05,
+      "loss": 1.1272,
+      "step": 1015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980959402303565e-05,
+      "loss": 1.0478,
+      "step": 1016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9809159454120687e-05,
+      "loss": 0.9462,
+      "step": 1017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9808724394633963e-05,
+      "loss": 1.1298,
+      "step": 1018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980828884459724e-05,
+      "loss": 1.0748,
+      "step": 1019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9807852804032306e-05,
+      "loss": 1.0331,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980741627296096e-05,
+      "loss": 1.0482,
+      "step": 1021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980697925140504e-05,
+      "loss": 1.1713,
+      "step": 1022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98065417393864e-05,
+      "loss": 0.9549,
+      "step": 1023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9806103736926916e-05,
+      "loss": 1.1529,
+      "step": 1024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9805665244048498e-05,
+      "loss": 1.0955,
+      "step": 1025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9805226260773075e-05,
+      "loss": 1.0463,
+      "step": 1026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9804786787122604e-05,
+      "loss": 1.0052,
+      "step": 1027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9804346823119058e-05,
+      "loss": 1.1104,
+      "step": 1028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803906368784445e-05,
+      "loss": 1.051,
+      "step": 1029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980346542414079e-05,
+      "loss": 1.0378,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9803023989210148e-05,
+      "loss": 1.0268,
+      "step": 1031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9802582064014593e-05,
+      "loss": 1.1041,
+      "step": 1032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980213964857623e-05,
+      "loss": 1.0327,
+      "step": 1033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9801696742917178e-05,
+      "loss": 0.8856,
+      "step": 1034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980125334705959e-05,
+      "loss": 1.2238,
+      "step": 1035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.980080946102565e-05,
+      "loss": 1.127,
+      "step": 1036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9800365084837548e-05,
+      "loss": 1.0424,
+      "step": 1037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979992021851751e-05,
+      "loss": 1.0417,
+      "step": 1038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799474862087787e-05,
+      "loss": 1.1108,
+      "step": 1039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9799029015570648e-05,
+      "loss": 1.0186,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9798582678988388e-05,
+      "loss": 1.144,
+      "step": 1041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979813585236334e-05,
+      "loss": 1.0839,
+      "step": 1042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979768853571784e-05,
+      "loss": 1.1263,
+      "step": 1043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9797240729074265e-05,
+      "loss": 0.7326,
+      "step": 1044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9796792432455007e-05,
+      "loss": 1.1435,
+      "step": 1045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979634364588249e-05,
+      "loss": 1.0954,
+      "step": 1046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979589436937915e-05,
+      "loss": 0.989,
+      "step": 1047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979544460296747e-05,
+      "loss": 1.0571,
+      "step": 1048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794994346669927e-05,
+      "loss": 1.1274,
+      "step": 1049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794543600509056e-05,
+      "loss": 1.045,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9794092364507387e-05,
+      "loss": 1.1658,
+      "step": 1051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979364063868749e-05,
+      "loss": 1.0248,
+      "step": 1052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9793188423071957e-05,
+      "loss": 1.1982,
+      "step": 1053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979273571768341e-05,
+      "loss": 1.0812,
+      "step": 1054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792282522544482e-05,
+      "loss": 1.1539,
+      "step": 1055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791828837677837e-05,
+      "loss": 1.118,
+      "step": 1056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979137466310617e-05,
+      "loss": 1.0633,
+      "step": 1057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979091999885219e-05,
+      "loss": 1.0768,
+      "step": 1058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790464844938643e-05,
+      "loss": 1.0384,
+      "step": 1059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790009201388283e-05,
+      "loss": 1.061,
+      "step": 1060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978955306822391e-05,
+      "loss": 1.0923,
+      "step": 1061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789096445468317e-05,
+      "loss": 1.1276,
+      "step": 1062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788639333144356e-05,
+      "loss": 1.2195,
+      "step": 1063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788181731274883e-05,
+      "loss": 1.0823,
+      "step": 1064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978772363988278e-05,
+      "loss": 1.1079,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9787265058990966e-05,
+      "loss": 1.0358,
+      "step": 1066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9786805988622367e-05,
+      "loss": 1.1313,
+      "step": 1067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9786346428799943e-05,
+      "loss": 1.0554,
+      "step": 1068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9785886379546676e-05,
+      "loss": 1.0687,
+      "step": 1069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9785425840885582e-05,
+      "loss": 0.7816,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9784964812839687e-05,
+      "loss": 1.0115,
+      "step": 1071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978450329543205e-05,
+      "loss": 1.1736,
+      "step": 1072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9784041288685743e-05,
+      "loss": 1.0319,
+      "step": 1073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978357879262389e-05,
+      "loss": 1.118,
+      "step": 1074
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9783115807269604e-05,
+      "loss": 0.9914,
+      "step": 1075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978265233264605e-05,
+      "loss": 1.067,
+      "step": 1076
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97821883687764e-05,
+      "loss": 1.0677,
+      "step": 1077
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.978172391568386e-05,
+      "loss": 0.7663,
+      "step": 1078
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9781258973391664e-05,
+      "loss": 1.0765,
+      "step": 1079
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9780793541923056e-05,
+      "loss": 1.0399,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9780327621301318e-05,
+      "loss": 1.1208,
+      "step": 1081
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977986121154975e-05,
+      "loss": 1.0632,
+      "step": 1082
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9779394312691678e-05,
+      "loss": 1.0669,
+      "step": 1083
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9778926924750452e-05,
+      "loss": 1.065,
+      "step": 1084
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9778459047749448e-05,
+      "loss": 1.0287,
+      "step": 1085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9777990681712064e-05,
+      "loss": 0.7474,
+      "step": 1086
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977752182666172e-05,
+      "loss": 1.0408,
+      "step": 1087
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9777052482621873e-05,
+      "loss": 0.9836,
+      "step": 1088
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977658264961599e-05,
+      "loss": 1.0821,
+      "step": 1089
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977611232766757e-05,
+      "loss": 1.1804,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9775641516800127e-05,
+      "loss": 1.0044,
+      "step": 1091
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977517021703722e-05,
+      "loss": 1.1054,
+      "step": 1092
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977469842840241e-05,
+      "loss": 1.0439,
+      "step": 1093
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9774226150919298e-05,
+      "loss": 1.0566,
+      "step": 1094
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9773753384611497e-05,
+      "loss": 1.1396,
+      "step": 1095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9773280129502655e-05,
+      "loss": 1.0728,
+      "step": 1096
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977280638561644e-05,
+      "loss": 0.946,
+      "step": 1097
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9772332152976547e-05,
+      "loss": 1.1141,
+      "step": 1098
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9771857431606687e-05,
+      "loss": 0.9764,
+      "step": 1099
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97713822215306e-05,
+      "loss": 1.1,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9770906522772066e-05,
+      "loss": 1.0688,
+      "step": 1101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.977043033535486e-05,
+      "loss": 1.0658,
+      "step": 1102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9769953659302806e-05,
+      "loss": 1.0023,
+      "step": 1103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976947649463974e-05,
+      "loss": 1.0328,
+      "step": 1104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9768998841389524e-05,
+      "loss": 1.0504,
+      "step": 1105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9768520699576048e-05,
+      "loss": 1.0299,
+      "step": 1106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9768042069223228e-05,
+      "loss": 0.7044,
+      "step": 1107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9767562950354998e-05,
+      "loss": 1.0896,
+      "step": 1108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9767083342995315e-05,
+      "loss": 0.7076,
+      "step": 1109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9766603247168174e-05,
+      "loss": 1.1479,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9766122662897584e-05,
+      "loss": 0.7062,
+      "step": 1111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976564159020757e-05,
+      "loss": 1.0982,
+      "step": 1112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9765160029122198e-05,
+      "loss": 1.09,
+      "step": 1113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764677979665555e-05,
+      "loss": 1.0046,
+      "step": 1114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9764195441861743e-05,
+      "loss": 1.0579,
+      "step": 1115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9763712415734902e-05,
+      "loss": 0.7455,
+      "step": 1116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9763228901309176e-05,
+      "loss": 1.0966,
+      "step": 1117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976274489860876e-05,
+      "loss": 1.0875,
+      "step": 1118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9762260407657855e-05,
+      "loss": 1.0013,
+      "step": 1119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761775428480686e-05,
+      "loss": 1.0465,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9761289961101516e-05,
+      "loss": 1.0203,
+      "step": 1121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760804005544617e-05,
+      "loss": 0.728,
+      "step": 1122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9760317561834295e-05,
+      "loss": 1.1704,
+      "step": 1123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975983062999488e-05,
+      "loss": 0.9727,
+      "step": 1124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975934321005072e-05,
+      "loss": 1.0608,
+      "step": 1125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758855302026198e-05,
+      "loss": 1.087,
+      "step": 1126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9758366905945706e-05,
+      "loss": 0.9402,
+      "step": 1127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757878021833676e-05,
+      "loss": 1.0651,
+      "step": 1128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9757388649714558e-05,
+      "loss": 0.9805,
+      "step": 1129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9756898789612823e-05,
+      "loss": 0.9749,
+      "step": 1130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975640844155297e-05,
+      "loss": 1.0927,
+      "step": 1131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755917605559526e-05,
+      "loss": 1.1058,
+      "step": 1132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9755426281657034e-05,
+      "loss": 1.0237,
+      "step": 1133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975493446987007e-05,
+      "loss": 1.1331,
+      "step": 1134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975444217022323e-05,
+      "loss": 1.1592,
+      "step": 1135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975394938274113e-05,
+      "loss": 1.1221,
+      "step": 1136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9753456107448416e-05,
+      "loss": 1.0084,
+      "step": 1137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975296234436976e-05,
+      "loss": 1.2236,
+      "step": 1138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9752468093529856e-05,
+      "loss": 1.0858,
+      "step": 1139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9751973354953425e-05,
+      "loss": 0.896,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97514781286652e-05,
+      "loss": 1.1149,
+      "step": 1141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.975098241468996e-05,
+      "loss": 1.1161,
+      "step": 1142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9750486213052488e-05,
+      "loss": 1.0308,
+      "step": 1143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97499895237776e-05,
+      "loss": 1.1402,
+      "step": 1144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9749492346890142e-05,
+      "loss": 0.9401,
+      "step": 1145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9748994682414974e-05,
+      "loss": 1.0522,
+      "step": 1146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974849653037699e-05,
+      "loss": 1.1728,
+      "step": 1147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9747997890801094e-05,
+      "loss": 1.0965,
+      "step": 1148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9747498763712235e-05,
+      "loss": 1.1056,
+      "step": 1149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9746999149135363e-05,
+      "loss": 1.0454,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9746499047095478e-05,
+      "loss": 0.9506,
+      "step": 1151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974599845761758e-05,
+      "loss": 1.0459,
+      "step": 1152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974549738072671e-05,
+      "loss": 1.1464,
+      "step": 1153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744995816447927e-05,
+      "loss": 1.0909,
+      "step": 1154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974449376480631e-05,
+      "loss": 1.0871,
+      "step": 1155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743991225826974e-05,
+      "loss": 0.9646,
+      "step": 1156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9743488199535052e-05,
+      "loss": 1.0453,
+      "step": 1157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9742984685955694e-05,
+      "loss": 1.0506,
+      "step": 1158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974248068511409e-05,
+      "loss": 1.0524,
+      "step": 1159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9741976197035438e-05,
+      "loss": 1.1111,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9741471221744973e-05,
+      "loss": 0.9825,
+      "step": 1161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.974096575926795e-05,
+      "loss": 0.6869,
+      "step": 1162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9740459809629646e-05,
+      "loss": 1.091,
+      "step": 1163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739953372855364e-05,
+      "loss": 1.0758,
+      "step": 1164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9739446448970436e-05,
+      "loss": 1.0136,
+      "step": 1165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9738939038000207e-05,
+      "loss": 1.0205,
+      "step": 1166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973843113997006e-05,
+      "loss": 1.0399,
+      "step": 1167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9737922754905388e-05,
+      "loss": 0.9853,
+      "step": 1168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9737413882831627e-05,
+      "loss": 0.9662,
+      "step": 1169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973690452377422e-05,
+      "loss": 1.0991,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9736394677758637e-05,
+      "loss": 1.0812,
+      "step": 1171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735884344810384e-05,
+      "loss": 1.0564,
+      "step": 1172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9735373524954978e-05,
+      "loss": 1.1929,
+      "step": 1173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973486221821797e-05,
+      "loss": 0.7633,
+      "step": 1174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973435042462493e-05,
+      "loss": 1.0917,
+      "step": 1175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9733838144201455e-05,
+      "loss": 1.0731,
+      "step": 1176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973332537697316e-05,
+      "loss": 1.0566,
+      "step": 1177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9732812122965695e-05,
+      "loss": 0.9618,
+      "step": 1178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973229838220472e-05,
+      "loss": 1.1351,
+      "step": 1179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973178415471594e-05,
+      "loss": 1.0042,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973126944052507e-05,
+      "loss": 1.0408,
+      "step": 1181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.973075423965784e-05,
+      "loss": 1.007,
+      "step": 1182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9730238552140026e-05,
+      "loss": 0.7149,
+      "step": 1183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9729722377997416e-05,
+      "loss": 1.0709,
+      "step": 1184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972920571725583e-05,
+      "loss": 1.1006,
+      "step": 1185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9728688569941098e-05,
+      "loss": 1.0686,
+      "step": 1186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9728170936079088e-05,
+      "loss": 1.1794,
+      "step": 1187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9727652815695686e-05,
+      "loss": 1.1156,
+      "step": 1188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972713420881681e-05,
+      "loss": 1.0821,
+      "step": 1189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9726615115468385e-05,
+      "loss": 1.1654,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9726095535676382e-05,
+      "loss": 0.9427,
+      "step": 1191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9725575469466786e-05,
+      "loss": 1.1033,
+      "step": 1192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9725054916865595e-05,
+      "loss": 0.997,
+      "step": 1193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9724533877898856e-05,
+      "loss": 1.117,
+      "step": 1194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972401235259262e-05,
+      "loss": 1.0812,
+      "step": 1195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972349034097297e-05,
+      "loss": 0.9958,
+      "step": 1196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972296784306601e-05,
+      "loss": 1.0674,
+      "step": 1197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972244485889788e-05,
+      "loss": 1.003,
+      "step": 1198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721921388494727e-05,
+      "loss": 1.1658,
+      "step": 1199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721397431882732e-05,
+      "loss": 1.0288,
+      "step": 1200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97208729890881e-05,
+      "loss": 1.1005,
+      "step": 1201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720348060137062e-05,
+      "loss": 1.1005,
+      "step": 1202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719822645055865e-05,
+      "loss": 0.9206,
+      "step": 1203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971929674387079e-05,
+      "loss": 1.125,
+      "step": 1204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718770356608138e-05,
+      "loss": 0.9941,
+      "step": 1205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971824348329423e-05,
+      "loss": 0.9999,
+      "step": 1206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.971771612395542e-05,
+      "loss": 0.722,
+      "step": 1207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9717188278618083e-05,
+      "loss": 1.0829,
+      "step": 1208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9716659947308616e-05,
+      "loss": 1.0698,
+      "step": 1209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.971613113005344e-05,
+      "loss": 1.103,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9715601826879002e-05,
+      "loss": 1.0475,
+      "step": 1211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9715072037811775e-05,
+      "loss": 1.0403,
+      "step": 1212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9714541762878253e-05,
+      "loss": 0.9557,
+      "step": 1213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.971401100210496e-05,
+      "loss": 1.1019,
+      "step": 1214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9713479755518436e-05,
+      "loss": 0.9751,
+      "step": 1215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.971294802314525e-05,
+      "loss": 1.0555,
+      "step": 1216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9712415805011997e-05,
+      "loss": 0.9905,
+      "step": 1217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9711883101145292e-05,
+      "loss": 1.022,
+      "step": 1218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9711349911571775e-05,
+      "loss": 1.0465,
+      "step": 1219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9710816236318114e-05,
+      "loss": 1.0475,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9710282075411002e-05,
+      "loss": 1.1566,
+      "step": 1221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9709747428877142e-05,
+      "loss": 1.1093,
+      "step": 1222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970921229674329e-05,
+      "loss": 1.0703,
+      "step": 1223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970867667903619e-05,
+      "loss": 1.064,
+      "step": 1224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9708140575782645e-05,
+      "loss": 1.0439,
+      "step": 1225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9707603987009457e-05,
+      "loss": 0.9871,
+      "step": 1226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9707066912743463e-05,
+      "loss": 1.0912,
+      "step": 1227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970652935301153e-05,
+      "loss": 1.0501,
+      "step": 1228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970599130784053e-05,
+      "loss": 1.0076,
+      "step": 1229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970545277725738e-05,
+      "loss": 1.0113,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9704913761289012e-05,
+      "loss": 1.0319,
+      "step": 1231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970437425996238e-05,
+      "loss": 0.9495,
+      "step": 1232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9703834273304466e-05,
+      "loss": 1.1134,
+      "step": 1233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9703293801342278e-05,
+      "loss": 1.013,
+      "step": 1234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9702752844102843e-05,
+      "loss": 1.0538,
+      "step": 1235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970221140161322e-05,
+      "loss": 1.0764,
+      "step": 1236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9701669473900484e-05,
+      "loss": 0.72,
+      "step": 1237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9701127060991736e-05,
+      "loss": 1.1147,
+      "step": 1238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9700584162914104e-05,
+      "loss": 1.0742,
+      "step": 1239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.970004077969474e-05,
+      "loss": 1.0116,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969949691136082e-05,
+      "loss": 1.1071,
+      "step": 1241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9698952557939543e-05,
+      "loss": 1.0218,
+      "step": 1242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9698407719458134e-05,
+      "loss": 0.9933,
+      "step": 1243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969786239594384e-05,
+      "loss": 1.1069,
+      "step": 1244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9697316587423934e-05,
+      "loss": 0.9524,
+      "step": 1245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969677029392571e-05,
+      "loss": 1.0267,
+      "step": 1246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9696223515476495e-05,
+      "loss": 1.1189,
+      "step": 1247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9695676252103632e-05,
+      "loss": 1.115,
+      "step": 1248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9695128503834486e-05,
+      "loss": 1.2156,
+      "step": 1249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694580270696454e-05,
+      "loss": 1.0325,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9694031552716953e-05,
+      "loss": 1.0724,
+      "step": 1251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9693482349923433e-05,
+      "loss": 1.0413,
+      "step": 1252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692932662343346e-05,
+      "loss": 1.0454,
+      "step": 1253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692382490004195e-05,
+      "loss": 1.0149,
+      "step": 1254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969183183293349e-05,
+      "loss": 1.1272,
+      "step": 1255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.969128069115877e-05,
+      "loss": 1.1159,
+      "step": 1256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690729064707602e-05,
+      "loss": 1.0734,
+      "step": 1257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9690176953607565e-05,
+      "loss": 1.0446,
+      "step": 1258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689624357886285e-05,
+      "loss": 1.0568,
+      "step": 1259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9689071277571384e-05,
+      "loss": 0.9989,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9688517712690533e-05,
+      "loss": 1.0347,
+      "step": 1261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968796366327141e-05,
+      "loss": 1.0889,
+      "step": 1262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968740912934173e-05,
+      "loss": 0.923,
+      "step": 1263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9686854110929217e-05,
+      "loss": 0.7445,
+      "step": 1264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9686298608061635e-05,
+      "loss": 1.0439,
+      "step": 1265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685742620766766e-05,
+      "loss": 1.1327,
+      "step": 1266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9685186149072412e-05,
+      "loss": 1.1343,
+      "step": 1267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968462919300641e-05,
+      "loss": 1.0424,
+      "step": 1268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9684071752596604e-05,
+      "loss": 1.1778,
+      "step": 1269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968351382787088e-05,
+      "loss": 1.0616,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968295541885714e-05,
+      "loss": 1.1332,
+      "step": 1271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9682396525583306e-05,
+      "loss": 0.681,
+      "step": 1272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9681837148077337e-05,
+      "loss": 1.1017,
+      "step": 1273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96812772863672e-05,
+      "loss": 1.1063,
+      "step": 1274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96807169404809e-05,
+      "loss": 1.1107,
+      "step": 1275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9680156110446464e-05,
+      "loss": 1.1187,
+      "step": 1276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967959479629193e-05,
+      "loss": 1.1041,
+      "step": 1277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9679032998045376e-05,
+      "loss": 0.6434,
+      "step": 1278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96784707157349e-05,
+      "loss": 1.0388,
+      "step": 1279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967790794938862e-05,
+      "loss": 1.12,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9677344699034684e-05,
+      "loss": 1.0148,
+      "step": 1281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9676780964701256e-05,
+      "loss": 1.1208,
+      "step": 1282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9676216746416533e-05,
+      "loss": 0.9654,
+      "step": 1283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9675652044208727e-05,
+      "loss": 1.0657,
+      "step": 1284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967508685810609e-05,
+      "loss": 0.9863,
+      "step": 1285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967452118813688e-05,
+      "loss": 1.1259,
+      "step": 1286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9673955034329387e-05,
+      "loss": 1.093,
+      "step": 1287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967338839671193e-05,
+      "loss": 1.1088,
+      "step": 1288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9672821275312843e-05,
+      "loss": 1.1575,
+      "step": 1289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9672253670160494e-05,
+      "loss": 1.0741,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967168558128326e-05,
+      "loss": 0.9968,
+      "step": 1291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967111700870956e-05,
+      "loss": 0.9392,
+      "step": 1292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.967054795246783e-05,
+      "loss": 1.0329,
+      "step": 1293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966997841258653e-05,
+      "loss": 1.0435,
+      "step": 1294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9669408389094132e-05,
+      "loss": 0.9769,
+      "step": 1295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9668837882019157e-05,
+      "loss": 0.9875,
+      "step": 1296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966826689139013e-05,
+      "loss": 1.1882,
+      "step": 1297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966769541723561e-05,
+      "loss": 1.1138,
+      "step": 1298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966712345958418e-05,
+      "loss": 1.0108,
+      "step": 1299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9666551018464436e-05,
+      "loss": 1.2096,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9665978093905014e-05,
+      "loss": 1.0307,
+      "step": 1301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9665404685934566e-05,
+      "loss": 1.0816,
+      "step": 1302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9664830794581767e-05,
+      "loss": 1.0216,
+      "step": 1303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966425641987532e-05,
+      "loss": 1.1393,
+      "step": 1304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663681561843948e-05,
+      "loss": 1.1339,
+      "step": 1305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9663106220516403e-05,
+      "loss": 1.0236,
+      "step": 1306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9662530395921457e-05,
+      "loss": 1.0932,
+      "step": 1307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966195408808791e-05,
+      "loss": 1.0723,
+      "step": 1308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966137729704458e-05,
+      "loss": 1.0612,
+      "step": 1309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9660800022820318e-05,
+      "loss": 1.1099,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.966022226544399e-05,
+      "loss": 1.0738,
+      "step": 1311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9659644024944494e-05,
+      "loss": 1.034,
+      "step": 1312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9659065301350752e-05,
+      "loss": 1.0563,
+      "step": 1313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9658486094691697e-05,
+      "loss": 1.0106,
+      "step": 1314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96579064049963e-05,
+      "loss": 1.013,
+      "step": 1315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965732623229356e-05,
+      "loss": 0.7187,
+      "step": 1316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656745576612482e-05,
+      "loss": 1.0186,
+      "step": 1317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9656164437982114e-05,
+      "loss": 1.089,
+      "step": 1318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9655582816431513e-05,
+      "loss": 1.1128,
+      "step": 1319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965500071198977e-05,
+      "loss": 1.032,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654418124685993e-05,
+      "loss": 0.9968,
+      "step": 1321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9653835054549326e-05,
+      "loss": 1.0041,
+      "step": 1322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.965325150160892e-05,
+      "loss": 1.1987,
+      "step": 1323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652667465893968e-05,
+      "loss": 1.0574,
+      "step": 1324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9652082947433672e-05,
+      "loss": 1.1668,
+      "step": 1325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9651497946257268e-05,
+      "loss": 0.6745,
+      "step": 1326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9650912462394013e-05,
+      "loss": 0.7173,
+      "step": 1327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9650326495873184e-05,
+      "loss": 1.0127,
+      "step": 1328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964974004672409e-05,
+      "loss": 1.0352,
+      "step": 1329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9649153114976062e-05,
+      "loss": 0.9995,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9648565700658452e-05,
+      "loss": 1.1758,
+      "step": 1331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9647977803800632e-05,
+      "loss": 0.6971,
+      "step": 1332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9647389424432012e-05,
+      "loss": 0.9885,
+      "step": 1333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964680056258201e-05,
+      "loss": 1.0798,
+      "step": 1334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9646211218280084e-05,
+      "loss": 1.0468,
+      "step": 1335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96456213915557e-05,
+      "loss": 1.0886,
+      "step": 1336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645031082438363e-05,
+      "loss": 1.0623,
+      "step": 1337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964444029095759e-05,
+      "loss": 1.0507,
+      "step": 1338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964384901714293e-05,
+      "loss": 1.1215,
+      "step": 1339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643257261023954e-05,
+      "loss": 1.1674,
+      "step": 1340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642665022630256e-05,
+      "loss": 1.064,
+      "step": 1341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642072301991453e-05,
+      "loss": 1.0458,
+      "step": 1342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641479099137194e-05,
+      "loss": 1.1067,
+      "step": 1343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640885414097138e-05,
+      "loss": 0.9914,
+      "step": 1344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964029124690098e-05,
+      "loss": 1.0702,
+      "step": 1345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639696597578438e-05,
+      "loss": 1.165,
+      "step": 1346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639101466159245e-05,
+      "loss": 1.0815,
+      "step": 1347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963850585267317e-05,
+      "loss": 1.0324,
+      "step": 1348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9637909757149997e-05,
+      "loss": 1.021,
+      "step": 1349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9637313179619538e-05,
+      "loss": 1.1399,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9636716120111635e-05,
+      "loss": 1.0791,
+      "step": 1351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9636118578656137e-05,
+      "loss": 1.0775,
+      "step": 1352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9635520555282935e-05,
+      "loss": 1.1015,
+      "step": 1353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.963492205002194e-05,
+      "loss": 1.0706,
+      "step": 1354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9634323062903078e-05,
+      "loss": 1.0816,
+      "step": 1355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.963372359395631e-05,
+      "loss": 0.9903,
+      "step": 1356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.963312364321161e-05,
+      "loss": 1.0105,
+      "step": 1357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9632523210698987e-05,
+      "loss": 1.1099,
+      "step": 1358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9631922296448468e-05,
+      "loss": 1.0846,
+      "step": 1359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.963132090049011e-05,
+      "loss": 1.0659,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9630719022853983e-05,
+      "loss": 1.0963,
+      "step": 1361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9630116663570192e-05,
+      "loss": 0.9709,
+      "step": 1362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9629513822668866e-05,
+      "loss": 1.1623,
+      "step": 1363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9628910500180143e-05,
+      "loss": 1.111,
+      "step": 1364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9628306696134207e-05,
+      "loss": 0.6868,
+      "step": 1365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9627702410561248e-05,
+      "loss": 1.0396,
+      "step": 1366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9627097643491488e-05,
+      "loss": 1.0854,
+      "step": 1367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9626492394955177e-05,
+      "loss": 0.9755,
+      "step": 1368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962588666498258e-05,
+      "loss": 1.0513,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9625280453603997e-05,
+      "loss": 1.0349,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962467376084974e-05,
+      "loss": 1.0189,
+      "step": 1371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962406658675015e-05,
+      "loss": 1.019,
+      "step": 1372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962345893133559e-05,
+      "loss": 1.0737,
+      "step": 1373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9622850794636456e-05,
+      "loss": 0.9941,
+      "step": 1374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9622242176683163e-05,
+      "loss": 0.963,
+      "step": 1375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9621633077506143e-05,
+      "loss": 0.9966,
+      "step": 1376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9621023497135865e-05,
+      "loss": 1.1031,
+      "step": 1377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.962041343560281e-05,
+      "loss": 1.0201,
+      "step": 1378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9619802892937485e-05,
+      "loss": 1.0189,
+      "step": 1379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9619191869170433e-05,
+      "loss": 1.1344,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9618580364332205e-05,
+      "loss": 0.9972,
+      "step": 1381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9617968378453386e-05,
+      "loss": 0.7691,
+      "step": 1382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9617355911564582e-05,
+      "loss": 1.0511,
+      "step": 1383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9616742963696428e-05,
+      "loss": 1.0673,
+      "step": 1384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9616129534879567e-05,
+      "loss": 1.0502,
+      "step": 1385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9615515625144694e-05,
+      "loss": 1.1312,
+      "step": 1386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9614901234522495e-05,
+      "loss": 1.1208,
+      "step": 1387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961428636304371e-05,
+      "loss": 1.1344,
+      "step": 1388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9613671010739082e-05,
+      "loss": 1.0712,
+      "step": 1389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9613055177639384e-05,
+      "loss": 1.0398,
+      "step": 1390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9612438863775424e-05,
+      "loss": 1.0536,
+      "step": 1391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9611822069178015e-05,
+      "loss": 1.0936,
+      "step": 1392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961120479387801e-05,
+      "loss": 1.0692,
+      "step": 1393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9610587037906275e-05,
+      "loss": 1.0427,
+      "step": 1394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9609968801293716e-05,
+      "loss": 0.9572,
+      "step": 1395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9609350084071238e-05,
+      "loss": 1.1176,
+      "step": 1396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960873088626979e-05,
+      "loss": 1.0056,
+      "step": 1397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960811120792034e-05,
+      "loss": 1.159,
+      "step": 1398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960749104905388e-05,
+      "loss": 1.119,
+      "step": 1399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606870409701422e-05,
+      "loss": 1.0737,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9606249289894007e-05,
+      "loss": 1.0045,
+      "step": 1401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9605627689662697e-05,
+      "loss": 1.0844,
+      "step": 1402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960500560903858e-05,
+      "loss": 1.106,
+      "step": 1403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9604383048052767e-05,
+      "loss": 0.9996,
+      "step": 1404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9603760006736396e-05,
+      "loss": 1.0561,
+      "step": 1405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960313648512062e-05,
+      "loss": 1.0755,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960251248323663e-05,
+      "loss": 1.0727,
+      "step": 1407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9601888001115627e-05,
+      "loss": 1.1189,
+      "step": 1408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9601263038788848e-05,
+      "loss": 1.0718,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960063759628754e-05,
+      "loss": 1.1007,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.960001167364299e-05,
+      "loss": 1.0706,
+      "step": 1411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95993852708865e-05,
+      "loss": 1.1101,
+      "step": 1412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9598758388049396e-05,
+      "loss": 0.9732,
+      "step": 1413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959813102516303e-05,
+      "loss": 1.0537,
+      "step": 1414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597503182258774e-05,
+      "loss": 0.6496,
+      "step": 1415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9596874859368035e-05,
+      "loss": 1.1622,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959624605652223e-05,
+      "loss": 1.0323,
+      "step": 1417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959561677375281e-05,
+      "loss": 0.739,
+      "step": 1418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959498701109124e-05,
+      "loss": 1.1224,
+      "step": 1419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9594356768569026e-05,
+      "loss": 1.012,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959372604621768e-05,
+      "loss": 1.0144,
+      "step": 1421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959309484406875e-05,
+      "loss": 1.0532,
+      "step": 1422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9592463162153797e-05,
+      "loss": 1.1015,
+      "step": 1423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9591831000504417e-05,
+      "loss": 1.1836,
+      "step": 1424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9591198359152223e-05,
+      "loss": 0.7332,
+      "step": 1425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.959056523812886e-05,
+      "loss": 1.0093,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9589931637465983e-05,
+      "loss": 1.1098,
+      "step": 1427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9589297557195285e-05,
+      "loss": 1.0797,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958866299734848e-05,
+      "loss": 0.9626,
+      "step": 1429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9588027957957295e-05,
+      "loss": 0.6675,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587392439053494e-05,
+      "loss": 1.0943,
+      "step": 1431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958675644066886e-05,
+      "loss": 1.0771,
+      "step": 1432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9586119962835203e-05,
+      "loss": 1.0879,
+      "step": 1433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9585483005584347e-05,
+      "loss": 1.0119,
+      "step": 1434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584845568948157e-05,
+      "loss": 1.074,
+      "step": 1435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9584207652958502e-05,
+      "loss": 1.1907,
+      "step": 1436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9583569257647296e-05,
+      "loss": 0.715,
+      "step": 1437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9582930383046456e-05,
+      "loss": 1.0947,
+      "step": 1438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9582291029187938e-05,
+      "loss": 1.0596,
+      "step": 1439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9581651196103718e-05,
+      "loss": 1.0389,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9581010883825794e-05,
+      "loss": 1.102,
+      "step": 1441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958037009238619e-05,
+      "loss": 1.0368,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957972882181695e-05,
+      "loss": 1.0164,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957908707215015e-05,
+      "loss": 0.9878,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9578444843417878e-05,
+      "loss": 1.1256,
+      "step": 1445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957780213565226e-05,
+      "loss": 1.0219,
+      "step": 1446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9577158948885428e-05,
+      "loss": 1.017,
+      "step": 1447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957651528314956e-05,
+      "loss": 1.0296,
+      "step": 1448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9575871138476847e-05,
+      "loss": 1.0266,
+      "step": 1449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9575226514899497e-05,
+      "loss": 1.0557,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9574581412449753e-05,
+      "loss": 1.1234,
+      "step": 1451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9573935831159873e-05,
+      "loss": 1.103,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9573289771062153e-05,
+      "loss": 1.0002,
+      "step": 1453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957264323218889e-05,
+      "loss": 1.0356,
+      "step": 1454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.957199621457243e-05,
+      "loss": 1.117,
+      "step": 1455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9571348718245123e-05,
+      "loss": 0.9913,
+      "step": 1456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9570700743239355e-05,
+      "loss": 1.0258,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9570052289587537e-05,
+      "loss": 1.054,
+      "step": 1458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956940335732209e-05,
+      "loss": 1.0424,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9568753946475475e-05,
+      "loss": 1.0211,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956810405708017e-05,
+      "loss": 0.7341,
+      "step": 1461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9567453689168673e-05,
+      "loss": 1.0522,
+      "step": 1462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.956680284277351e-05,
+      "loss": 1.0851,
+      "step": 1463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9566151517927235e-05,
+      "loss": 1.0549,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9565499714662418e-05,
+      "loss": 0.6978,
+      "step": 1465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564847433011657e-05,
+      "loss": 1.1121,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9564194673007577e-05,
+      "loss": 0.7435,
+      "step": 1467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9563541434682823e-05,
+      "loss": 1.02,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9562887718070063e-05,
+      "loss": 0.9725,
+      "step": 1469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9562233523201986e-05,
+      "loss": 1.1223,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9561578850111318e-05,
+      "loss": 1.1488,
+      "step": 1471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560923698830795e-05,
+      "loss": 0.9848,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9560268069393183e-05,
+      "loss": 1.0254,
+      "step": 1473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955961196183127e-05,
+      "loss": 1.1494,
+      "step": 1474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955895537617787e-05,
+      "loss": 1.1017,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955829831246582e-05,
+      "loss": 1.054,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9557640770727984e-05,
+      "loss": 1.1024,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955698275099724e-05,
+      "loss": 1.0981,
+      "step": 1478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95563242533065e-05,
+      "loss": 1.0039,
+      "step": 1479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95556652776887e-05,
+      "loss": 1.0824,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555005824176795e-05,
+      "loss": 1.1053,
+      "step": 1481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9554345892803758e-05,
+      "loss": 1.0328,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95536854836026e-05,
+      "loss": 1.0847,
+      "step": 1483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553024596606348e-05,
+      "loss": 1.0535,
+      "step": 1484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552363231848054e-05,
+      "loss": 1.0379,
+      "step": 1485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551701389360796e-05,
+      "loss": 1.1486,
+      "step": 1486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551039069177672e-05,
+      "loss": 1.1591,
+      "step": 1487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95503762713318e-05,
+      "loss": 0.6621,
+      "step": 1488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549712995856334e-05,
+      "loss": 1.0716,
+      "step": 1489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549049242784447e-05,
+      "loss": 1.059,
+      "step": 1490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9548385012149332e-05,
+      "loss": 1.0008,
+      "step": 1491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9547720303984205e-05,
+      "loss": 1.13,
+      "step": 1492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9547055118322315e-05,
+      "loss": 0.9878,
+      "step": 1493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9546389455196925e-05,
+      "loss": 1.0341,
+      "step": 1494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9545723314641326e-05,
+      "loss": 0.708,
+      "step": 1495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9545056696688834e-05,
+      "loss": 1.0296,
+      "step": 1496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9544389601372788e-05,
+      "loss": 1.0753,
+      "step": 1497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.954372202872655e-05,
+      "loss": 1.1512,
+      "step": 1498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9543053978783505e-05,
+      "loss": 1.058,
+      "step": 1499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9542385451577066e-05,
+      "loss": 0.9869,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9541716447140663e-05,
+      "loss": 1.1011,
+      "step": 1501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9541046965507758e-05,
+      "loss": 1.1246,
+      "step": 1502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9540377006711832e-05,
+      "loss": 0.7028,
+      "step": 1503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.953970657078639e-05,
+      "loss": 0.9322,
+      "step": 1504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.953903565776496e-05,
+      "loss": 1.0802,
+      "step": 1505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9538364267681098e-05,
+      "loss": 1.1284,
+      "step": 1506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9537692400568378e-05,
+      "loss": 1.1246,
+      "step": 1507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.953702005646041e-05,
+      "loss": 0.9526,
+      "step": 1508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9536347235390808e-05,
+      "loss": 0.9947,
+      "step": 1509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9535673937393226e-05,
+      "loss": 1.1086,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9535000162501335e-05,
+      "loss": 1.0985,
+      "step": 1511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.953432591074883e-05,
+      "loss": 1.085,
+      "step": 1512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9533651182169443e-05,
+      "loss": 1.1156,
+      "step": 1513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9532975976796903e-05,
+      "loss": 1.0128,
+      "step": 1514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9532300294664984e-05,
+      "loss": 0.8886,
+      "step": 1515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.953162413580748e-05,
+      "loss": 1.0136,
+      "step": 1516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9530947500258202e-05,
+      "loss": 1.0552,
+      "step": 1517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9530270388050997e-05,
+      "loss": 0.9533,
+      "step": 1518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9529592799219725e-05,
+      "loss": 0.9545,
+      "step": 1519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9528914733798266e-05,
+      "loss": 1.0233,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952823619182054e-05,
+      "loss": 1.0105,
+      "step": 1521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9527557173320483e-05,
+      "loss": 0.9901,
+      "step": 1522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9526877678332053e-05,
+      "loss": 1.0275,
+      "step": 1523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9526197706889226e-05,
+      "loss": 0.9609,
+      "step": 1524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952551725902601e-05,
+      "loss": 1.1043,
+      "step": 1525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9524836334776442e-05,
+      "loss": 1.1004,
+      "step": 1526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9524154934174568e-05,
+      "loss": 0.6749,
+      "step": 1527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9523473057254474e-05,
+      "loss": 1.0554,
+      "step": 1528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9522790704050253e-05,
+      "loss": 1.0498,
+      "step": 1529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952210787459604e-05,
+      "loss": 1.0614,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9521424568925977e-05,
+      "loss": 1.1412,
+      "step": 1531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.952074078707424e-05,
+      "loss": 0.9663,
+      "step": 1532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9520056529075027e-05,
+      "loss": 0.7126,
+      "step": 1533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9519371794962554e-05,
+      "loss": 1.0671,
+      "step": 1534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9518686584771074e-05,
+      "loss": 1.0224,
+      "step": 1535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951800089853485e-05,
+      "loss": 1.0378,
+      "step": 1536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9517314736288172e-05,
+      "loss": 1.135,
+      "step": 1537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951662809806536e-05,
+      "loss": 1.0626,
+      "step": 1538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9515940983900753e-05,
+      "loss": 1.054,
+      "step": 1539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951525339382872e-05,
+      "loss": 1.0492,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9514565327883637e-05,
+      "loss": 1.1619,
+      "step": 1541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9513876786099922e-05,
+      "loss": 1.0776,
+      "step": 1542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951318776851201e-05,
+      "loss": 1.0847,
+      "step": 1543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.951249827515436e-05,
+      "loss": 1.0782,
+      "step": 1544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9511808306061454e-05,
+      "loss": 0.9141,
+      "step": 1545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95111178612678e-05,
+      "loss": 1.1217,
+      "step": 1546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9510426940807924e-05,
+      "loss": 1.0281,
+      "step": 1547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9509735544716385e-05,
+      "loss": 1.0553,
+      "step": 1548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9509043673027755e-05,
+      "loss": 1.1311,
+      "step": 1549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950835132577664e-05,
+      "loss": 1.0454,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950765850299767e-05,
+      "loss": 1.0122,
+      "step": 1551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9506965204725484e-05,
+      "loss": 1.0722,
+      "step": 1552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950627143099476e-05,
+      "loss": 1.0482,
+      "step": 1553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9505577181840195e-05,
+      "loss": 0.7153,
+      "step": 1554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950488245729651e-05,
+      "loss": 0.8813,
+      "step": 1555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9504187257398445e-05,
+      "loss": 1.0771,
+      "step": 1556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9503491582180777e-05,
+      "loss": 1.1567,
+      "step": 1557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.950279543167829e-05,
+      "loss": 1.1015,
+      "step": 1558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9502098805925797e-05,
+      "loss": 0.9778,
+      "step": 1559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9501401704958145e-05,
+      "loss": 1.0421,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9500704128810196e-05,
+      "loss": 1.0979,
+      "step": 1561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9500006077516833e-05,
+      "loss": 1.1436,
+      "step": 1562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949930755111297e-05,
+      "loss": 0.6908,
+      "step": 1563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949860854963354e-05,
+      "loss": 1.1119,
+      "step": 1564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9497909073113502e-05,
+      "loss": 1.0637,
+      "step": 1565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9497209121587837e-05,
+      "loss": 1.107,
+      "step": 1566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949650869509155e-05,
+      "loss": 1.0542,
+      "step": 1567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949580779365967e-05,
+      "loss": 1.0608,
+      "step": 1568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9495106417327257e-05,
+      "loss": 1.014,
+      "step": 1569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949440456612938e-05,
+      "loss": 1.1338,
+      "step": 1570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9493702240101143e-05,
+      "loss": 1.0942,
+      "step": 1571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9492999439277668e-05,
+      "loss": 1.0284,
+      "step": 1572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9492296163694107e-05,
+      "loss": 1.1655,
+      "step": 1573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9491592413385627e-05,
+      "loss": 0.9556,
+      "step": 1574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949088818838743e-05,
+      "loss": 1.027,
+      "step": 1575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.949018348873473e-05,
+      "loss": 1.0486,
+      "step": 1576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948947831446277e-05,
+      "loss": 1.0243,
+      "step": 1577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9488772665606824e-05,
+      "loss": 1.0757,
+      "step": 1578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9488066542202176e-05,
+      "loss": 1.018,
+      "step": 1579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948735994428414e-05,
+      "loss": 0.9702,
+      "step": 1580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9486652871888058e-05,
+      "loss": 0.7132,
+      "step": 1581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948594532504929e-05,
+      "loss": 0.9991,
+      "step": 1582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948523730380322e-05,
+      "loss": 1.1385,
+      "step": 1583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9484528808185258e-05,
+      "loss": 0.9486,
+      "step": 1584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9483819838230842e-05,
+      "loss": 1.0935,
+      "step": 1585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.948311039397542e-05,
+      "loss": 1.0348,
+      "step": 1586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9482400475454482e-05,
+      "loss": 0.6716,
+      "step": 1587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9481690082703524e-05,
+      "loss": 1.0272,
+      "step": 1588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9480979215758077e-05,
+      "loss": 1.0903,
+      "step": 1589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9480267874653692e-05,
+      "loss": 1.0854,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947955605942595e-05,
+      "loss": 1.0982,
+      "step": 1591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9478843770110434e-05,
+      "loss": 1.1174,
+      "step": 1592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947813100674279e-05,
+      "loss": 1.0319,
+      "step": 1593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9477417769358644e-05,
+      "loss": 0.9786,
+      "step": 1594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9476704057993676e-05,
+      "loss": 1.0372,
+      "step": 1595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947598987268358e-05,
+      "loss": 0.9718,
+      "step": 1596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947527521346407e-05,
+      "loss": 1.058,
+      "step": 1597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947456008037089e-05,
+      "loss": 0.7348,
+      "step": 1598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9473844473439803e-05,
+      "loss": 1.0411,
+      "step": 1599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94731283927066e-05,
+      "loss": 1.0172,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947241183820709e-05,
+      "loss": 0.9983,
+      "step": 1601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.947169480997711e-05,
+      "loss": 1.1128,
+      "step": 1602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9470977308052524e-05,
+      "loss": 1.0321,
+      "step": 1603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9470259332469207e-05,
+      "loss": 1.1428,
+      "step": 1604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9469540883263077e-05,
+      "loss": 1.134,
+      "step": 1605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9468821960470056e-05,
+      "loss": 1.0936,
+      "step": 1606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9468102564126097e-05,
+      "loss": 1.0493,
+      "step": 1607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946738269426719e-05,
+      "loss": 0.9956,
+      "step": 1608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9466662350929325e-05,
+      "loss": 1.0771,
+      "step": 1609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9465941534148533e-05,
+      "loss": 1.0316,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9465220243960863e-05,
+      "loss": 1.0181,
+      "step": 1611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9464498480402384e-05,
+      "loss": 1.1023,
+      "step": 1612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9463776243509202e-05,
+      "loss": 1.0028,
+      "step": 1613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9463053533317424e-05,
+      "loss": 0.6897,
+      "step": 1614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9462330349863202e-05,
+      "loss": 1.0865,
+      "step": 1615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9461606693182707e-05,
+      "loss": 1.0974,
+      "step": 1616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946088256331212e-05,
+      "loss": 1.002,
+      "step": 1617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460157960287662e-05,
+      "loss": 1.0011,
+      "step": 1618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9459432884145573e-05,
+      "loss": 1.0956,
+      "step": 1619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458707334922113e-05,
+      "loss": 1.061,
+      "step": 1620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9457981312653566e-05,
+      "loss": 0.6747,
+      "step": 1621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9457254817376245e-05,
+      "loss": 1.1066,
+      "step": 1622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456527849126484e-05,
+      "loss": 1.0941,
+      "step": 1623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455800407940633e-05,
+      "loss": 1.032,
+      "step": 1624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945507249385508e-05,
+      "loss": 0.7193,
+      "step": 1625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9454344106906225e-05,
+      "loss": 1.0212,
+      "step": 1626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453615247130496e-05,
+      "loss": 1.0468,
+      "step": 1627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9452885914564347e-05,
+      "loss": 1.0181,
+      "step": 1628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945215610924425e-05,
+      "loss": 1.0561,
+      "step": 1629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451425831206706e-05,
+      "loss": 1.0503,
+      "step": 1630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450695080488238e-05,
+      "loss": 0.7082,
+      "step": 1631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9449963857125393e-05,
+      "loss": 1.0994,
+      "step": 1632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9449232161154734e-05,
+      "loss": 0.9853,
+      "step": 1633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9448499992612858e-05,
+      "loss": 1.0194,
+      "step": 1634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9447767351536386e-05,
+      "loss": 1.0107,
+      "step": 1635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.944703423796195e-05,
+      "loss": 1.1333,
+      "step": 1636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9446300651926223e-05,
+      "loss": 1.17,
+      "step": 1637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.944556659346589e-05,
+      "loss": 1.1138,
+      "step": 1638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.944483206261766e-05,
+      "loss": 1.1467,
+      "step": 1639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.944409705941827e-05,
+      "loss": 1.0006,
+      "step": 1640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9443361583904477e-05,
+      "loss": 0.9628,
+      "step": 1641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9442625636113067e-05,
+      "loss": 1.0789,
+      "step": 1642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.944188921608084e-05,
+      "loss": 0.9201,
+      "step": 1643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.944115232384463e-05,
+      "loss": 1.053,
+      "step": 1644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9440414959441293e-05,
+      "loss": 1.1086,
+      "step": 1645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.94396771229077e-05,
+      "loss": 1.0491,
+      "step": 1646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.943893881428075e-05,
+      "loss": 1.0631,
+      "step": 1647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9438200033597372e-05,
+      "loss": 1.0971,
+      "step": 1648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9437460780894516e-05,
+      "loss": 0.9542,
+      "step": 1649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.943672105620914e-05,
+      "loss": 0.9449,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9435980859578258e-05,
+      "loss": 1.1267,
+      "step": 1651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.943524019103887e-05,
+      "loss": 1.0763,
+      "step": 1652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9434499050628032e-05,
+      "loss": 0.7097,
+      "step": 1653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9433757438382806e-05,
+      "loss": 0.9812,
+      "step": 1654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9433015354340275e-05,
+      "loss": 1.085,
+      "step": 1655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9432272798537553e-05,
+      "loss": 1.0868,
+      "step": 1656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9431529771011786e-05,
+      "loss": 1.1049,
+      "step": 1657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9430786271800125e-05,
+      "loss": 1.0111,
+      "step": 1658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9430042300939754e-05,
+      "loss": 1.1307,
+      "step": 1659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9429297858467888e-05,
+      "loss": 1.1245,
+      "step": 1660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9428552944421744e-05,
+      "loss": 1.1012,
+      "step": 1661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9427807558838586e-05,
+      "loss": 1.1332,
+      "step": 1662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9427061701755696e-05,
+      "loss": 1.1228,
+      "step": 1663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9426315373210365e-05,
+      "loss": 1.0128,
+      "step": 1664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9425568573239923e-05,
+      "loss": 1.0374,
+      "step": 1665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.942482130188172e-05,
+      "loss": 0.9908,
+      "step": 1666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9424073559173125e-05,
+      "loss": 0.9072,
+      "step": 1667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.942332534515154e-05,
+      "loss": 1.1352,
+      "step": 1668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9422576659854374e-05,
+      "loss": 1.0473,
+      "step": 1669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9421827503319078e-05,
+      "loss": 1.0546,
+      "step": 1670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9421077875583118e-05,
+      "loss": 1.165,
+      "step": 1671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.942032777668398e-05,
+      "loss": 1.0012,
+      "step": 1672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9419577206659183e-05,
+      "loss": 1.1024,
+      "step": 1673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941882616554626e-05,
+      "loss": 1.1346,
+      "step": 1674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941807465338277e-05,
+      "loss": 1.1098,
+      "step": 1675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9417322670206303e-05,
+      "loss": 0.9849,
+      "step": 1676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9416570216054463e-05,
+      "loss": 1.0056,
+      "step": 1677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9415817290964885e-05,
+      "loss": 1.0459,
+      "step": 1678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941506389497522e-05,
+      "loss": 0.6925,
+      "step": 1679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9414310028123153e-05,
+      "loss": 1.0678,
+      "step": 1680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9413555690446374e-05,
+      "loss": 1.0352,
+      "step": 1681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941280088198262e-05,
+      "loss": 1.0595,
+      "step": 1682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9412045602769634e-05,
+      "loss": 1.0843,
+      "step": 1683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9411289852845194e-05,
+      "loss": 1.1527,
+      "step": 1684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.941053363224709e-05,
+      "loss": 0.9851,
+      "step": 1685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9409776941013146e-05,
+      "loss": 0.9569,
+      "step": 1686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9409019779181206e-05,
+      "loss": 0.9483,
+      "step": 1687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940826214678913e-05,
+      "loss": 1.0312,
+      "step": 1688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9407504043874817e-05,
+      "loss": 1.0809,
+      "step": 1689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940674547047618e-05,
+      "loss": 1.0261,
+      "step": 1690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940598642663115e-05,
+      "loss": 1.0162,
+      "step": 1691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9405226912377693e-05,
+      "loss": 1.0424,
+      "step": 1692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9404466927753795e-05,
+      "loss": 1.1549,
+      "step": 1693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9403706472797462e-05,
+      "loss": 1.0337,
+      "step": 1694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940294554754672e-05,
+      "loss": 1.0481,
+      "step": 1695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.940218415203963e-05,
+      "loss": 1.0234,
+      "step": 1696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9401422286314274e-05,
+      "loss": 1.0282,
+      "step": 1697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9400659950408748e-05,
+      "loss": 0.9657,
+      "step": 1698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9399897144361176e-05,
+      "loss": 1.0229,
+      "step": 1699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9399133868209715e-05,
+      "loss": 1.0675,
+      "step": 1700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9398370121992533e-05,
+      "loss": 1.0271,
+      "step": 1701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9397605905747827e-05,
+      "loss": 0.9083,
+      "step": 1702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9396841219513812e-05,
+      "loss": 0.9928,
+      "step": 1703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.939607606332874e-05,
+      "loss": 1.0465,
+      "step": 1704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9395310437230873e-05,
+      "loss": 0.9829,
+      "step": 1705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9394544341258498e-05,
+      "loss": 1.0753,
+      "step": 1706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9393777775449933e-05,
+      "loss": 1.0396,
+      "step": 1707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9393010739843512e-05,
+      "loss": 1.0099,
+      "step": 1708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9392243234477603e-05,
+      "loss": 1.0542,
+      "step": 1709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9391475259390583e-05,
+      "loss": 1.1405,
+      "step": 1710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9390706814620863e-05,
+      "loss": 1.0223,
+      "step": 1711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9389937900206867e-05,
+      "loss": 1.0262,
+      "step": 1712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938916851618706e-05,
+      "loss": 0.6739,
+      "step": 1713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9388398662599915e-05,
+      "loss": 0.955,
+      "step": 1714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9387628339483935e-05,
+      "loss": 1.0201,
+      "step": 1715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9386857546877642e-05,
+      "loss": 1.0864,
+      "step": 1716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938608628481959e-05,
+      "loss": 1.0326,
+      "step": 1717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9385314553348346e-05,
+      "loss": 0.9657,
+      "step": 1718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9384542352502507e-05,
+      "loss": 0.993,
+      "step": 1719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9383769682320693e-05,
+      "loss": 1.1149,
+      "step": 1720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9382996542841542e-05,
+      "loss": 0.9106,
+      "step": 1721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9382222934103732e-05,
+      "loss": 1.0619,
+      "step": 1722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9381448856145936e-05,
+      "loss": 1.0611,
+      "step": 1723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.938067430900688e-05,
+      "loss": 1.0612,
+      "step": 1724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9379899292725294e-05,
+      "loss": 1.0555,
+      "step": 1725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937912380733994e-05,
+      "loss": 1.0476,
+      "step": 1726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93783478528896e-05,
+      "loss": 1.1026,
+      "step": 1727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9377571429413083e-05,
+      "loss": 1.1062,
+      "step": 1728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9376794536949216e-05,
+      "loss": 1.0242,
+      "step": 1729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9376017175536854e-05,
+      "loss": 1.0362,
+      "step": 1730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937523934521487e-05,
+      "loss": 1.039,
+      "step": 1731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9374461046022172e-05,
+      "loss": 1.0112,
+      "step": 1732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.937368227799768e-05,
+      "loss": 0.9697,
+      "step": 1733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9372903041180347e-05,
+      "loss": 1.0923,
+      "step": 1734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9372123335609134e-05,
+      "loss": 1.0779,
+      "step": 1735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9371343161323042e-05,
+      "loss": 1.0729,
+      "step": 1736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9370562518361087e-05,
+      "loss": 0.9462,
+      "step": 1737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936978140676231e-05,
+      "loss": 1.0229,
+      "step": 1738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9368999826565776e-05,
+      "loss": 0.9913,
+      "step": 1739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936821777781057e-05,
+      "loss": 0.9643,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936743526053581e-05,
+      "loss": 1.0225,
+      "step": 1741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936665227478063e-05,
+      "loss": 1.0363,
+      "step": 1742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936586882058418e-05,
+      "loss": 1.0427,
+      "step": 1743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936508489798565e-05,
+      "loss": 1.0017,
+      "step": 1744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9364300507024243e-05,
+      "loss": 1.0062,
+      "step": 1745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936351564773919e-05,
+      "loss": 0.9287,
+      "step": 1746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9362730320169735e-05,
+      "loss": 1.0987,
+      "step": 1747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9361944524355162e-05,
+      "loss": 0.9196,
+      "step": 1748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.936115826033477e-05,
+      "loss": 0.9238,
+      "step": 1749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9360371528147874e-05,
+      "loss": 0.9918,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9359584327833826e-05,
+      "loss": 1.0173,
+      "step": 1751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9358796659431993e-05,
+      "loss": 1.0005,
+      "step": 1752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9358008522981766e-05,
+      "loss": 1.0398,
+      "step": 1753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9357219918522562e-05,
+      "loss": 0.9403,
+      "step": 1754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9356430846093823e-05,
+      "loss": 1.0676,
+      "step": 1755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9355641305735008e-05,
+      "loss": 1.0337,
+      "step": 1756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9354851297485607e-05,
+      "loss": 0.9662,
+      "step": 1757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935406082138513e-05,
+      "loss": 1.0719,
+      "step": 1758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9353269877473104e-05,
+      "loss": 1.1447,
+      "step": 1759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9352478465789087e-05,
+      "loss": 1.0853,
+      "step": 1760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9351686586372667e-05,
+      "loss": 1.0952,
+      "step": 1761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350894239263434e-05,
+      "loss": 1.0872,
+      "step": 1762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350101424501026e-05,
+      "loss": 1.0219,
+      "step": 1763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349308142125088e-05,
+      "loss": 1.0047,
+      "step": 1764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9348514392175293e-05,
+      "loss": 1.0164,
+      "step": 1765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9347720174691336e-05,
+      "loss": 0.7104,
+      "step": 1766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934692548971294e-05,
+      "loss": 1.0922,
+      "step": 1767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9346130337279852e-05,
+      "loss": 0.961,
+      "step": 1768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934533471743183e-05,
+      "loss": 1.1453,
+      "step": 1769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934453863020867e-05,
+      "loss": 0.9975,
+      "step": 1770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343742075650187e-05,
+      "loss": 0.9868,
+      "step": 1771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9342945053796215e-05,
+      "loss": 1.1818,
+      "step": 1772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934214756468661e-05,
+      "loss": 0.9798,
+      "step": 1773
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9341349608361267e-05,
+      "loss": 0.9711,
+      "step": 1774
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.934055118486008e-05,
+      "loss": 1.0576,
+      "step": 1775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9339752294222992e-05,
+      "loss": 0.7024,
+      "step": 1776
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9338952936489948e-05,
+      "loss": 1.0392,
+      "step": 1777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.933815311170093e-05,
+      "loss": 0.997,
+      "step": 1778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9337352819895937e-05,
+      "loss": 1.0935,
+      "step": 1779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9336552061114988e-05,
+      "loss": 1.0375,
+      "step": 1780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.933575083539814e-05,
+      "loss": 1.0969,
+      "step": 1781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9334949142785453e-05,
+      "loss": 1.1597,
+      "step": 1782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.933414698331703e-05,
+      "loss": 1.0016,
+      "step": 1783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.933334435703298e-05,
+      "loss": 0.715,
+      "step": 1784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9332541263973455e-05,
+      "loss": 0.8857,
+      "step": 1785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.933173770417861e-05,
+      "loss": 1.0041,
+      "step": 1786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.933093367768863e-05,
+      "loss": 1.0688,
+      "step": 1787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9330129184543735e-05,
+      "loss": 1.0401,
+      "step": 1788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9329324224784147e-05,
+      "loss": 0.9864,
+      "step": 1789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9328518798450138e-05,
+      "loss": 0.9787,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9327712905581976e-05,
+      "loss": 1.0422,
+      "step": 1791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9326906546219973e-05,
+      "loss": 1.109,
+      "step": 1792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.932609972040445e-05,
+      "loss": 1.0849,
+      "step": 1793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9325292428175762e-05,
+      "loss": 1.1474,
+      "step": 1794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9324484669574284e-05,
+      "loss": 1.0372,
+      "step": 1795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9323676444640412e-05,
+      "loss": 1.0812,
+      "step": 1796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9322867753414564e-05,
+      "loss": 0.9457,
+      "step": 1797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9322058595937183e-05,
+      "loss": 0.8793,
+      "step": 1798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9321248972248742e-05,
+      "loss": 1.0594,
+      "step": 1799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9320438882389728e-05,
+      "loss": 1.0785,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9319628326400657e-05,
+      "loss": 0.9053,
+      "step": 1801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.931881730432206e-05,
+      "loss": 1.0596,
+      "step": 1802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9318005816194508e-05,
+      "loss": 1.0818,
+      "step": 1803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.931719386205858e-05,
+      "loss": 0.9887,
+      "step": 1804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9316381441954878e-05,
+      "loss": 1.0541,
+      "step": 1805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9315568555924035e-05,
+      "loss": 1.0981,
+      "step": 1806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.931475520400671e-05,
+      "loss": 1.0318,
+      "step": 1807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9313941386243575e-05,
+      "loss": 0.9174,
+      "step": 1808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9313127102675332e-05,
+      "loss": 0.9799,
+      "step": 1809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9312312353342705e-05,
+      "loss": 1.0555,
+      "step": 1810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9311497138286438e-05,
+      "loss": 0.7046,
+      "step": 1811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9310681457547306e-05,
+      "loss": 0.997,
+      "step": 1812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.93098653111661e-05,
+      "loss": 1.0552,
+      "step": 1813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9309048699183633e-05,
+      "loss": 0.9899,
+      "step": 1814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930823162164075e-05,
+      "loss": 0.9984,
+      "step": 1815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9307414078578314e-05,
+      "loss": 1.0835,
+      "step": 1816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9306596070037212e-05,
+      "loss": 0.9039,
+      "step": 1817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930577759605835e-05,
+      "loss": 1.045,
+      "step": 1818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9304958656682666e-05,
+      "loss": 1.0222,
+      "step": 1819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9304139251951116e-05,
+      "loss": 0.9651,
+      "step": 1820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9303319381904676e-05,
+      "loss": 1.0103,
+      "step": 1821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.930249904658435e-05,
+      "loss": 1.0012,
+      "step": 1822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9301678246031167e-05,
+      "loss": 1.0254,
+      "step": 1823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9300856980286175e-05,
+      "loss": 1.0261,
+      "step": 1824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9300035249390444e-05,
+      "loss": 0.9328,
+      "step": 1825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9299213053385077e-05,
+      "loss": 1.0624,
+      "step": 1826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9298390392311184e-05,
+      "loss": 1.0807,
+      "step": 1827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929756726620992e-05,
+      "loss": 1.039,
+      "step": 1828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9296743675122436e-05,
+      "loss": 1.0409,
+      "step": 1829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9295919619089935e-05,
+      "loss": 1.0742,
+      "step": 1830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9295095098153616e-05,
+      "loss": 0.7285,
+      "step": 1831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9294270112354726e-05,
+      "loss": 0.9637,
+      "step": 1832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929344466173452e-05,
+      "loss": 1.0638,
+      "step": 1833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929261874633428e-05,
+      "loss": 1.0746,
+      "step": 1834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9291792366195306e-05,
+      "loss": 0.9986,
+      "step": 1835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929096552135894e-05,
+      "loss": 0.9517,
+      "step": 1836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.929013821186652e-05,
+      "loss": 0.91,
+      "step": 1837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9289310437759427e-05,
+      "loss": 1.0099,
+      "step": 1838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928848219907906e-05,
+      "loss": 0.9916,
+      "step": 1839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928765349586684e-05,
+      "loss": 1.0961,
+      "step": 1840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9286824328164208e-05,
+      "loss": 1.0269,
+      "step": 1841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9285994696012636e-05,
+      "loss": 0.7372,
+      "step": 1842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9285164599453617e-05,
+      "loss": 0.9678,
+      "step": 1843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928433403852866e-05,
+      "loss": 1.1129,
+      "step": 1844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.928350301327931e-05,
+      "loss": 0.7095,
+      "step": 1845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9282671523747117e-05,
+      "loss": 1.0146,
+      "step": 1846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9281839569973673e-05,
+      "loss": 0.9381,
+      "step": 1847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9281007152000587e-05,
+      "loss": 1.0301,
+      "step": 1848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9280174269869482e-05,
+      "loss": 0.9987,
+      "step": 1849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9279340923622022e-05,
+      "loss": 0.6915,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9278507113299873e-05,
+      "loss": 1.0041,
+      "step": 1851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927767283894474e-05,
+      "loss": 0.9841,
+      "step": 1852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9276838100598355e-05,
+      "loss": 1.081,
+      "step": 1853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9276002898302447e-05,
+      "loss": 1.0773,
+      "step": 1854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92751672320988e-05,
+      "loss": 1.0089,
+      "step": 1855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92743311020292e-05,
+      "loss": 1.0407,
+      "step": 1856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9273494508135466e-05,
+      "loss": 0.9823,
+      "step": 1857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9272657450459438e-05,
+      "loss": 1.0666,
+      "step": 1858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9271819929042975e-05,
+      "loss": 0.7004,
+      "step": 1859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9270981943927967e-05,
+      "loss": 1.0446,
+      "step": 1860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.927014349515632e-05,
+      "loss": 1.0572,
+      "step": 1861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9269304582769967e-05,
+      "loss": 1.0615,
+      "step": 1862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9268465206810864e-05,
+      "loss": 1.0045,
+      "step": 1863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926762536732099e-05,
+      "loss": 0.9982,
+      "step": 1864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9266785064342346e-05,
+      "loss": 0.9782,
+      "step": 1865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9265944297916953e-05,
+      "loss": 1.1449,
+      "step": 1866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926510306808687e-05,
+      "loss": 0.945,
+      "step": 1867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9264261374894154e-05,
+      "loss": 0.9833,
+      "step": 1868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926341921838091e-05,
+      "loss": 1.0444,
+      "step": 1869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926257659858925e-05,
+      "loss": 1.079,
+      "step": 1870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926173351556132e-05,
+      "loss": 1.1351,
+      "step": 1871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.926088996933928e-05,
+      "loss": 0.9643,
+      "step": 1872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9260045959965317e-05,
+      "loss": 1.0358,
+      "step": 1873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9259201487481643e-05,
+      "loss": 1.0539,
+      "step": 1874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9258356551930492e-05,
+      "loss": 0.9474,
+      "step": 1875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9257511153354112e-05,
+      "loss": 0.9365,
+      "step": 1876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9256665291794796e-05,
+      "loss": 1.0969,
+      "step": 1877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925581896729484e-05,
+      "loss": 1.0066,
+      "step": 1878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9254972179896572e-05,
+      "loss": 0.9938,
+      "step": 1879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9254124929642337e-05,
+      "loss": 1.0304,
+      "step": 1880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925327721657451e-05,
+      "loss": 1.0154,
+      "step": 1881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925242904073549e-05,
+      "loss": 1.0139,
+      "step": 1882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925158040216769e-05,
+      "loss": 1.0727,
+      "step": 1883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.925073130091356e-05,
+      "loss": 1.0052,
+      "step": 1884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9249881737015553e-05,
+      "loss": 0.9625,
+      "step": 1885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9249031710516162e-05,
+      "loss": 1.0536,
+      "step": 1886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9248181221457904e-05,
+      "loss": 1.1006,
+      "step": 1887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9247330269883308e-05,
+      "loss": 1.115,
+      "step": 1888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924647885583493e-05,
+      "loss": 1.0303,
+      "step": 1889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9245626979355352e-05,
+      "loss": 1.0064,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9244774640487183e-05,
+      "loss": 1.0699,
+      "step": 1891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9243921839273043e-05,
+      "loss": 0.9693,
+      "step": 1892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924306857575558e-05,
+      "loss": 1.1479,
+      "step": 1893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9242214849977476e-05,
+      "loss": 0.9828,
+      "step": 1894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924136066198142e-05,
+      "loss": 1.027,
+      "step": 1895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.924050601181014e-05,
+      "loss": 1.0986,
+      "step": 1896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9239650899506362e-05,
+      "loss": 1.0,
+      "step": 1897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 1.022,
+      "step": 1898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923793928867244e-05,
+      "loss": 0.9862,
+      "step": 1899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923708279022789e-05,
+      "loss": 1.1206,
+      "step": 1900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9236225829822053e-05,
+      "loss": 1.0788,
+      "step": 1901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9235368407497788e-05,
+      "loss": 1.0393,
+      "step": 1902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9234510523297972e-05,
+      "loss": 1.0725,
+      "step": 1903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9233652177265514e-05,
+      "loss": 1.073,
+      "step": 1904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232793369443338e-05,
+      "loss": 1.0472,
+      "step": 1905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9231934099874398e-05,
+      "loss": 0.7498,
+      "step": 1906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9231074368601666e-05,
+      "loss": 1.0701,
+      "step": 1907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230214175668137e-05,
+      "loss": 1.062,
+      "step": 1908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229353521116828e-05,
+      "loss": 1.1097,
+      "step": 1909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9228492404990788e-05,
+      "loss": 1.0954,
+      "step": 1910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9227630827333083e-05,
+      "loss": 0.9907,
+      "step": 1911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9226768788186793e-05,
+      "loss": 1.0249,
+      "step": 1912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9225906287595036e-05,
+      "loss": 1.0839,
+      "step": 1913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9225043325600953e-05,
+      "loss": 1.0824,
+      "step": 1914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922417990224769e-05,
+      "loss": 1.0257,
+      "step": 1915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9223316017578434e-05,
+      "loss": 1.0652,
+      "step": 1916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9222451671636392e-05,
+      "loss": 1.1591,
+      "step": 1917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9221586864464788e-05,
+      "loss": 1.0436,
+      "step": 1918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.922072159610687e-05,
+      "loss": 1.1822,
+      "step": 1919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.921985586660592e-05,
+      "loss": 1.1541,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9218989676005225e-05,
+      "loss": 1.0032,
+      "step": 1921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9218123024348112e-05,
+      "loss": 1.1192,
+      "step": 1922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9217255911677916e-05,
+      "loss": 1.0712,
+      "step": 1923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.921638833803801e-05,
+      "loss": 0.8966,
+      "step": 1924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.921552030347178e-05,
+      "loss": 1.1095,
+      "step": 1925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9214651808022635e-05,
+      "loss": 0.8732,
+      "step": 1926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9213782851734014e-05,
+      "loss": 1.0377,
+      "step": 1927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9212913434649372e-05,
+      "loss": 0.9722,
+      "step": 1928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.921204355681219e-05,
+      "loss": 0.9917,
+      "step": 1929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9211173218265977e-05,
+      "loss": 0.7469,
+      "step": 1930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9210302419054257e-05,
+      "loss": 1.0602,
+      "step": 1931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9209431159220575e-05,
+      "loss": 1.0442,
+      "step": 1932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.920855943880851e-05,
+      "loss": 1.0048,
+      "step": 1933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9207687257861654e-05,
+      "loss": 1.1142,
+      "step": 1934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9206814616423632e-05,
+      "loss": 0.9297,
+      "step": 1935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9205941514538083e-05,
+      "loss": 1.086,
+      "step": 1936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.920506795224867e-05,
+      "loss": 0.9905,
+      "step": 1937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9204193929599082e-05,
+      "loss": 0.9881,
+      "step": 1938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9203319446633034e-05,
+      "loss": 1.0371,
+      "step": 1939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9202444503394257e-05,
+      "loss": 1.0089,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.920156909992651e-05,
+      "loss": 1.0756,
+      "step": 1941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.920069323627357e-05,
+      "loss": 0.9909,
+      "step": 1942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9199816912479245e-05,
+      "loss": 1.0695,
+      "step": 1943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.919894012858736e-05,
+      "loss": 1.0626,
+      "step": 1944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.919806288464176e-05,
+      "loss": 0.6868,
+      "step": 1945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9197185180686326e-05,
+      "loss": 1.0899,
+      "step": 1946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9196307016764947e-05,
+      "loss": 1.0587,
+      "step": 1947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.919542839292154e-05,
+      "loss": 1.0431,
+      "step": 1948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9194549309200048e-05,
+      "loss": 0.9721,
+      "step": 1949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.919366976564444e-05,
+      "loss": 1.0166,
+      "step": 1950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91927897622987e-05,
+      "loss": 1.0137,
+      "step": 1951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9191909299206836e-05,
+      "loss": 1.1055,
+      "step": 1952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9191028376412883e-05,
+      "loss": 0.8987,
+      "step": 1953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91901469939609e-05,
+      "loss": 0.954,
+      "step": 1954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9189265151894963e-05,
+      "loss": 1.0535,
+      "step": 1955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9188382850259175e-05,
+      "loss": 1.1115,
+      "step": 1956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918750008909766e-05,
+      "loss": 1.0003,
+      "step": 1957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9186616868454574e-05,
+      "loss": 1.1412,
+      "step": 1958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918573318837408e-05,
+      "loss": 1.0147,
+      "step": 1959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9184849048900375e-05,
+      "loss": 1.005,
+      "step": 1960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.918396445007767e-05,
+      "loss": 0.9561,
+      "step": 1961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9183079391950217e-05,
+      "loss": 0.983,
+      "step": 1962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9182193874562275e-05,
+      "loss": 1.1507,
+      "step": 1963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9181307897958125e-05,
+      "loss": 0.8256,
+      "step": 1964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9180421462182082e-05,
+      "loss": 1.1234,
+      "step": 1965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9179534567278474e-05,
+      "loss": 1.0318,
+      "step": 1966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917864721329166e-05,
+      "loss": 1.0069,
+      "step": 1967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9177759400266015e-05,
+      "loss": 0.7108,
+      "step": 1968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9176871128245938e-05,
+      "loss": 1.0161,
+      "step": 1969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917598239727586e-05,
+      "loss": 0.9515,
+      "step": 1970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9175093207400218e-05,
+      "loss": 1.0174,
+      "step": 1971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.917420355866349e-05,
+      "loss": 1.0364,
+      "step": 1972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9173313451110165e-05,
+      "loss": 0.9571,
+      "step": 1973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9172422884784758e-05,
+      "loss": 0.9312,
+      "step": 1974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9171531859731813e-05,
+      "loss": 0.9826,
+      "step": 1975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9170640375995887e-05,
+      "loss": 1.1032,
+      "step": 1976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9169748433621565e-05,
+      "loss": 1.0661,
+      "step": 1977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9168856032653457e-05,
+      "loss": 1.0865,
+      "step": 1978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916796317313619e-05,
+      "loss": 1.0104,
+      "step": 1979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9167069855114414e-05,
+      "loss": 0.6836,
+      "step": 1980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9166176078632813e-05,
+      "loss": 0.9478,
+      "step": 1981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9165281843736084e-05,
+      "loss": 1.0523,
+      "step": 1982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916438715046895e-05,
+      "loss": 0.9202,
+      "step": 1983
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916349199887615e-05,
+      "loss": 1.109,
+      "step": 1984
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.916259638900246e-05,
+      "loss": 0.9824,
+      "step": 1985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9161700320892665e-05,
+      "loss": 1.0198,
+      "step": 1986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9160803794591582e-05,
+      "loss": 0.6898,
+      "step": 1987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9159906810144045e-05,
+      "loss": 1.0467,
+      "step": 1988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9159009367594917e-05,
+      "loss": 1.117,
+      "step": 1989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9158111466989076e-05,
+      "loss": 1.1116,
+      "step": 1990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9157213108371434e-05,
+      "loss": 1.0675,
+      "step": 1991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9156314291786912e-05,
+      "loss": 1.0144,
+      "step": 1992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9155415017280465e-05,
+      "loss": 1.0478,
+      "step": 1993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915451528489707e-05,
+      "loss": 1.1115,
+      "step": 1994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9153615094681722e-05,
+      "loss": 1.0201,
+      "step": 1995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9152714446679435e-05,
+      "loss": 1.0871,
+      "step": 1996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9151813340935258e-05,
+      "loss": 0.9796,
+      "step": 1997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915091177749426e-05,
+      "loss": 1.0346,
+      "step": 1998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.915000975640152e-05,
+      "loss": 0.998,
+      "step": 1999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9149107277702156e-05,
+      "loss": 1.0297,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9148204341441304e-05,
+      "loss": 1.1611,
+      "step": 2001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9147300947664116e-05,
+      "loss": 1.03,
+      "step": 2002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9146397096415774e-05,
+      "loss": 1.004,
+      "step": 2003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9145492787741484e-05,
+      "loss": 1.0116,
+      "step": 2004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9144588021686466e-05,
+      "loss": 1.0601,
+      "step": 2005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9143682798295974e-05,
+      "loss": 0.9654,
+      "step": 2006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9142777117615278e-05,
+      "loss": 1.0079,
+      "step": 2007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9141870979689672e-05,
+      "loss": 0.9502,
+      "step": 2008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9140964384564478e-05,
+      "loss": 1.1691,
+      "step": 2009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.914005733228503e-05,
+      "loss": 1.0035,
+      "step": 2010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913914982289669e-05,
+      "loss": 1.0187,
+      "step": 2011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9138241856444853e-05,
+      "loss": 0.9987,
+      "step": 2012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9137333432974915e-05,
+      "loss": 0.6978,
+      "step": 2013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913642455253232e-05,
+      "loss": 1.0284,
+      "step": 2014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9135515215162517e-05,
+      "loss": 1.0792,
+      "step": 2015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9134605420910985e-05,
+      "loss": 1.1098,
+      "step": 2016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913369516982322e-05,
+      "loss": 1.1212,
+      "step": 2017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913278446194475e-05,
+      "loss": 1.0506,
+      "step": 2018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.913187329732112e-05,
+      "loss": 1.0533,
+      "step": 2019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9130961675997897e-05,
+      "loss": 1.0574,
+      "step": 2020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9130049598020672e-05,
+      "loss": 1.0698,
+      "step": 2021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9129137063435067e-05,
+      "loss": 0.998,
+      "step": 2022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9128224072286706e-05,
+      "loss": 1.1309,
+      "step": 2023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912731062462126e-05,
+      "loss": 1.0511,
+      "step": 2024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9126396720484412e-05,
+      "loss": 1.0316,
+      "step": 2025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9125482359921863e-05,
+      "loss": 0.958,
+      "step": 2026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9124567542979337e-05,
+      "loss": 1.0371,
+      "step": 2027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91236522697026e-05,
+      "loss": 1.1221,
+      "step": 2028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912273654013741e-05,
+      "loss": 1.0015,
+      "step": 2029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9121820354329578e-05,
+      "loss": 1.0287,
+      "step": 2030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9120903712324913e-05,
+      "loss": 1.041,
+      "step": 2031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9119986614169263e-05,
+      "loss": 1.0948,
+      "step": 2032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9119069059908495e-05,
+      "loss": 1.0145,
+      "step": 2033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9118151049588496e-05,
+      "loss": 1.0154,
+      "step": 2034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911723258325518e-05,
+      "loss": 0.973,
+      "step": 2035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911631366095447e-05,
+      "loss": 1.0591,
+      "step": 2036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9115394282732334e-05,
+      "loss": 0.9774,
+      "step": 2037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911447444863475e-05,
+      "loss": 0.9943,
+      "step": 2038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9113554158707713e-05,
+      "loss": 0.9352,
+      "step": 2039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911263341299726e-05,
+      "loss": 0.7272,
+      "step": 2040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9111712211549428e-05,
+      "loss": 0.6937,
+      "step": 2041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9110790554410294e-05,
+      "loss": 1.1486,
+      "step": 2042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9109868441625947e-05,
+      "loss": 1.0282,
+      "step": 2043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108945873242507e-05,
+      "loss": 1.0425,
+      "step": 2044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108022849306117e-05,
+      "loss": 1.0566,
+      "step": 2045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910709936986293e-05,
+      "loss": 1.0858,
+      "step": 2046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106175434959135e-05,
+      "loss": 1.009,
+      "step": 2047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105251044640938e-05,
+      "loss": 1.0807,
+      "step": 2048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104326198954572e-05,
+      "loss": 1.0298,
+      "step": 2049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910340089794629e-05,
+      "loss": 1.0016,
+      "step": 2050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9102475141662365e-05,
+      "loss": 1.0546,
+      "step": 2051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9101548930149093e-05,
+      "loss": 0.9925,
+      "step": 2052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9100622263452805e-05,
+      "loss": 1.0578,
+      "step": 2053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9099695141619835e-05,
+      "loss": 0.9763,
+      "step": 2054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9098767564696558e-05,
+      "loss": 0.8858,
+      "step": 2055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909783953272936e-05,
+      "loss": 0.9797,
+      "step": 2056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9096911045764648e-05,
+      "loss": 1.1545,
+      "step": 2057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9095982103848864e-05,
+      "loss": 1.1111,
+      "step": 2058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.909505270702847e-05,
+      "loss": 1.0032,
+      "step": 2059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9094122855349934e-05,
+      "loss": 1.0298,
+      "step": 2060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9093192548859766e-05,
+      "loss": 1.0555,
+      "step": 2061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9092261787604492e-05,
+      "loss": 1.0373,
+      "step": 2062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9091330571630664e-05,
+      "loss": 0.6996,
+      "step": 2063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.909039890098485e-05,
+      "loss": 1.0414,
+      "step": 2064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.908946677571364e-05,
+      "loss": 1.0724,
+      "step": 2065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9088534195863665e-05,
+      "loss": 0.6761,
+      "step": 2066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9087601161481552e-05,
+      "loss": 0.9721,
+      "step": 2067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9086667672613967e-05,
+      "loss": 1.0216,
+      "step": 2068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9085733729307595e-05,
+      "loss": 0.975,
+      "step": 2069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9084799331609145e-05,
+      "loss": 1.0686,
+      "step": 2070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.908386447956535e-05,
+      "loss": 1.0703,
+      "step": 2071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9082929173222957e-05,
+      "loss": 1.0509,
+      "step": 2072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9081993412628745e-05,
+      "loss": 1.0755,
+      "step": 2073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9081057197829517e-05,
+      "loss": 1.0357,
+      "step": 2074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.908012052887209e-05,
+      "loss": 0.7509,
+      "step": 2075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9079183405803313e-05,
+      "loss": 0.9845,
+      "step": 2076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9078245828670045e-05,
+      "loss": 0.9817,
+      "step": 2077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9077307797519183e-05,
+      "loss": 1.0125,
+      "step": 2078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9076369312397638e-05,
+      "loss": 1.0925,
+      "step": 2079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9075430373352342e-05,
+      "loss": 0.9291,
+      "step": 2080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9074490980430256e-05,
+      "loss": 1.0078,
+      "step": 2081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.907355113367836e-05,
+      "loss": 1.0697,
+      "step": 2082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9072610833143655e-05,
+      "loss": 0.9419,
+      "step": 2083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.907167007887317e-05,
+      "loss": 1.0257,
+      "step": 2084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.907072887091395e-05,
+      "loss": 1.052,
+      "step": 2085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9069787209313072e-05,
+      "loss": 1.0247,
+      "step": 2086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9068845094117625e-05,
+      "loss": 1.0815,
+      "step": 2087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.906790252537473e-05,
+      "loss": 1.1158,
+      "step": 2088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.906695950313152e-05,
+      "loss": 0.9854,
+      "step": 2089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9066016027435164e-05,
+      "loss": 1.0186,
+      "step": 2090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.906507209833284e-05,
+      "loss": 0.9783,
+      "step": 2091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9064127715871762e-05,
+      "loss": 1.0742,
+      "step": 2092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9063182880099153e-05,
+      "loss": 1.1216,
+      "step": 2093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.906223759106227e-05,
+      "loss": 0.9973,
+      "step": 2094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.906129184880839e-05,
+      "loss": 1.0955,
+      "step": 2095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9060345653384814e-05,
+      "loss": 1.0305,
+      "step": 2096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905939900483885e-05,
+      "loss": 1.2189,
+      "step": 2097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9058451903217855e-05,
+      "loss": 0.942,
+      "step": 2098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905750434856919e-05,
+      "loss": 1.0532,
+      "step": 2099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9056556340940234e-05,
+      "loss": 1.0885,
+      "step": 2100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905560788037842e-05,
+      "loss": 1.0323,
+      "step": 2101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905465896693116e-05,
+      "loss": 1.0675,
+      "step": 2102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9053709600645925e-05,
+      "loss": 1.023,
+      "step": 2103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.905275978157019e-05,
+      "loss": 0.9724,
+      "step": 2104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9051809509751452e-05,
+      "loss": 1.0617,
+      "step": 2105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9050858785237244e-05,
+      "loss": 0.9992,
+      "step": 2106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9049907608075108e-05,
+      "loss": 1.1387,
+      "step": 2107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9048955978312612e-05,
+      "loss": 1.1183,
+      "step": 2108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9048003895997357e-05,
+      "loss": 1.0291,
+      "step": 2109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9047051361176955e-05,
+      "loss": 0.9652,
+      "step": 2110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9046098373899037e-05,
+      "loss": 1.0038,
+      "step": 2111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9045144934211268e-05,
+      "loss": 1.0476,
+      "step": 2112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.904419104216133e-05,
+      "loss": 0.9565,
+      "step": 2113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9043236697796936e-05,
+      "loss": 1.0452,
+      "step": 2114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9042281901165803e-05,
+      "loss": 1.0252,
+      "step": 2115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9041326652315686e-05,
+      "loss": 0.9612,
+      "step": 2116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9040370951294364e-05,
+      "loss": 0.6479,
+      "step": 2117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9039414798149627e-05,
+      "loss": 1.0256,
+      "step": 2118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9038458192929297e-05,
+      "loss": 0.9961,
+      "step": 2119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9037501135681215e-05,
+      "loss": 1.0542,
+      "step": 2120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9036543626453236e-05,
+      "loss": 0.9676,
+      "step": 2121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903558566529326e-05,
+      "loss": 1.0066,
+      "step": 2122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9034627252249193e-05,
+      "loss": 0.9517,
+      "step": 2123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9033668387368963e-05,
+      "loss": 1.0508,
+      "step": 2124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903270907070052e-05,
+      "loss": 1.0792,
+      "step": 2125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9031749302291853e-05,
+      "loss": 0.9489,
+      "step": 2126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.903078908219095e-05,
+      "loss": 0.9483,
+      "step": 2127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9029828410445846e-05,
+      "loss": 1.0304,
+      "step": 2128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9028867287104572e-05,
+      "loss": 0.9912,
+      "step": 2129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9027905712215204e-05,
+      "loss": 1.041,
+      "step": 2130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9026943685825826e-05,
+      "loss": 0.9573,
+      "step": 2131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9025981207984558e-05,
+      "loss": 1.1537,
+      "step": 2132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9025018278739526e-05,
+      "loss": 1.0353,
+      "step": 2133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9024054898138894e-05,
+      "loss": 1.0194,
+      "step": 2134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9023091066230844e-05,
+      "loss": 0.6825,
+      "step": 2135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9022126783063576e-05,
+      "loss": 1.1526,
+      "step": 2136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9021162048685312e-05,
+      "loss": 1.078,
+      "step": 2137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.90201968631443e-05,
+      "loss": 1.0933,
+      "step": 2138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9019231226488818e-05,
+      "loss": 1.0148,
+      "step": 2139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9018265138767153e-05,
+      "loss": 0.9441,
+      "step": 2140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901729860002762e-05,
+      "loss": 1.0976,
+      "step": 2141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901633161031856e-05,
+      "loss": 1.1024,
+      "step": 2142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9015364169688332e-05,
+      "loss": 1.0675,
+      "step": 2143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.901439627818532e-05,
+      "loss": 1.0498,
+      "step": 2144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9013427935857932e-05,
+      "loss": 0.9368,
+      "step": 2145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9012459142754592e-05,
+      "loss": 1.052,
+      "step": 2146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9011489898923753e-05,
+      "loss": 1.1,
+      "step": 2147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9010520204413887e-05,
+      "loss": 1.0182,
+      "step": 2148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9009550059273494e-05,
+      "loss": 1.0421,
+      "step": 2149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9008579463551088e-05,
+      "loss": 0.9781,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9007608417295212e-05,
+      "loss": 1.0287,
+      "step": 2151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9006636920554432e-05,
+      "loss": 1.0918,
+      "step": 2152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900566497337733e-05,
+      "loss": 1.0162,
+      "step": 2153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9004692575812516e-05,
+      "loss": 1.0319,
+      "step": 2154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.900371972790862e-05,
+      "loss": 0.9903,
+      "step": 2155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.90027464297143e-05,
+      "loss": 0.9872,
+      "step": 2156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9001772681278227e-05,
+      "loss": 0.9619,
+      "step": 2157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9000798482649105e-05,
+      "loss": 1.0439,
+      "step": 2158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899982383387565e-05,
+      "loss": 1.0943,
+      "step": 2159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8998848735006606e-05,
+      "loss": 0.978,
+      "step": 2160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8997873186090742e-05,
+      "loss": 1.0485,
+      "step": 2161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899689718717685e-05,
+      "loss": 1.0929,
+      "step": 2162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8995920738313734e-05,
+      "loss": 1.1563,
+      "step": 2163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8994943839550232e-05,
+      "loss": 0.9777,
+      "step": 2164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89939664909352e-05,
+      "loss": 1.0689,
+      "step": 2165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8992988692517512e-05,
+      "loss": 0.7159,
+      "step": 2166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8992010444346075e-05,
+      "loss": 1.0594,
+      "step": 2167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899103174646981e-05,
+      "loss": 0.925,
+      "step": 2168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899005259893767e-05,
+      "loss": 1.0208,
+      "step": 2169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8989073001798614e-05,
+      "loss": 1.0,
+      "step": 2170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8988092955101636e-05,
+      "loss": 1.0015,
+      "step": 2171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898711245889575e-05,
+      "loss": 1.0532,
+      "step": 2172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8986131513229998e-05,
+      "loss": 0.9451,
+      "step": 2173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898515011815343e-05,
+      "loss": 1.113,
+      "step": 2174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8984168273715135e-05,
+      "loss": 1.1243,
+      "step": 2175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.898318597996421e-05,
+      "loss": 1.0274,
+      "step": 2176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8982203236949784e-05,
+      "loss": 0.9803,
+      "step": 2177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8981220044721004e-05,
+      "loss": 0.667,
+      "step": 2178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8980236403327047e-05,
+      "loss": 0.9375,
+      "step": 2179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8979252312817096e-05,
+      "loss": 0.9612,
+      "step": 2180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8978267773240376e-05,
+      "loss": 0.9524,
+      "step": 2181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977282784646117e-05,
+      "loss": 1.0638,
+      "step": 2182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897629734708359e-05,
+      "loss": 0.6536,
+      "step": 2183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975311460602074e-05,
+      "loss": 1.0138,
+      "step": 2184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897432512525087e-05,
+      "loss": 0.6906,
+      "step": 2185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897333834107931e-05,
+      "loss": 1.0235,
+      "step": 2186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897235110813675e-05,
+      "loss": 1.0217,
+      "step": 2187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8971363426472552e-05,
+      "loss": 1.0649,
+      "step": 2188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897037529613612e-05,
+      "loss": 1.0232,
+      "step": 2189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896938671717687e-05,
+      "loss": 0.8669,
+      "step": 2190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968397689644242e-05,
+      "loss": 1.0447,
+      "step": 2191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89674082135877e-05,
+      "loss": 1.0789,
+      "step": 2192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8966418289056723e-05,
+      "loss": 0.9572,
+      "step": 2193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8965427916100827e-05,
+      "loss": 1.1317,
+      "step": 2194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8964437094769538e-05,
+      "loss": 1.0732,
+      "step": 2195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896344582511241e-05,
+      "loss": 0.9223,
+      "step": 2196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8962454107179017e-05,
+      "loss": 0.9495,
+      "step": 2197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896146194101896e-05,
+      "loss": 1.0155,
+      "step": 2198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8960469326681852e-05,
+      "loss": 0.9915,
+      "step": 2199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8959476264217337e-05,
+      "loss": 0.99,
+      "step": 2200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8958482753675086e-05,
+      "loss": 0.9604,
+      "step": 2201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8957488795104776e-05,
+      "loss": 0.9711,
+      "step": 2202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8956494388556125e-05,
+      "loss": 1.0962,
+      "step": 2203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8955499534078867e-05,
+      "loss": 0.8663,
+      "step": 2204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8954504231722743e-05,
+      "loss": 1.0067,
+      "step": 2205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8953508481537542e-05,
+      "loss": 0.9492,
+      "step": 2206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8952512283573056e-05,
+      "loss": 1.0604,
+      "step": 2207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8951515637879114e-05,
+      "loss": 0.9609,
+      "step": 2208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.895051854450555e-05,
+      "loss": 1.0951,
+      "step": 2209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8949521003502237e-05,
+      "loss": 0.9893,
+      "step": 2210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8948523014919065e-05,
+      "loss": 1.0243,
+      "step": 2211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.894752457880594e-05,
+      "loss": 1.041,
+      "step": 2212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8946525695212795e-05,
+      "loss": 1.0265,
+      "step": 2213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.894552636418959e-05,
+      "loss": 1.0154,
+      "step": 2214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89445265857863e-05,
+      "loss": 1.0452,
+      "step": 2215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.894352636005293e-05,
+      "loss": 1.1185,
+      "step": 2216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8942525687039498e-05,
+      "loss": 0.9964,
+      "step": 2217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.894152456679605e-05,
+      "loss": 0.9099,
+      "step": 2218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8940522999372656e-05,
+      "loss": 1.0373,
+      "step": 2219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8939520984819402e-05,
+      "loss": 0.975,
+      "step": 2220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8938518523186408e-05,
+      "loss": 0.9902,
+      "step": 2221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89375156145238e-05,
+      "loss": 0.9449,
+      "step": 2222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.893651225888174e-05,
+      "loss": 1.0109,
+      "step": 2223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89355084563104e-05,
+      "loss": 0.9532,
+      "step": 2224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8934504206859995e-05,
+      "loss": 1.0626,
+      "step": 2225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8933499510580742e-05,
+      "loss": 1.0276,
+      "step": 2226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8932494367522885e-05,
+      "loss": 1.0346,
+      "step": 2227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8931488777736694e-05,
+      "loss": 1.053,
+      "step": 2228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8930482741272466e-05,
+      "loss": 1.0794,
+      "step": 2229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8929476258180504e-05,
+      "loss": 1.0755,
+      "step": 2230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8928469328511156e-05,
+      "loss": 0.9516,
+      "step": 2231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.892746195231477e-05,
+      "loss": 0.9977,
+      "step": 2232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.892645412964173e-05,
+      "loss": 1.0766,
+      "step": 2233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.892544586054244e-05,
+      "loss": 0.9751,
+      "step": 2234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8924437145067327e-05,
+      "loss": 0.9011,
+      "step": 2235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8923427983266833e-05,
+      "loss": 1.0929,
+      "step": 2236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.892241837519143e-05,
+      "loss": 1.0318,
+      "step": 2237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8921408320891615e-05,
+      "loss": 0.739,
+      "step": 2238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8920397820417892e-05,
+      "loss": 1.0101,
+      "step": 2239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891938687382081e-05,
+      "loss": 1.0949,
+      "step": 2240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891837548115092e-05,
+      "loss": 1.1152,
+      "step": 2241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8917363642458805e-05,
+      "loss": 1.026,
+      "step": 2242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891635135779507e-05,
+      "loss": 0.9628,
+      "step": 2243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8915338627210336e-05,
+      "loss": 1.0055,
+      "step": 2244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891432545075526e-05,
+      "loss": 1.0887,
+      "step": 2245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8913311828480504e-05,
+      "loss": 1.2021,
+      "step": 2246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8912297760436768e-05,
+      "loss": 0.6781,
+      "step": 2247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.891128324667476e-05,
+      "loss": 0.8218,
+      "step": 2248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8910268287245227e-05,
+      "loss": 1.0989,
+      "step": 2249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890925288219892e-05,
+      "loss": 0.9033,
+      "step": 2250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8908237031586622e-05,
+      "loss": 1.0491,
+      "step": 2251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890722073545914e-05,
+      "loss": 1.0728,
+      "step": 2252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.89062039938673e-05,
+      "loss": 1.0333,
+      "step": 2253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8905186806861957e-05,
+      "loss": 1.1023,
+      "step": 2254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890416917449397e-05,
+      "loss": 0.9448,
+      "step": 2255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890315109681424e-05,
+      "loss": 1.0713,
+      "step": 2256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8902132573873682e-05,
+      "loss": 1.0836,
+      "step": 2257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8901113605723235e-05,
+      "loss": 1.0111,
+      "step": 2258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.890009419241385e-05,
+      "loss": 1.0098,
+      "step": 2259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8899074333996525e-05,
+      "loss": 1.0094,
+      "step": 2260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889805403052225e-05,
+      "loss": 0.7296,
+      "step": 2261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8897033282042065e-05,
+      "loss": 1.0624,
+      "step": 2262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8896012088607008e-05,
+      "loss": 1.0564,
+      "step": 2263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8894990450268156e-05,
+      "loss": 0.9967,
+      "step": 2264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8893968367076606e-05,
+      "loss": 1.1082,
+      "step": 2265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8892945839083468e-05,
+      "loss": 1.0637,
+      "step": 2266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8891922866339882e-05,
+      "loss": 0.9947,
+      "step": 2267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.889089944889701e-05,
+      "loss": 1.0613,
+      "step": 2268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.888987558680603e-05,
+      "loss": 1.0997,
+      "step": 2269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8888851280118153e-05,
+      "loss": 0.9964,
+      "step": 2270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8887826528884604e-05,
+      "loss": 1.0242,
+      "step": 2271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8886801333156635e-05,
+      "loss": 1.0465,
+      "step": 2272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8885775692985514e-05,
+      "loss": 1.0744,
+      "step": 2273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8884749608422532e-05,
+      "loss": 0.9925,
+      "step": 2274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8883723079519012e-05,
+      "loss": 1.0202,
+      "step": 2275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8882696106326288e-05,
+      "loss": 1.0332,
+      "step": 2276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8881668688895726e-05,
+      "loss": 0.9603,
+      "step": 2277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88806408272787e-05,
+      "loss": 1.0442,
+      "step": 2278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8879612521526623e-05,
+      "loss": 0.9659,
+      "step": 2279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8878583771690918e-05,
+      "loss": 1.0558,
+      "step": 2280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887755457782303e-05,
+      "loss": 0.8781,
+      "step": 2281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8876524939974444e-05,
+      "loss": 1.0024,
+      "step": 2282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887549485819664e-05,
+      "loss": 0.6741,
+      "step": 2283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8874464332541146e-05,
+      "loss": 1.0569,
+      "step": 2284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8873433363059492e-05,
+      "loss": 1.0024,
+      "step": 2285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887240194980324e-05,
+      "loss": 0.9617,
+      "step": 2286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.887137009282397e-05,
+      "loss": 1.0746,
+      "step": 2287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8870337792173292e-05,
+      "loss": 1.1077,
+      "step": 2288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8869305047902836e-05,
+      "loss": 1.0774,
+      "step": 2289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886827186006424e-05,
+      "loss": 1.0893,
+      "step": 2290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8867238228709184e-05,
+      "loss": 1.0001,
+      "step": 2291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886620415388936e-05,
+      "loss": 0.9964,
+      "step": 2292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886516963565648e-05,
+      "loss": 0.6709,
+      "step": 2293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8864134674062288e-05,
+      "loss": 1.03,
+      "step": 2294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.886309926915854e-05,
+      "loss": 0.9721,
+      "step": 2295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8862063420997017e-05,
+      "loss": 1.0275,
+      "step": 2296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8861027129629524e-05,
+      "loss": 0.9877,
+      "step": 2297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8859990395107895e-05,
+      "loss": 1.0656,
+      "step": 2298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885895321748397e-05,
+      "loss": 1.0291,
+      "step": 2299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8857915596809625e-05,
+      "loss": 0.9775,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885687753313675e-05,
+      "loss": 0.6529,
+      "step": 2301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8855839026517257e-05,
+      "loss": 1.1799,
+      "step": 2302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8854800077003092e-05,
+      "loss": 1.0887,
+      "step": 2303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8853760684646207e-05,
+      "loss": 0.9614,
+      "step": 2304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885272084949859e-05,
+      "loss": 0.9836,
+      "step": 2305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8851680571612242e-05,
+      "loss": 1.0069,
+      "step": 2306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.885063985103918e-05,
+      "loss": 1.0089,
+      "step": 2307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8849598687831473e-05,
+      "loss": 0.737,
+      "step": 2308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8848557082041172e-05,
+      "loss": 1.0279,
+      "step": 2309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8847515033720375e-05,
+      "loss": 0.6525,
+      "step": 2310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88464725429212e-05,
+      "loss": 1.0797,
+      "step": 2311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8845429609695783e-05,
+      "loss": 1.2032,
+      "step": 2312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8844386234096278e-05,
+      "loss": 0.9107,
+      "step": 2313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884334241617487e-05,
+      "loss": 0.9358,
+      "step": 2314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884229815598376e-05,
+      "loss": 1.0937,
+      "step": 2315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.884125345357518e-05,
+      "loss": 1.0807,
+      "step": 2316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8840208309001365e-05,
+      "loss": 0.6683,
+      "step": 2317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883916272231459e-05,
+      "loss": 1.1145,
+      "step": 2318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883811669356715e-05,
+      "loss": 1.0194,
+      "step": 2319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8837070222811357e-05,
+      "loss": 1.0832,
+      "step": 2320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883602331009954e-05,
+      "loss": 1.0282,
+      "step": 2321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883497595548407e-05,
+      "loss": 0.9184,
+      "step": 2322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8833928159017313e-05,
+      "loss": 1.063,
+      "step": 2323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832879920751678e-05,
+      "loss": 1.0541,
+      "step": 2324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883183124073959e-05,
+      "loss": 0.9161,
+      "step": 2325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830782119033493e-05,
+      "loss": 0.9922,
+      "step": 2326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8829732555685853e-05,
+      "loss": 1.0682,
+      "step": 2327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8828682550749166e-05,
+      "loss": 1.0975,
+      "step": 2328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882763210427594e-05,
+      "loss": 1.0293,
+      "step": 2329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882658121631871e-05,
+      "loss": 1.0016,
+      "step": 2330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8825529886930036e-05,
+      "loss": 0.9602,
+      "step": 2331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882447811616249e-05,
+      "loss": 1.1194,
+      "step": 2332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823425904068678e-05,
+      "loss": 1.1208,
+      "step": 2333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8822373250701224e-05,
+      "loss": 1.0297,
+      "step": 2334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882132015611277e-05,
+      "loss": 1.1065,
+      "step": 2335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8820266620355984e-05,
+      "loss": 0.9999,
+      "step": 2336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 0.9799,
+      "step": 2337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8818158225548188e-05,
+      "loss": 0.9434,
+      "step": 2338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8817103366602624e-05,
+      "loss": 0.9948,
+      "step": 2339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881604806669962e-05,
+      "loss": 0.9658,
+      "step": 2340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8814992325891945e-05,
+      "loss": 1.1034,
+      "step": 2341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8813936144232402e-05,
+      "loss": 1.0975,
+      "step": 2342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8812879521773813e-05,
+      "loss": 0.9437,
+      "step": 2343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8811822458569023e-05,
+      "loss": 1.0273,
+      "step": 2344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8810764954670893e-05,
+      "loss": 1.0423,
+      "step": 2345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8809707010132316e-05,
+      "loss": 1.1468,
+      "step": 2346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8808648625006197e-05,
+      "loss": 1.0142,
+      "step": 2347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8807589799345473e-05,
+      "loss": 1.071,
+      "step": 2348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.880653053320309e-05,
+      "loss": 0.886,
+      "step": 2349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8805470826632023e-05,
+      "loss": 1.0214,
+      "step": 2350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8804410679685278e-05,
+      "loss": 1.0526,
+      "step": 2351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8803350092415872e-05,
+      "loss": 1.063,
+      "step": 2352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8802289064876843e-05,
+      "loss": 1.0217,
+      "step": 2353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8801227597121256e-05,
+      "loss": 1.1403,
+      "step": 2354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.88001656892022e-05,
+      "loss": 0.9817,
+      "step": 2355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8799103341172778e-05,
+      "loss": 0.958,
+      "step": 2356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8798040553086122e-05,
+      "loss": 1.0208,
+      "step": 2357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.879697732499538e-05,
+      "loss": 0.9844,
+      "step": 2358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8795913656953736e-05,
+      "loss": 0.9592,
+      "step": 2359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8794849549014375e-05,
+      "loss": 1.0652,
+      "step": 2360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.879378500123052e-05,
+      "loss": 0.999,
+      "step": 2361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.879272001365541e-05,
+      "loss": 1.0348,
+      "step": 2362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8791654586342307e-05,
+      "loss": 1.0603,
+      "step": 2363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8790588719344492e-05,
+      "loss": 1.0427,
+      "step": 2364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8789522412715273e-05,
+      "loss": 1.0676,
+      "step": 2365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.878845566650798e-05,
+      "loss": 1.0023,
+      "step": 2366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8787388480775958e-05,
+      "loss": 1.0118,
+      "step": 2367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8786320855572582e-05,
+      "loss": 1.1046,
+      "step": 2368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.878525279095124e-05,
+      "loss": 0.9365,
+      "step": 2369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.878418428696536e-05,
+      "loss": 0.9909,
+      "step": 2370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.878311534366837e-05,
+      "loss": 0.9699,
+      "step": 2371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8782045961113727e-05,
+      "loss": 1.1384,
+      "step": 2372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8780976139354917e-05,
+      "loss": 0.9125,
+      "step": 2373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8779905878445445e-05,
+      "loss": 0.9748,
+      "step": 2374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8778835178438833e-05,
+      "loss": 0.9864,
+      "step": 2375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8777764039388635e-05,
+      "loss": 0.9163,
+      "step": 2376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8776692461348408e-05,
+      "loss": 1.0966,
+      "step": 2377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8775620444371755e-05,
+      "loss": 1.1309,
+      "step": 2378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8774547988512288e-05,
+      "loss": 1.0155,
+      "step": 2379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8773475093823635e-05,
+      "loss": 0.9542,
+      "step": 2380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8772401760359457e-05,
+      "loss": 0.6817,
+      "step": 2381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8771327988173436e-05,
+      "loss": 0.9688,
+      "step": 2382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.877025377731927e-05,
+      "loss": 1.0674,
+      "step": 2383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8769179127850685e-05,
+      "loss": 0.8781,
+      "step": 2384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876810403982142e-05,
+      "loss": 1.0152,
+      "step": 2385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8767028513285247e-05,
+      "loss": 1.0217,
+      "step": 2386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8765952548295955e-05,
+      "loss": 1.07,
+      "step": 2387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876487614490735e-05,
+      "loss": 1.1568,
+      "step": 2388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876379930317327e-05,
+      "loss": 0.9847,
+      "step": 2389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8762722023147567e-05,
+      "loss": 0.9038,
+      "step": 2390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8761644304884116e-05,
+      "loss": 0.9824,
+      "step": 2391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.876056614843682e-05,
+      "loss": 1.0675,
+      "step": 2392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8759487553859594e-05,
+      "loss": 1.0656,
+      "step": 2393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8758408521206384e-05,
+      "loss": 1.1322,
+      "step": 2394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8757329050531153e-05,
+      "loss": 1.0459,
+      "step": 2395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8756249141887884e-05,
+      "loss": 0.971,
+      "step": 2396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8755168795330594e-05,
+      "loss": 1.077,
+      "step": 2397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8754088010913306e-05,
+      "loss": 1.1129,
+      "step": 2398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8753006788690068e-05,
+      "loss": 0.9746,
+      "step": 2399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.875192512871496e-05,
+      "loss": 1.0601,
+      "step": 2400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8750843031042075e-05,
+      "loss": 1.1128,
+      "step": 2401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8749760495725536e-05,
+      "loss": 1.0012,
+      "step": 2402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874867752281947e-05,
+      "loss": 1.1131,
+      "step": 2403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874759411237805e-05,
+      "loss": 1.0919,
+      "step": 2404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8746510264455455e-05,
+      "loss": 0.9977,
+      "step": 2405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8745425979105892e-05,
+      "loss": 0.7299,
+      "step": 2406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874434125638358e-05,
+      "loss": 1.0533,
+      "step": 2407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8743256096342778e-05,
+      "loss": 1.1308,
+      "step": 2408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874217049903775e-05,
+      "loss": 1.008,
+      "step": 2409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874108446452279e-05,
+      "loss": 0.9705,
+      "step": 2410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8739997992852216e-05,
+      "loss": 1.0975,
+      "step": 2411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8738911084080354e-05,
+      "loss": 0.9902,
+      "step": 2412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8737823738261576e-05,
+      "loss": 0.9682,
+      "step": 2413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8736735955450253e-05,
+      "loss": 0.9794,
+      "step": 2414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873564773570079e-05,
+      "loss": 1.0862,
+      "step": 2415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8734559079067605e-05,
+      "loss": 0.9766,
+      "step": 2416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8733469985605147e-05,
+      "loss": 0.953,
+      "step": 2417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873238045536789e-05,
+      "loss": 1.0381,
+      "step": 2418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8731290488410315e-05,
+      "loss": 0.9058,
+      "step": 2419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.873020008478693e-05,
+      "loss": 1.088,
+      "step": 2420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8729109244552277e-05,
+      "loss": 1.0332,
+      "step": 2421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8728017967760906e-05,
+      "loss": 1.1,
+      "step": 2422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8726926254467396e-05,
+      "loss": 1.0629,
+      "step": 2423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8725834104726343e-05,
+      "loss": 1.101,
+      "step": 2424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8724741518592367e-05,
+      "loss": 1.0928,
+      "step": 2425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.872364849612011e-05,
+      "loss": 1.0566,
+      "step": 2426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8722555037364236e-05,
+      "loss": 0.9203,
+      "step": 2427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8721461142379432e-05,
+      "loss": 1.0224,
+      "step": 2428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8720366811220403e-05,
+      "loss": 1.1477,
+      "step": 2429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8719272043941883e-05,
+      "loss": 1.0148,
+      "step": 2430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8718176840598618e-05,
+      "loss": 1.0979,
+      "step": 2431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871708120124538e-05,
+      "loss": 0.9976,
+      "step": 2432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871598512593697e-05,
+      "loss": 1.0223,
+      "step": 2433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8714888614728198e-05,
+      "loss": 0.9656,
+      "step": 2434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.871379166767391e-05,
+      "loss": 1.1059,
+      "step": 2435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8712694284828955e-05,
+      "loss": 1.0229,
+      "step": 2436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8711596466248223e-05,
+      "loss": 0.9189,
+      "step": 2437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8710498211986618e-05,
+      "loss": 1.1277,
+      "step": 2438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8709399522099062e-05,
+      "loss": 0.9229,
+      "step": 2439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8708300396640503e-05,
+      "loss": 0.9688,
+      "step": 2440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870720083566591e-05,
+      "loss": 1.0083,
+      "step": 2441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8706100839230273e-05,
+      "loss": 0.9925,
+      "step": 2442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870500040738861e-05,
+      "loss": 1.1087,
+      "step": 2443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870389954019595e-05,
+      "loss": 0.9507,
+      "step": 2444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8702798237707346e-05,
+      "loss": 0.9144,
+      "step": 2445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8701696499977884e-05,
+      "loss": 1.0095,
+      "step": 2446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.870059432706266e-05,
+      "loss": 1.01,
+      "step": 2447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8699491719016794e-05,
+      "loss": 1.0633,
+      "step": 2448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869838867589543e-05,
+      "loss": 1.007,
+      "step": 2449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8697285197753733e-05,
+      "loss": 1.0658,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8696181284646893e-05,
+      "loss": 1.0086,
+      "step": 2451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869507693663011e-05,
+      "loss": 0.9911,
+      "step": 2452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8693972153758622e-05,
+      "loss": 0.6796,
+      "step": 2453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8692866936087683e-05,
+      "loss": 1.0378,
+      "step": 2454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869176128367256e-05,
+      "loss": 1.0985,
+      "step": 2455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869065519656855e-05,
+      "loss": 1.1257,
+      "step": 2456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8689548674830972e-05,
+      "loss": 1.119,
+      "step": 2457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8688441718515163e-05,
+      "loss": 1.07,
+      "step": 2458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8687334327676487e-05,
+      "loss": 0.9435,
+      "step": 2459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868622650237032e-05,
+      "loss": 1.153,
+      "step": 2460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8685118242652077e-05,
+      "loss": 1.0065,
+      "step": 2461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868400954857717e-05,
+      "loss": 1.0366,
+      "step": 2462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8682900420201058e-05,
+      "loss": 0.9984,
+      "step": 2463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8681790857579203e-05,
+      "loss": 0.9857,
+      "step": 2464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86806808607671e-05,
+      "loss": 1.0227,
+      "step": 2465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8679570429820263e-05,
+      "loss": 0.8504,
+      "step": 2466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867845956479422e-05,
+      "loss": 1.0471,
+      "step": 2467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8677348265744532e-05,
+      "loss": 0.9025,
+      "step": 2468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8676236532726774e-05,
+      "loss": 1.0266,
+      "step": 2469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8675124365796554e-05,
+      "loss": 1.1363,
+      "step": 2470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867401176500948e-05,
+      "loss": 0.9782,
+      "step": 2471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8672898730421206e-05,
+      "loss": 1.0653,
+      "step": 2472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8671785262087388e-05,
+      "loss": 0.9219,
+      "step": 2473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8670671360063722e-05,
+      "loss": 0.9975,
+      "step": 2474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866955702440591e-05,
+      "loss": 0.9628,
+      "step": 2475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866844225516968e-05,
+      "loss": 1.0786,
+      "step": 2476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8667327052410788e-05,
+      "loss": 1.091,
+      "step": 2477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8666211416185e-05,
+      "loss": 1.042,
+      "step": 2478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8665095346548123e-05,
+      "loss": 0.9626,
+      "step": 2479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866397884355596e-05,
+      "loss": 0.671,
+      "step": 2480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866286190726436e-05,
+      "loss": 1.1287,
+      "step": 2481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8661744537729175e-05,
+      "loss": 1.0783,
+      "step": 2482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8660626735006287e-05,
+      "loss": 0.946,
+      "step": 2483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8659508499151605e-05,
+      "loss": 0.9678,
+      "step": 2484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8658389830221048e-05,
+      "loss": 1.0435,
+      "step": 2485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.865727072827057e-05,
+      "loss": 1.0939,
+      "step": 2486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.865615119335613e-05,
+      "loss": 1.0264,
+      "step": 2487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8655031225533718e-05,
+      "loss": 0.9825,
+      "step": 2488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8653910824859354e-05,
+      "loss": 0.9591,
+      "step": 2489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8652789991389064e-05,
+      "loss": 1.0866,
+      "step": 2490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8651668725178905e-05,
+      "loss": 0.9409,
+      "step": 2491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8650547026284953e-05,
+      "loss": 1.0187,
+      "step": 2492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8649424894763305e-05,
+      "loss": 0.6619,
+      "step": 2493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8648302330670083e-05,
+      "loss": 1.0354,
+      "step": 2494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8647179334061425e-05,
+      "loss": 0.9496,
+      "step": 2495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8646055904993495e-05,
+      "loss": 1.01,
+      "step": 2496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.864493204352248e-05,
+      "loss": 0.9588,
+      "step": 2497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8643807749704584e-05,
+      "loss": 0.8484,
+      "step": 2498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8642683023596034e-05,
+      "loss": 1.0211,
+      "step": 2499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.864155786525308e-05,
+      "loss": 1.0258,
+      "step": 2500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8640432274731995e-05,
+      "loss": 1.0122,
+      "step": 2501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.863930625208907e-05,
+      "loss": 1.0163,
+      "step": 2502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8638179797380614e-05,
+      "loss": 1.0643,
+      "step": 2503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.863705291066297e-05,
+      "loss": 1.051,
+      "step": 2504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8635925591992496e-05,
+      "loss": 1.0548,
+      "step": 2505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8634797841425567e-05,
+      "loss": 0.6686,
+      "step": 2506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.863366965901858e-05,
+      "loss": 0.696,
+      "step": 2507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.863254104482797e-05,
+      "loss": 0.9304,
+      "step": 2508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.863141199891017e-05,
+      "loss": 0.9931,
+      "step": 2509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8630282521321644e-05,
+      "loss": 1.0297,
+      "step": 2510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.862915261211889e-05,
+      "loss": 0.9405,
+      "step": 2511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8628022271358404e-05,
+      "loss": 1.1333,
+      "step": 2512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8626891499096724e-05,
+      "loss": 0.9897,
+      "step": 2513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.86257602953904e-05,
+      "loss": 0.9747,
+      "step": 2514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8624628660296004e-05,
+      "loss": 1.0419,
+      "step": 2515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8623496593870132e-05,
+      "loss": 1.1347,
+      "step": 2516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8622364096169398e-05,
+      "loss": 0.9647,
+      "step": 2517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8621231167250445e-05,
+      "loss": 1.0759,
+      "step": 2518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.862009780716993e-05,
+      "loss": 0.9453,
+      "step": 2519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8618964015984536e-05,
+      "loss": 1.0968,
+      "step": 2520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.861782979375096e-05,
+      "loss": 1.167,
+      "step": 2521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.861669514052593e-05,
+      "loss": 0.9645,
+      "step": 2522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8615560056366193e-05,
+      "loss": 1.0544,
+      "step": 2523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8614424541328516e-05,
+      "loss": 1.0461,
+      "step": 2524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8613288595469685e-05,
+      "loss": 1.089,
+      "step": 2525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8612152218846513e-05,
+      "loss": 1.0329,
+      "step": 2526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8611015411515832e-05,
+      "loss": 1.0216,
+      "step": 2527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8609878173534495e-05,
+      "loss": 0.9033,
+      "step": 2528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8608740504959382e-05,
+      "loss": 0.9936,
+      "step": 2529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8607602405847377e-05,
+      "loss": 0.9628,
+      "step": 2530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8606463876255412e-05,
+      "loss": 1.0466,
+      "step": 2531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8605324916240416e-05,
+      "loss": 0.9881,
+      "step": 2532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.860418552585936e-05,
+      "loss": 0.9668,
+      "step": 2533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8603045705169217e-05,
+      "loss": 1.043,
+      "step": 2534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8601905454226997e-05,
+      "loss": 1.0284,
+      "step": 2535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8600764773089727e-05,
+      "loss": 0.9742,
+      "step": 2536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859962366181445e-05,
+      "loss": 1.1082,
+      "step": 2537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8598482120458238e-05,
+      "loss": 1.0692,
+      "step": 2538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8597340149078177e-05,
+      "loss": 1.0109,
+      "step": 2539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8596197747731387e-05,
+      "loss": 1.0986,
+      "step": 2540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8595054916474994e-05,
+      "loss": 0.9238,
+      "step": 2541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8593911655366154e-05,
+      "loss": 0.943,
+      "step": 2542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8592767964462043e-05,
+      "loss": 1.0558,
+      "step": 2543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8591623843819863e-05,
+      "loss": 0.995,
+      "step": 2544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.859047929349683e-05,
+      "loss": 0.9867,
+      "step": 2545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8589334313550185e-05,
+      "loss": 0.9451,
+      "step": 2546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8588188904037187e-05,
+      "loss": 1.023,
+      "step": 2547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8587043065015126e-05,
+      "loss": 1.0003,
+      "step": 2548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8585896796541307e-05,
+      "loss": 1.1301,
+      "step": 2549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8584750098673052e-05,
+      "loss": 0.9627,
+      "step": 2550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.858360297146771e-05,
+      "loss": 0.9114,
+      "step": 2551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8582455414982658e-05,
+      "loss": 0.9205,
+      "step": 2552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8581307429275275e-05,
+      "loss": 1.0334,
+      "step": 2553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8580159014402982e-05,
+      "loss": 0.9661,
+      "step": 2554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8579010170423212e-05,
+      "loss": 0.8712,
+      "step": 2555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8577860897393418e-05,
+      "loss": 1.009,
+      "step": 2556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8576711195371078e-05,
+      "loss": 0.9406,
+      "step": 2557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.857556106441369e-05,
+      "loss": 1.0473,
+      "step": 2558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8574410504578777e-05,
+      "loss": 0.9982,
+      "step": 2559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8573259515923877e-05,
+      "loss": 1.0831,
+      "step": 2560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8572108098506554e-05,
+      "loss": 0.9745,
+      "step": 2561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8570956252384392e-05,
+      "loss": 1.0955,
+      "step": 2562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8569803977615e-05,
+      "loss": 0.9403,
+      "step": 2563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8568651274255997e-05,
+      "loss": 0.9335,
+      "step": 2564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8567498142365038e-05,
+      "loss": 1.0591,
+      "step": 2565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8566344581999793e-05,
+      "loss": 0.9513,
+      "step": 2566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.856519059321795e-05,
+      "loss": 1.1472,
+      "step": 2567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8564036176077227e-05,
+      "loss": 0.9241,
+      "step": 2568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8562881330635353e-05,
+      "loss": 0.9916,
+      "step": 2569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8561726056950084e-05,
+      "loss": 1.0491,
+      "step": 2570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8560570355079198e-05,
+      "loss": 0.9933,
+      "step": 2571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8559414225080495e-05,
+      "loss": 1.0208,
+      "step": 2572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8558257667011795e-05,
+      "loss": 1.0332,
+      "step": 2573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8557100680930937e-05,
+      "loss": 0.9368,
+      "step": 2574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8555943266895786e-05,
+      "loss": 1.13,
+      "step": 2575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8554785424964225e-05,
+      "loss": 1.0216,
+      "step": 2576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8553627155194158e-05,
+      "loss": 0.9019,
+      "step": 2577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8552468457643517e-05,
+      "loss": 0.9425,
+      "step": 2578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8551309332370248e-05,
+      "loss": 0.6408,
+      "step": 2579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8550149779432313e-05,
+      "loss": 1.0794,
+      "step": 2580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8548989798887714e-05,
+      "loss": 0.7133,
+      "step": 2581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854782939079446e-05,
+      "loss": 0.981,
+      "step": 2582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8546668555210584e-05,
+      "loss": 1.0876,
+      "step": 2583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8545507292194142e-05,
+      "loss": 0.9444,
+      "step": 2584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8544345601803205e-05,
+      "loss": 1.0045,
+      "step": 2585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8543183484095883e-05,
+      "loss": 0.9429,
+      "step": 2586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.854202093913028e-05,
+      "loss": 1.0444,
+      "step": 2587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8540857966964552e-05,
+      "loss": 1.0103,
+      "step": 2588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853969456765685e-05,
+      "loss": 1.0876,
+      "step": 2589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8538530741265367e-05,
+      "loss": 1.0421,
+      "step": 2590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8537366487848297e-05,
+      "loss": 1.027,
+      "step": 2591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8536201807463874e-05,
+      "loss": 0.9343,
+      "step": 2592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8535036700170343e-05,
+      "loss": 0.9155,
+      "step": 2593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8533871166025972e-05,
+      "loss": 0.9629,
+      "step": 2594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8532705205089055e-05,
+      "loss": 1.0424,
+      "step": 2595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8531538817417898e-05,
+      "loss": 0.99,
+      "step": 2596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.853037200307084e-05,
+      "loss": 1.025,
+      "step": 2597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852920476210623e-05,
+      "loss": 0.9573,
+      "step": 2598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8528037094582444e-05,
+      "loss": 0.8979,
+      "step": 2599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8526869000557882e-05,
+      "loss": 0.9539,
+      "step": 2600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852570048009096e-05,
+      "loss": 0.9908,
+      "step": 2601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852453153324012e-05,
+      "loss": 0.8877,
+      "step": 2602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8523362160063822e-05,
+      "loss": 1.0625,
+      "step": 2603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522192360620542e-05,
+      "loss": 0.9119,
+      "step": 2604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8521022134968792e-05,
+      "loss": 0.9936,
+      "step": 2605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8519851483167097e-05,
+      "loss": 1.0125,
+      "step": 2606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8518680405273998e-05,
+      "loss": 0.9807,
+      "step": 2607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8517508901348063e-05,
+      "loss": 1.0941,
+      "step": 2608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8516336971447884e-05,
+      "loss": 0.9098,
+      "step": 2609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8515164615632067e-05,
+      "loss": 1.1265,
+      "step": 2610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851399183395925e-05,
+      "loss": 1.1274,
+      "step": 2611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8512818626488078e-05,
+      "loss": 1.091,
+      "step": 2612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851164499327723e-05,
+      "loss": 0.69,
+      "step": 2613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8510470934385398e-05,
+      "loss": 0.9895,
+      "step": 2614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8509296449871302e-05,
+      "loss": 1.0006,
+      "step": 2615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8508121539793678e-05,
+      "loss": 1.0956,
+      "step": 2616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8506946204211285e-05,
+      "loss": 0.9973,
+      "step": 2617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505770443182904e-05,
+      "loss": 0.9845,
+      "step": 2618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8504594256767334e-05,
+      "loss": 1.1029,
+      "step": 2619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8503417645023405e-05,
+      "loss": 0.9651,
+      "step": 2620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8502240608009953e-05,
+      "loss": 0.6823,
+      "step": 2621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8501063145785846e-05,
+      "loss": 0.9834,
+      "step": 2622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8499885258409977e-05,
+      "loss": 1.0865,
+      "step": 2623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849870694594124e-05,
+      "loss": 0.9847,
+      "step": 2624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8497528208438583e-05,
+      "loss": 1.1202,
+      "step": 2625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.849634904596094e-05,
+      "loss": 0.9304,
+      "step": 2626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8495169458567294e-05,
+      "loss": 0.8642,
+      "step": 2627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.849398944631663e-05,
+      "loss": 0.9368,
+      "step": 2628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.849280900926797e-05,
+      "loss": 0.991,
+      "step": 2629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8491628147480343e-05,
+      "loss": 1.0557,
+      "step": 2630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8490446861012807e-05,
+      "loss": 1.0442,
+      "step": 2631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8489265149924442e-05,
+      "loss": 0.9658,
+      "step": 2632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.848808301427435e-05,
+      "loss": 0.6875,
+      "step": 2633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8486900454121642e-05,
+      "loss": 1.0161,
+      "step": 2634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.848571746952547e-05,
+      "loss": 1.0397,
+      "step": 2635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.848453406054499e-05,
+      "loss": 1.0392,
+      "step": 2636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.848335022723939e-05,
+      "loss": 1.1125,
+      "step": 2637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8482165969667873e-05,
+      "loss": 1.0097,
+      "step": 2638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.848098128788967e-05,
+      "loss": 1.0499,
+      "step": 2639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8479796181964024e-05,
+      "loss": 1.0426,
+      "step": 2640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84786106519502e-05,
+      "loss": 1.1662,
+      "step": 2641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8477424697907504e-05,
+      "loss": 0.9214,
+      "step": 2642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.847623831989523e-05,
+      "loss": 1.0444,
+      "step": 2643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.847505151797272e-05,
+      "loss": 1.0679,
+      "step": 2644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8473864292199327e-05,
+      "loss": 0.8928,
+      "step": 2645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8472676642634425e-05,
+      "loss": 1.0076,
+      "step": 2646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8471488569337412e-05,
+      "loss": 0.9764,
+      "step": 2647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84703000723677e-05,
+      "loss": 0.9475,
+      "step": 2648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8469111151784736e-05,
+      "loss": 1.0304,
+      "step": 2649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8467921807647972e-05,
+      "loss": 1.1064,
+      "step": 2650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8466732040016894e-05,
+      "loss": 1.1424,
+      "step": 2651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8465541848951e-05,
+      "loss": 1.0061,
+      "step": 2652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8464351234509818e-05,
+      "loss": 1.0035,
+      "step": 2653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8463160196752886e-05,
+      "loss": 0.6649,
+      "step": 2654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8461968735739777e-05,
+      "loss": 1.0225,
+      "step": 2655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8460776851530074e-05,
+      "loss": 1.071,
+      "step": 2656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.845958454418339e-05,
+      "loss": 0.996,
+      "step": 2657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8458391813759346e-05,
+      "loss": 1.025,
+      "step": 2658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8457198660317597e-05,
+      "loss": 0.6535,
+      "step": 2659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8456005083917815e-05,
+      "loss": 0.6665,
+      "step": 2660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8454811084619687e-05,
+      "loss": 0.9366,
+      "step": 2661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8453616662482936e-05,
+      "loss": 1.0318,
+      "step": 2662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.845242181756729e-05,
+      "loss": 0.9901,
+      "step": 2663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8451226549932514e-05,
+      "loss": 1.0044,
+      "step": 2664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.845003085963837e-05,
+      "loss": 0.9962,
+      "step": 2665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8448834746744668e-05,
+      "loss": 1.0901,
+      "step": 2666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8447638211311227e-05,
+      "loss": 0.9772,
+      "step": 2667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844644125339788e-05,
+      "loss": 0.6865,
+      "step": 2668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.84452438730645e-05,
+      "loss": 0.6898,
+      "step": 2669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8444046070370963e-05,
+      "loss": 0.9546,
+      "step": 2670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.844284784537717e-05,
+      "loss": 1.0603,
+      "step": 2671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8441649198143053e-05,
+      "loss": 0.9998,
+      "step": 2672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8440450128728558e-05,
+      "loss": 1.0641,
+      "step": 2673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8439250637193648e-05,
+      "loss": 1.0712,
+      "step": 2674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843805072359831e-05,
+      "loss": 0.9346,
+      "step": 2675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8436850388002557e-05,
+      "loss": 0.9311,
+      "step": 2676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8435649630466424e-05,
+      "loss": 0.9571,
+      "step": 2677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8434448451049956e-05,
+      "loss": 1.0106,
+      "step": 2678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.843324684981323e-05,
+      "loss": 0.9664,
+      "step": 2679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8432044826816337e-05,
+      "loss": 0.9556,
+      "step": 2680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8430842382119393e-05,
+      "loss": 1.0475,
+      "step": 2681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8429639515782533e-05,
+      "loss": 0.9757,
+      "step": 2682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8428436227865917e-05,
+      "loss": 1.0653,
+      "step": 2683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8427232518429723e-05,
+      "loss": 1.0217,
+      "step": 2684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842602838753415e-05,
+      "loss": 0.9552,
+      "step": 2685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8424823835239416e-05,
+      "loss": 0.9938,
+      "step": 2686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8423618861605766e-05,
+      "loss": 0.9572,
+      "step": 2687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8422413466693464e-05,
+      "loss": 0.9812,
+      "step": 2688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8421207650562786e-05,
+      "loss": 0.9485,
+      "step": 2689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8420001413274045e-05,
+      "loss": 1.0228,
+      "step": 2690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8418794754887564e-05,
+      "loss": 0.9866,
+      "step": 2691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841758767546369e-05,
+      "loss": 1.0388,
+      "step": 2692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8416380175062788e-05,
+      "loss": 1.0303,
+      "step": 2693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841517225374525e-05,
+      "loss": 0.9722,
+      "step": 2694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841396391157149e-05,
+      "loss": 0.9424,
+      "step": 2695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.841275514860193e-05,
+      "loss": 0.9638,
+      "step": 2696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8411545964897027e-05,
+      "loss": 0.678,
+      "step": 2697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8410336360517257e-05,
+      "loss": 0.9564,
+      "step": 2698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840912633552311e-05,
+      "loss": 0.8815,
+      "step": 2699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8407915889975107e-05,
+      "loss": 1.0594,
+      "step": 2700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8406705023933775e-05,
+      "loss": 1.0156,
+      "step": 2701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840549373745968e-05,
+      "loss": 1.0596,
+      "step": 2702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840428203061339e-05,
+      "loss": 0.8817,
+      "step": 2703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8403069903455518e-05,
+      "loss": 0.9194,
+      "step": 2704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8401857356046674e-05,
+      "loss": 1.0799,
+      "step": 2705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.840064438844751e-05,
+      "loss": 0.9889,
+      "step": 2706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8399431000718674e-05,
+      "loss": 1.045,
+      "step": 2707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839821719292086e-05,
+      "loss": 1.0703,
+      "step": 2708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839700296511477e-05,
+      "loss": 0.9587,
+      "step": 2709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8395788317361126e-05,
+      "loss": 1.0265,
+      "step": 2710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839457324972068e-05,
+      "loss": 1.0131,
+      "step": 2711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8393357762254194e-05,
+      "loss": 0.9585,
+      "step": 2712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8392141855022462e-05,
+      "loss": 1.0442,
+      "step": 2713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.839092552808629e-05,
+      "loss": 1.0276,
+      "step": 2714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8389708781506512e-05,
+      "loss": 1.0406,
+      "step": 2715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838849161534397e-05,
+      "loss": 0.9873,
+      "step": 2716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.838727402965955e-05,
+      "loss": 1.1101,
+      "step": 2717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8386056024514136e-05,
+      "loss": 1.0043,
+      "step": 2718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8384837599968644e-05,
+      "loss": 0.9678,
+      "step": 2719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8383618756084015e-05,
+      "loss": 1.0573,
+      "step": 2720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8382399492921194e-05,
+      "loss": 0.9972,
+      "step": 2721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8381179810541167e-05,
+      "loss": 1.083,
+      "step": 2722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8379959709004932e-05,
+      "loss": 0.9651,
+      "step": 2723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8378739188373508e-05,
+      "loss": 0.9803,
+      "step": 2724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8377518248707932e-05,
+      "loss": 0.9575,
+      "step": 2725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8376296890069266e-05,
+      "loss": 0.9701,
+      "step": 2726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.837507511251859e-05,
+      "loss": 1.0692,
+      "step": 2727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8373852916117014e-05,
+      "loss": 1.0864,
+      "step": 2728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8372630300925656e-05,
+      "loss": 1.0918,
+      "step": 2729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8371407267005663e-05,
+      "loss": 1.0266,
+      "step": 2730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83701838144182e-05,
+      "loss": 1.0537,
+      "step": 2731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836895994322446e-05,
+      "loss": 1.0479,
+      "step": 2732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836773565348564e-05,
+      "loss": 0.6849,
+      "step": 2733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8366510945262974e-05,
+      "loss": 0.9276,
+      "step": 2734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8365285818617713e-05,
+      "loss": 0.9778,
+      "step": 2735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8364060273611127e-05,
+      "loss": 1.0389,
+      "step": 2736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8362834310304506e-05,
+      "loss": 1.0198,
+      "step": 2737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836160792875916e-05,
+      "loss": 1.0403,
+      "step": 2738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836038112903643e-05,
+      "loss": 1.0451,
+      "step": 2739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8359153911197665e-05,
+      "loss": 1.0013,
+      "step": 2740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8357926275304238e-05,
+      "loss": 0.6693,
+      "step": 2741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835669822141755e-05,
+      "loss": 0.705,
+      "step": 2742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8355469749599014e-05,
+      "loss": 1.1073,
+      "step": 2743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835424085991007e-05,
+      "loss": 0.9868,
+      "step": 2744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8353011552412176e-05,
+      "loss": 1.0534,
+      "step": 2745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8351781827166813e-05,
+      "loss": 1.0021,
+      "step": 2746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8350551684235477e-05,
+      "loss": 0.9855,
+      "step": 2747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83493211236797e-05,
+      "loss": 1.0408,
+      "step": 2748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834809014556101e-05,
+      "loss": 0.9258,
+      "step": 2749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834685874994098e-05,
+      "loss": 1.0509,
+      "step": 2750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8345626936881192e-05,
+      "loss": 0.9183,
+      "step": 2751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834439470644325e-05,
+      "loss": 1.0019,
+      "step": 2752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834316205868878e-05,
+      "loss": 0.9516,
+      "step": 2753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8341928993679427e-05,
+      "loss": 0.9716,
+      "step": 2754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8340695511476863e-05,
+      "loss": 1.1467,
+      "step": 2755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8339461612142772e-05,
+      "loss": 0.9714,
+      "step": 2756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8338227295738864e-05,
+      "loss": 0.8938,
+      "step": 2757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8336992562326873e-05,
+      "loss": 0.996,
+      "step": 2758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8335757411968545e-05,
+      "loss": 1.0384,
+      "step": 2759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8334521844725656e-05,
+      "loss": 0.9367,
+      "step": 2760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8333285860659998e-05,
+      "loss": 0.9283,
+      "step": 2761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8332049459833378e-05,
+      "loss": 0.7202,
+      "step": 2762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833081264230764e-05,
+      "loss": 0.9806,
+      "step": 2763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8329575408144635e-05,
+      "loss": 0.9281,
+      "step": 2764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8328337757406236e-05,
+      "loss": 0.929,
+      "step": 2765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8327099690154344e-05,
+      "loss": 1.0541,
+      "step": 2766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325861206450874e-05,
+      "loss": 1.0007,
+      "step": 2767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8324622306357767e-05,
+      "loss": 0.9892,
+      "step": 2768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8323382989936983e-05,
+      "loss": 1.0067,
+      "step": 2769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8322143257250502e-05,
+      "loss": 0.9962,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8320903108360323e-05,
+      "loss": 0.9398,
+      "step": 2771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8319662543328467e-05,
+      "loss": 0.9879,
+      "step": 2772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.831842156221698e-05,
+      "loss": 1.1157,
+      "step": 2773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.831718016508792e-05,
+      "loss": 0.9698,
+      "step": 2774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.831593835200338e-05,
+      "loss": 0.939,
+      "step": 2775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8314696123025456e-05,
+      "loss": 1.002,
+      "step": 2776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8313453478216277e-05,
+      "loss": 1.0307,
+      "step": 2777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.831221041763799e-05,
+      "loss": 1.0384,
+      "step": 2778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8310966941352765e-05,
+      "loss": 0.6767,
+      "step": 2779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.830972304942279e-05,
+      "loss": 0.6649,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.830847874191027e-05,
+      "loss": 0.9633,
+      "step": 2781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8307234018877437e-05,
+      "loss": 1.0909,
+      "step": 2782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.830598888038654e-05,
+      "loss": 0.9867,
+      "step": 2783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8304743326499856e-05,
+      "loss": 0.9199,
+      "step": 2784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8303497357279673e-05,
+      "loss": 0.9627,
+      "step": 2785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.83022509727883e-05,
+      "loss": 1.0455,
+      "step": 2786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8301004173088077e-05,
+      "loss": 0.9987,
+      "step": 2787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8299756958241353e-05,
+      "loss": 0.8771,
+      "step": 2788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8298509328310508e-05,
+      "loss": 0.6702,
+      "step": 2789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.829726128335794e-05,
+      "loss": 0.989,
+      "step": 2790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.829601282344606e-05,
+      "loss": 0.9898,
+      "step": 2791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8294763948637306e-05,
+      "loss": 0.9899,
+      "step": 2792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8293514658994135e-05,
+      "loss": 1.1043,
+      "step": 2793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8292264954579032e-05,
+      "loss": 1.0678,
+      "step": 2794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8291014835454495e-05,
+      "loss": 1.0386,
+      "step": 2795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8289764301683043e-05,
+      "loss": 1.0715,
+      "step": 2796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8288513353327212e-05,
+      "loss": 1.1569,
+      "step": 2797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.828726199044957e-05,
+      "loss": 0.9716,
+      "step": 2798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8286010213112705e-05,
+      "loss": 0.8661,
+      "step": 2799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.828475802137921e-05,
+      "loss": 0.9797,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8283505415311712e-05,
+      "loss": 0.9897,
+      "step": 2801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.828225239497286e-05,
+      "loss": 1.0244,
+      "step": 2802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8280998960425313e-05,
+      "loss": 0.9625,
+      "step": 2803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827974511173176e-05,
+      "loss": 0.984,
+      "step": 2804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8278490848954913e-05,
+      "loss": 1.0075,
+      "step": 2805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8277236172157492e-05,
+      "loss": 1.0444,
+      "step": 2806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8275981081402252e-05,
+      "loss": 1.0392,
+      "step": 2807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827472557675196e-05,
+      "loss": 1.0462,
+      "step": 2808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8273469658269397e-05,
+      "loss": 0.9618,
+      "step": 2809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.827221332601739e-05,
+      "loss": 0.8927,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8270956580058755e-05,
+      "loss": 0.6258,
+      "step": 2811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8269699420456357e-05,
+      "loss": 0.9065,
+      "step": 2812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8268441847273056e-05,
+      "loss": 0.9713,
+      "step": 2813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8267183860571753e-05,
+      "loss": 1.0223,
+      "step": 2814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826592546041536e-05,
+      "loss": 0.9921,
+      "step": 2815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8264666646866812e-05,
+      "loss": 0.8742,
+      "step": 2816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826340741998906e-05,
+      "loss": 0.9971,
+      "step": 2817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826214777984509e-05,
+      "loss": 0.9704,
+      "step": 2818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.826088772649789e-05,
+      "loss": 1.0318,
+      "step": 2819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825962726001048e-05,
+      "loss": 1.0397,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8258366380445897e-05,
+      "loss": 0.9062,
+      "step": 2821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82571050878672e-05,
+      "loss": 1.0519,
+      "step": 2822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.825584338233747e-05,
+      "loss": 1.0002,
+      "step": 2823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8254581263919804e-05,
+      "loss": 1.0572,
+      "step": 2824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8253318732677326e-05,
+      "loss": 1.0368,
+      "step": 2825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8252055788673175e-05,
+      "loss": 0.9248,
+      "step": 2826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8250792431970513e-05,
+      "loss": 1.0459,
+      "step": 2827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824952866263252e-05,
+      "loss": 1.1456,
+      "step": 2828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8248264480722403e-05,
+      "loss": 1.0736,
+      "step": 2829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8246999886303384e-05,
+      "loss": 1.0901,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824573487943871e-05,
+      "loss": 0.9743,
+      "step": 2831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824446946019164e-05,
+      "loss": 1.0198,
+      "step": 2832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.824320362862547e-05,
+      "loss": 1.0307,
+      "step": 2833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8241937384803497e-05,
+      "loss": 1.1149,
+      "step": 2834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8240670728789046e-05,
+      "loss": 1.0952,
+      "step": 2835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8239403660645477e-05,
+      "loss": 0.9482,
+      "step": 2836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8238136180436147e-05,
+      "loss": 1.0371,
+      "step": 2837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8236868288224445e-05,
+      "loss": 1.0989,
+      "step": 2838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8235599984073787e-05,
+      "loss": 0.9856,
+      "step": 2839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8234331268047595e-05,
+      "loss": 1.0082,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.823306214020933e-05,
+      "loss": 1.0202,
+      "step": 2841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8231792600622452e-05,
+      "loss": 1.0087,
+      "step": 2842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8230522649350458e-05,
+      "loss": 0.952,
+      "step": 2843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822925228645686e-05,
+      "loss": 1.0282,
+      "step": 2844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822798151200519e-05,
+      "loss": 1.0121,
+      "step": 2845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8226710326059004e-05,
+      "loss": 0.9478,
+      "step": 2846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8225438728681876e-05,
+      "loss": 1.1086,
+      "step": 2847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8224166719937396e-05,
+      "loss": 1.005,
+      "step": 2848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8222894299889184e-05,
+      "loss": 1.0102,
+      "step": 2849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8221621468600872e-05,
+      "loss": 0.9637,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.822034822613612e-05,
+      "loss": 1.0353,
+      "step": 2851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.82190745725586e-05,
+      "loss": 0.9416,
+      "step": 2852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8217800507932015e-05,
+      "loss": 1.0012,
+      "step": 2853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821652603232008e-05,
+      "loss": 0.927,
+      "step": 2854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.821525114578653e-05,
+      "loss": 1.0302,
+      "step": 2855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8213975848395136e-05,
+      "loss": 1.0394,
+      "step": 2856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8212700140209665e-05,
+      "loss": 0.9652,
+      "step": 2857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8211424021293923e-05,
+      "loss": 1.063,
+      "step": 2858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8210147491711732e-05,
+      "loss": 0.9922,
+      "step": 2859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8208870551526925e-05,
+      "loss": 1.0252,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8207593200803376e-05,
+      "loss": 0.9745,
+      "step": 2861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820631543960496e-05,
+      "loss": 1.0378,
+      "step": 2862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820503726799558e-05,
+      "loss": 0.9707,
+      "step": 2863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.820375868603916e-05,
+      "loss": 0.9953,
+      "step": 2864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8202479693799645e-05,
+      "loss": 0.985,
+      "step": 2865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8201200291341e-05,
+      "loss": 1.0119,
+      "step": 2866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8199920478727206e-05,
+      "loss": 1.0167,
+      "step": 2867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8198640256022276e-05,
+      "loss": 1.0272,
+      "step": 2868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819735962329023e-05,
+      "loss": 0.9425,
+      "step": 2869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8196078580595112e-05,
+      "loss": 0.9966,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8194797128000992e-05,
+      "loss": 1.0194,
+      "step": 2871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8193515265571963e-05,
+      "loss": 1.0192,
+      "step": 2872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.819223299337213e-05,
+      "loss": 1.1267,
+      "step": 2873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8190950311465615e-05,
+      "loss": 1.0093,
+      "step": 2874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8189667219916573e-05,
+      "loss": 1.0051,
+      "step": 2875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8188383718789174e-05,
+      "loss": 1.0326,
+      "step": 2876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8187099808147604e-05,
+      "loss": 1.1218,
+      "step": 2877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8185815488056077e-05,
+      "loss": 0.9919,
+      "step": 2878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8184530758578822e-05,
+      "loss": 1.0653,
+      "step": 2879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818324561978009e-05,
+      "loss": 0.993,
+      "step": 2880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8181960071724155e-05,
+      "loss": 1.0714,
+      "step": 2881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818067411447531e-05,
+      "loss": 0.9958,
+      "step": 2882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817938774809786e-05,
+      "loss": 0.9545,
+      "step": 2883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817810097265615e-05,
+      "loss": 0.9406,
+      "step": 2884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8176813788214526e-05,
+      "loss": 1.0441,
+      "step": 2885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817552619483736e-05,
+      "loss": 1.1203,
+      "step": 2886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8174238192589055e-05,
+      "loss": 1.1263,
+      "step": 2887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817294978153402e-05,
+      "loss": 1.0158,
+      "step": 2888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8171660961736694e-05,
+      "loss": 1.0137,
+      "step": 2889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817037173326153e-05,
+      "loss": 1.0334,
+      "step": 2890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8169082096173004e-05,
+      "loss": 1.0689,
+      "step": 2891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8167792050535616e-05,
+      "loss": 0.8862,
+      "step": 2892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8166501596413878e-05,
+      "loss": 0.9987,
+      "step": 2893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8165210733872335e-05,
+      "loss": 1.0787,
+      "step": 2894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816391946297554e-05,
+      "loss": 1.0912,
+      "step": 2895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8162627783788075e-05,
+      "loss": 0.9791,
+      "step": 2896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8161335696374535e-05,
+      "loss": 0.9587,
+      "step": 2897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8160043200799538e-05,
+      "loss": 0.7002,
+      "step": 2898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815875029712773e-05,
+      "loss": 1.0744,
+      "step": 2899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815745698542377e-05,
+      "loss": 1.0879,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815616326575233e-05,
+      "loss": 0.9322,
+      "step": 2901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815486913817812e-05,
+      "loss": 1.0507,
+      "step": 2902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815357460276586e-05,
+      "loss": 0.6722,
+      "step": 2903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8152279659580295e-05,
+      "loss": 1.1214,
+      "step": 2904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8150984308686178e-05,
+      "loss": 0.6483,
+      "step": 2905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8149688550148296e-05,
+      "loss": 0.938,
+      "step": 2906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8148392384031455e-05,
+      "loss": 1.1087,
+      "step": 2907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8147095810400474e-05,
+      "loss": 0.9491,
+      "step": 2908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81457988293202e-05,
+      "loss": 0.9305,
+      "step": 2909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8144501440855497e-05,
+      "loss": 0.9485,
+      "step": 2910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8143203645071247e-05,
+      "loss": 0.9415,
+      "step": 2911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8141905442032353e-05,
+      "loss": 0.9849,
+      "step": 2912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8140606831803746e-05,
+      "loss": 1.0694,
+      "step": 2913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8139307814450368e-05,
+      "loss": 1.0103,
+      "step": 2914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8138008390037185e-05,
+      "loss": 1.0741,
+      "step": 2915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8136708558629187e-05,
+      "loss": 1.0649,
+      "step": 2916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.813540832029137e-05,
+      "loss": 0.8779,
+      "step": 2917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8134107675088778e-05,
+      "loss": 1.0305,
+      "step": 2918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.813280662308644e-05,
+      "loss": 0.9832,
+      "step": 2919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.813150516434944e-05,
+      "loss": 0.8908,
+      "step": 2920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8130203298942856e-05,
+      "loss": 1.047,
+      "step": 2921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8128901026931797e-05,
+      "loss": 1.0884,
+      "step": 2922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.81275983483814e-05,
+      "loss": 1.0715,
+      "step": 2923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.81262952633568e-05,
+      "loss": 0.9668,
+      "step": 2924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.812499177192318e-05,
+      "loss": 1.1076,
+      "step": 2925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.812368787414572e-05,
+      "loss": 1.0509,
+      "step": 2926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8122383570089638e-05,
+      "loss": 1.0033,
+      "step": 2927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8121078859820155e-05,
+      "loss": 0.9474,
+      "step": 2928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.811977374340253e-05,
+      "loss": 1.1378,
+      "step": 2929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8118468220902034e-05,
+      "loss": 1.0216,
+      "step": 2930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8117162292383948e-05,
+      "loss": 1.0448,
+      "step": 2931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8115855957913596e-05,
+      "loss": 1.0775,
+      "step": 2932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.81145492175563e-05,
+      "loss": 1.0039,
+      "step": 2933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8113242071377423e-05,
+      "loss": 1.0429,
+      "step": 2934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8111934519442326e-05,
+      "loss": 1.0561,
+      "step": 2935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.811062656181641e-05,
+      "loss": 0.9675,
+      "step": 2936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.810931819856508e-05,
+      "loss": 0.9072,
+      "step": 2937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.810800942975378e-05,
+      "loss": 0.9523,
+      "step": 2938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8106700255447956e-05,
+      "loss": 0.9873,
+      "step": 2939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.810539067571308e-05,
+      "loss": 1.0007,
+      "step": 2940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.810408069061466e-05,
+      "loss": 1.037,
+      "step": 2941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.810277030021819e-05,
+      "loss": 1.0927,
+      "step": 2942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.810145950458922e-05,
+      "loss": 1.116,
+      "step": 2943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.81001483037933e-05,
+      "loss": 0.9041,
+      "step": 2944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8098836697896004e-05,
+      "loss": 0.9615,
+      "step": 2945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809752468696293e-05,
+      "loss": 0.9673,
+      "step": 2946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8096212271059692e-05,
+      "loss": 1.0283,
+      "step": 2947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809489945025193e-05,
+      "loss": 0.9912,
+      "step": 2948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.809358622460529e-05,
+      "loss": 1.021,
+      "step": 2949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8092272594185464e-05,
+      "loss": 1.0632,
+      "step": 2950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8090958559058135e-05,
+      "loss": 0.979,
+      "step": 2951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8089644119289026e-05,
+      "loss": 1.0268,
+      "step": 2952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8088329274943873e-05,
+      "loss": 1.033,
+      "step": 2953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8087014026088435e-05,
+      "loss": 0.9844,
+      "step": 2954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808569837278849e-05,
+      "loss": 0.9335,
+      "step": 2955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8084382315109833e-05,
+      "loss": 0.9625,
+      "step": 2956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8083065853118283e-05,
+      "loss": 1.0166,
+      "step": 2957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.808174898687968e-05,
+      "loss": 1.1676,
+      "step": 2958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8080431716459883e-05,
+      "loss": 0.9628,
+      "step": 2959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807911404192477e-05,
+      "loss": 0.9798,
+      "step": 2960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8077795963340232e-05,
+      "loss": 1.0051,
+      "step": 2961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8076477480772207e-05,
+      "loss": 1.0712,
+      "step": 2962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8075158594286614e-05,
+      "loss": 1.0178,
+      "step": 2963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8073839303949428e-05,
+      "loss": 1.0125,
+      "step": 2964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8072519609826617e-05,
+      "loss": 0.988,
+      "step": 2965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.807119951198419e-05,
+      "loss": 0.9177,
+      "step": 2966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8069879010488166e-05,
+      "loss": 1.0536,
+      "step": 2967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.806855810540458e-05,
+      "loss": 0.9978,
+      "step": 2968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8067236796799496e-05,
+      "loss": 0.9375,
+      "step": 2969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8065915084738996e-05,
+      "loss": 0.9532,
+      "step": 2970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8064592969289178e-05,
+      "loss": 0.953,
+      "step": 2971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8063270450516165e-05,
+      "loss": 1.0437,
+      "step": 2972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8061947528486098e-05,
+      "loss": 1.0343,
+      "step": 2973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8060624203265136e-05,
+      "loss": 1.0501,
+      "step": 2974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8059300474919466e-05,
+      "loss": 1.0083,
+      "step": 2975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8057976343515286e-05,
+      "loss": 0.9968,
+      "step": 2976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805665180911882e-05,
+      "loss": 0.6685,
+      "step": 2977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8055326871796305e-05,
+      "loss": 0.9575,
+      "step": 2978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805400153161401e-05,
+      "loss": 0.9851,
+      "step": 2979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8052675788638214e-05,
+      "loss": 0.9506,
+      "step": 2980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8051349642935217e-05,
+      "loss": 0.9114,
+      "step": 2981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.805002309457135e-05,
+      "loss": 1.0839,
+      "step": 2982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8048696143612947e-05,
+      "loss": 0.9213,
+      "step": 2983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8047368790126375e-05,
+      "loss": 1.0135,
+      "step": 2984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8046041034178015e-05,
+      "loss": 1.0082,
+      "step": 2985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8044712875834273e-05,
+      "loss": 0.9755,
+      "step": 2986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804338431516157e-05,
+      "loss": 1.1028,
+      "step": 2987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804205535222635e-05,
+      "loss": 0.9247,
+      "step": 2988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.804072598709508e-05,
+      "loss": 1.0341,
+      "step": 2989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8039396219834236e-05,
+      "loss": 1.0664,
+      "step": 2990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803806605051033e-05,
+      "loss": 1.0111,
+      "step": 2991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803673547918988e-05,
+      "loss": 0.9937,
+      "step": 2992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803540450593943e-05,
+      "loss": 0.6984,
+      "step": 2993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803407313082555e-05,
+      "loss": 0.9225,
+      "step": 2994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803274135391482e-05,
+      "loss": 1.0073,
+      "step": 2995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8031409175273847e-05,
+      "loss": 0.9604,
+      "step": 2996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.803007659496925e-05,
+      "loss": 0.9635,
+      "step": 2997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802874361306768e-05,
+      "loss": 1.0909,
+      "step": 2998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8027410229635792e-05,
+      "loss": 0.9945,
+      "step": 2999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802607644474028e-05,
+      "loss": 1.0475,
+      "step": 3000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802474225844785e-05,
+      "loss": 0.9998,
+      "step": 3001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.802340767082522e-05,
+      "loss": 0.9975,
+      "step": 3002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8022072681939133e-05,
+      "loss": 0.9511,
+      "step": 3003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8020737291856362e-05,
+      "loss": 1.0561,
+      "step": 3004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8019401500643687e-05,
+      "loss": 1.0284,
+      "step": 3005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8018065308367915e-05,
+      "loss": 1.0791,
+      "step": 3006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801672871509587e-05,
+      "loss": 0.9685,
+      "step": 3007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8015391720894396e-05,
+      "loss": 0.7376,
+      "step": 3008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.801405432583036e-05,
+      "loss": 0.9446,
+      "step": 3009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8012716529970648e-05,
+      "loss": 1.0464,
+      "step": 3010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8011378333382164e-05,
+      "loss": 1.0164,
+      "step": 3011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8010039736131834e-05,
+      "loss": 1.0418,
+      "step": 3012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8008700738286605e-05,
+      "loss": 1.0755,
+      "step": 3013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8007361339913437e-05,
+      "loss": 1.1165,
+      "step": 3014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.800602154107932e-05,
+      "loss": 0.6728,
+      "step": 3015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.800468134185126e-05,
+      "loss": 1.0817,
+      "step": 3016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.800334074229628e-05,
+      "loss": 0.981,
+      "step": 3017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8001999742481426e-05,
+      "loss": 1.1704,
+      "step": 3018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8000658342473764e-05,
+      "loss": 1.0111,
+      "step": 3019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7999316542340383e-05,
+      "loss": 0.978,
+      "step": 3020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799797434214838e-05,
+      "loss": 0.8864,
+      "step": 3021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7996631741964888e-05,
+      "loss": 0.9941,
+      "step": 3022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799528874185705e-05,
+      "loss": 1.0018,
+      "step": 3023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7993945341892032e-05,
+      "loss": 1.0182,
+      "step": 3024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992601542137018e-05,
+      "loss": 0.9733,
+      "step": 3025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7991257342659217e-05,
+      "loss": 1.019,
+      "step": 3026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798991274352585e-05,
+      "loss": 1.0039,
+      "step": 3027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7988567744804167e-05,
+      "loss": 1.024,
+      "step": 3028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7987222346561433e-05,
+      "loss": 1.0204,
+      "step": 3029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798587654886493e-05,
+      "loss": 0.9375,
+      "step": 3030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7984530351781967e-05,
+      "loss": 0.9936,
+      "step": 3031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798318375537987e-05,
+      "loss": 1.0151,
+      "step": 3032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7981836759725978e-05,
+      "loss": 0.9774,
+      "step": 3033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7980489364887663e-05,
+      "loss": 1.09,
+      "step": 3034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797914157093231e-05,
+      "loss": 1.082,
+      "step": 3035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7977793377927324e-05,
+      "loss": 1.005,
+      "step": 3036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7976444785940127e-05,
+      "loss": 1.0264,
+      "step": 3037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7975095795038165e-05,
+      "loss": 0.9887,
+      "step": 3038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7973746405288907e-05,
+      "loss": 0.9953,
+      "step": 3039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7972396616759834e-05,
+      "loss": 1.0602,
+      "step": 3040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7971046429518448e-05,
+      "loss": 0.7172,
+      "step": 3041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7969695843632284e-05,
+      "loss": 1.0154,
+      "step": 3042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7968344859168882e-05,
+      "loss": 1.0151,
+      "step": 3043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7966993476195807e-05,
+      "loss": 1.0666,
+      "step": 3044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796564169478064e-05,
+      "loss": 1.0938,
+      "step": 3045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7964289514990994e-05,
+      "loss": 1.0158,
+      "step": 3046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7962936936894482e-05,
+      "loss": 1.0273,
+      "step": 3047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796158396055876e-05,
+      "loss": 1.012,
+      "step": 3048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7960230586051488e-05,
+      "loss": 0.996,
+      "step": 3049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795887681344035e-05,
+      "loss": 1.0204,
+      "step": 3050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7957522642793052e-05,
+      "loss": 0.9116,
+      "step": 3051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7956168074177313e-05,
+      "loss": 0.9798,
+      "step": 3052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.795481310766088e-05,
+      "loss": 1.0225,
+      "step": 3053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7953457743311526e-05,
+      "loss": 1.1078,
+      "step": 3054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.795210198119702e-05,
+      "loss": 1.0562,
+      "step": 3055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7950745821385173e-05,
+      "loss": 1.0341,
+      "step": 3056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.794938926394381e-05,
+      "loss": 0.954,
+      "step": 3057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7948032308940776e-05,
+      "loss": 0.9475,
+      "step": 3058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7946674956443926e-05,
+      "loss": 0.9932,
+      "step": 3059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7945317206521156e-05,
+      "loss": 0.9646,
+      "step": 3060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7943959059240357e-05,
+      "loss": 1.0928,
+      "step": 3061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.794260051466946e-05,
+      "loss": 1.0258,
+      "step": 3062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7941241572876405e-05,
+      "loss": 1.113,
+      "step": 3063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7939882233929156e-05,
+      "loss": 1.0502,
+      "step": 3064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7938522497895695e-05,
+      "loss": 0.7826,
+      "step": 3065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.793716236484402e-05,
+      "loss": 1.0171,
+      "step": 3066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7935801834842162e-05,
+      "loss": 1.0828,
+      "step": 3067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.793444090795816e-05,
+      "loss": 1.0852,
+      "step": 3068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7933079584260073e-05,
+      "loss": 0.9037,
+      "step": 3069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7931717863815987e-05,
+      "loss": 1.0428,
+      "step": 3070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7930355746694e-05,
+      "loss": 1.0023,
+      "step": 3071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.792899323296224e-05,
+      "loss": 0.9488,
+      "step": 3072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.792763032268884e-05,
+      "loss": 0.8875,
+      "step": 3073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7926267015941968e-05,
+      "loss": 1.0084,
+      "step": 3074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7924903312789803e-05,
+      "loss": 1.0535,
+      "step": 3075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7923539213300544e-05,
+      "loss": 0.9979,
+      "step": 3076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7922174717542414e-05,
+      "loss": 1.0905,
+      "step": 3077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7920809825583655e-05,
+      "loss": 0.9383,
+      "step": 3078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7919444537492523e-05,
+      "loss": 1.0157,
+      "step": 3079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7918078853337302e-05,
+      "loss": 0.8607,
+      "step": 3080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.791671277318629e-05,
+      "loss": 1.0086,
+      "step": 3081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.791534629710781e-05,
+      "loss": 0.9752,
+      "step": 3082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.79139794251702e-05,
+      "loss": 0.7417,
+      "step": 3083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7912612157441817e-05,
+      "loss": 1.0122,
+      "step": 3084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7911244493991044e-05,
+      "loss": 1.0222,
+      "step": 3085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7909876434886272e-05,
+      "loss": 0.6792,
+      "step": 3086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.790850798019593e-05,
+      "loss": 0.972,
+      "step": 3087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7907139129988454e-05,
+      "loss": 0.9191,
+      "step": 3088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.79057698843323e-05,
+      "loss": 1.0389,
+      "step": 3089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7904400243295948e-05,
+      "loss": 0.6631,
+      "step": 3090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7903030206947893e-05,
+      "loss": 1.0314,
+      "step": 3091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7901659775356656e-05,
+      "loss": 0.956,
+      "step": 3092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7900288948590777e-05,
+      "loss": 1.1061,
+      "step": 3093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7898917726718804e-05,
+      "loss": 0.706,
+      "step": 3094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789754610980932e-05,
+      "loss": 1.1215,
+      "step": 3095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7896174097930922e-05,
+      "loss": 0.9234,
+      "step": 3096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7894801691152227e-05,
+      "loss": 1.0115,
+      "step": 3097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.789342888954187e-05,
+      "loss": 1.0775,
+      "step": 3098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7892055693168506e-05,
+      "loss": 0.986,
+      "step": 3099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7890682102100815e-05,
+      "loss": 1.0225,
+      "step": 3100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788930811640748e-05,
+      "loss": 1.0429,
+      "step": 3101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7887933736157235e-05,
+      "loss": 0.9965,
+      "step": 3102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78865589614188e-05,
+      "loss": 0.9695,
+      "step": 3103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.788518379226094e-05,
+      "loss": 1.097,
+      "step": 3104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7883808228752423e-05,
+      "loss": 1.1412,
+      "step": 3105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7882432270962043e-05,
+      "loss": 0.8935,
+      "step": 3106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7881055918958615e-05,
+      "loss": 0.9426,
+      "step": 3107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7879679172810974e-05,
+      "loss": 1.0572,
+      "step": 3108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7878302032587977e-05,
+      "loss": 0.9632,
+      "step": 3109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7876924498358486e-05,
+      "loss": 0.9961,
+      "step": 3110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7875546570191404e-05,
+      "loss": 1.0011,
+      "step": 3111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7874168248155634e-05,
+      "loss": 1.0006,
+      "step": 3112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7872789532320118e-05,
+      "loss": 0.676,
+      "step": 3113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7871410422753802e-05,
+      "loss": 0.9611,
+      "step": 3114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.787003091952566e-05,
+      "loss": 1.0389,
+      "step": 3115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.786865102270468e-05,
+      "loss": 0.9433,
+      "step": 3116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7867270732359877e-05,
+      "loss": 1.0665,
+      "step": 3117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7865890048560277e-05,
+      "loss": 0.9423,
+      "step": 3118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7864508971374934e-05,
+      "loss": 1.0524,
+      "step": 3119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7863127500872915e-05,
+      "loss": 1.0908,
+      "step": 3120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7861745637123314e-05,
+      "loss": 1.0513,
+      "step": 3121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7860363380195232e-05,
+      "loss": 1.0612,
+      "step": 3122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7858980730157808e-05,
+      "loss": 0.8975,
+      "step": 3123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7857597687080182e-05,
+      "loss": 0.9617,
+      "step": 3124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.785621425103153e-05,
+      "loss": 1.0044,
+      "step": 3125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7854830422081033e-05,
+      "loss": 1.0244,
+      "step": 3126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.78534462002979e-05,
+      "loss": 0.898,
+      "step": 3127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7852061585751362e-05,
+      "loss": 1.1143,
+      "step": 3128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7850676578510665e-05,
+      "loss": 1.0627,
+      "step": 3129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7849291178645067e-05,
+      "loss": 1.1175,
+      "step": 3130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7847905386223865e-05,
+      "loss": 1.0641,
+      "step": 3131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7846519201316358e-05,
+      "loss": 1.0598,
+      "step": 3132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7845132623991873e-05,
+      "loss": 0.9714,
+      "step": 3133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7843745654319758e-05,
+      "loss": 1.071,
+      "step": 3134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784235829236938e-05,
+      "loss": 1.0064,
+      "step": 3135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.784097053821011e-05,
+      "loss": 0.9892,
+      "step": 3136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7839582391911362e-05,
+      "loss": 1.0177,
+      "step": 3137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7838193853542557e-05,
+      "loss": 0.9261,
+      "step": 3138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783680492317314e-05,
+      "loss": 1.0595,
+      "step": 3139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7835415600872572e-05,
+      "loss": 0.9284,
+      "step": 3140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7834025886710336e-05,
+      "loss": 0.9277,
+      "step": 3141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783263578075593e-05,
+      "loss": 1.021,
+      "step": 3142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.783124528307888e-05,
+      "loss": 1.0778,
+      "step": 3143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782985439374873e-05,
+      "loss": 0.9956,
+      "step": 3144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7828463112835034e-05,
+      "loss": 0.9767,
+      "step": 3145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7827071440407373e-05,
+      "loss": 1.0116,
+      "step": 3146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782567937653535e-05,
+      "loss": 0.9654,
+      "step": 3147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.782428692128858e-05,
+      "loss": 0.9473,
+      "step": 3148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7822894074736704e-05,
+      "loss": 0.9254,
+      "step": 3149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7821500836949385e-05,
+      "loss": 1.0071,
+      "step": 3150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7820107207996294e-05,
+      "loss": 0.9456,
+      "step": 3151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7818713187947133e-05,
+      "loss": 1.0033,
+      "step": 3152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7817318776871618e-05,
+      "loss": 0.9378,
+      "step": 3153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7815923974839485e-05,
+      "loss": 1.0132,
+      "step": 3154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781452878192049e-05,
+      "loss": 1.1611,
+      "step": 3155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7813133198184413e-05,
+      "loss": 1.0069,
+      "step": 3156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.781173722370104e-05,
+      "loss": 0.9072,
+      "step": 3157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7810340858540195e-05,
+      "loss": 0.9531,
+      "step": 3158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780894410277171e-05,
+      "loss": 0.9931,
+      "step": 3159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7807546956465438e-05,
+      "loss": 1.0317,
+      "step": 3160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7806149419691255e-05,
+      "loss": 0.9464,
+      "step": 3161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780475149251905e-05,
+      "loss": 1.11,
+      "step": 3162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7803353175018736e-05,
+      "loss": 0.6833,
+      "step": 3163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7801954467260248e-05,
+      "loss": 1.1224,
+      "step": 3164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7800555369313538e-05,
+      "loss": 1.1902,
+      "step": 3165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779915588124857e-05,
+      "loss": 1.0002,
+      "step": 3166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7797756003135344e-05,
+      "loss": 1.0332,
+      "step": 3167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7796355735043867e-05,
+      "loss": 0.96,
+      "step": 3168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794955077044167e-05,
+      "loss": 0.9111,
+      "step": 3169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779355402920629e-05,
+      "loss": 0.9371,
+      "step": 3170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7792152591600314e-05,
+      "loss": 0.9769,
+      "step": 3171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779075076429632e-05,
+      "loss": 1.0908,
+      "step": 3172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7789348547364417e-05,
+      "loss": 0.9281,
+      "step": 3173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778794594087473e-05,
+      "loss": 0.9922,
+      "step": 3174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778654294489741e-05,
+      "loss": 0.9877,
+      "step": 3175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7785139559502623e-05,
+      "loss": 1.0202,
+      "step": 3176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778373578476055e-05,
+      "loss": 0.6409,
+      "step": 3177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7782331620741403e-05,
+      "loss": 1.0173,
+      "step": 3178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7780927067515395e-05,
+      "loss": 0.6721,
+      "step": 3179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7779522125152782e-05,
+      "loss": 0.9268,
+      "step": 3180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7778116793723823e-05,
+      "loss": 0.9293,
+      "step": 3181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77767110732988e-05,
+      "loss": 0.9844,
+      "step": 3182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7775304963948014e-05,
+      "loss": 1.0192,
+      "step": 3183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777389846574179e-05,
+      "loss": 0.944,
+      "step": 3184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777249157875047e-05,
+      "loss": 0.9942,
+      "step": 3185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7771084303044407e-05,
+      "loss": 1.0393,
+      "step": 3186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7769676638693987e-05,
+      "loss": 0.971,
+      "step": 3187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7768268585769612e-05,
+      "loss": 1.0678,
+      "step": 3188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7766860144341697e-05,
+      "loss": 0.9312,
+      "step": 3189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776545131448068e-05,
+      "loss": 0.9752,
+      "step": 3190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7764042096257027e-05,
+      "loss": 1.0228,
+      "step": 3191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77626324897412e-05,
+      "loss": 0.9544,
+      "step": 3192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.776122249500371e-05,
+      "loss": 1.1033,
+      "step": 3193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7759812112115065e-05,
+      "loss": 0.6804,
+      "step": 3194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.77584013411458e-05,
+      "loss": 0.7316,
+      "step": 3195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.775699018216648e-05,
+      "loss": 0.9602,
+      "step": 3196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7755578635247666e-05,
+      "loss": 0.9812,
+      "step": 3197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7754166700459958e-05,
+      "loss": 1.0238,
+      "step": 3198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7752754377873972e-05,
+      "loss": 0.9476,
+      "step": 3199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7751341667560336e-05,
+      "loss": 0.9734,
+      "step": 3200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.77499285695897e-05,
+      "loss": 0.9076,
+      "step": 3201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7748515084032742e-05,
+      "loss": 0.9934,
+      "step": 3202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.774710121096015e-05,
+      "loss": 1.0691,
+      "step": 3203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.774568695044263e-05,
+      "loss": 0.9702,
+      "step": 3204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7744272302550918e-05,
+      "loss": 1.0833,
+      "step": 3205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7742857267355756e-05,
+      "loss": 1.056,
+      "step": 3206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.774144184492792e-05,
+      "loss": 1.0026,
+      "step": 3207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7740026035338195e-05,
+      "loss": 0.6837,
+      "step": 3208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7738609838657378e-05,
+      "loss": 1.035,
+      "step": 3209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7737193254956312e-05,
+      "loss": 0.9437,
+      "step": 3210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7735776284305828e-05,
+      "loss": 0.8647,
+      "step": 3211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.77343589267768e-05,
+      "loss": 1.0501,
+      "step": 3212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.773294118244011e-05,
+      "loss": 0.9426,
+      "step": 3213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.773152305136666e-05,
+      "loss": 0.9019,
+      "step": 3214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.773010453362737e-05,
+      "loss": 0.8531,
+      "step": 3215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772868562929319e-05,
+      "loss": 0.971,
+      "step": 3216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772726633843508e-05,
+      "loss": 1.0128,
+      "step": 3217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7725846661124016e-05,
+      "loss": 1.0241,
+      "step": 3218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7724426597431e-05,
+      "loss": 0.9982,
+      "step": 3219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7723006147427056e-05,
+      "loss": 1.0006,
+      "step": 3220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772158531118322e-05,
+      "loss": 1.0322,
+      "step": 3221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.772016408877055e-05,
+      "loss": 1.001,
+      "step": 3222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7718742480260124e-05,
+      "loss": 1.0868,
+      "step": 3223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7717320485723036e-05,
+      "loss": 1.0529,
+      "step": 3224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.771589810523041e-05,
+      "loss": 0.9865,
+      "step": 3225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7714475338853372e-05,
+      "loss": 1.1024,
+      "step": 3226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7713052186663083e-05,
+      "loss": 1.067,
+      "step": 3227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7711628648730718e-05,
+      "loss": 0.9504,
+      "step": 3228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7710204725127464e-05,
+      "loss": 1.0113,
+      "step": 3229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770878041592454e-05,
+      "loss": 0.9378,
+      "step": 3230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7707355721193175e-05,
+      "loss": 0.999,
+      "step": 3231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7705930641004622e-05,
+      "loss": 0.9899,
+      "step": 3232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7704505175430146e-05,
+      "loss": 1.0352,
+      "step": 3233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7703079324541046e-05,
+      "loss": 0.9985,
+      "step": 3234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7701653088408625e-05,
+      "loss": 0.9646,
+      "step": 3235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.770022646710421e-05,
+      "loss": 1.0647,
+      "step": 3236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7698799460699155e-05,
+      "loss": 1.0459,
+      "step": 3237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769737206926482e-05,
+      "loss": 0.9863,
+      "step": 3238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7695944292872597e-05,
+      "loss": 0.6606,
+      "step": 3239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769451613159389e-05,
+      "loss": 1.0109,
+      "step": 3240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7693087585500116e-05,
+      "loss": 1.0028,
+      "step": 3241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7691658654662733e-05,
+      "loss": 0.6799,
+      "step": 3242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.769022933915319e-05,
+      "loss": 0.932,
+      "step": 3243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7688799639042977e-05,
+      "loss": 0.957,
+      "step": 3244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7687369554403594e-05,
+      "loss": 0.6921,
+      "step": 3245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768593908530656e-05,
+      "loss": 0.8827,
+      "step": 3246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7684508231823422e-05,
+      "loss": 1.0322,
+      "step": 3247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7683076994025735e-05,
+      "loss": 0.9779,
+      "step": 3248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7681645371985073e-05,
+      "loss": 1.1238,
+      "step": 3249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.768021336577304e-05,
+      "loss": 0.9293,
+      "step": 3250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7678780975461248e-05,
+      "loss": 1.0698,
+      "step": 3251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7677348201121335e-05,
+      "loss": 0.9487,
+      "step": 3252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7675915042824958e-05,
+      "loss": 0.9583,
+      "step": 3253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7674481500643793e-05,
+      "loss": 1.0265,
+      "step": 3254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767304757464953e-05,
+      "loss": 0.8811,
+      "step": 3255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7671613264913882e-05,
+      "loss": 0.9758,
+      "step": 3256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.767017857150858e-05,
+      "loss": 1.1669,
+      "step": 3257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7668743494505383e-05,
+      "loss": 1.0221,
+      "step": 3258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7667308033976055e-05,
+      "loss": 1.0119,
+      "step": 3259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7665872189992383e-05,
+      "loss": 1.0009,
+      "step": 3260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7664435962626184e-05,
+      "loss": 1.0468,
+      "step": 3261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7662999351949278e-05,
+      "loss": 0.8998,
+      "step": 3262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.766156235803352e-05,
+      "loss": 1.0148,
+      "step": 3263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7660124980950767e-05,
+      "loss": 0.6999,
+      "step": 3264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7658687220772912e-05,
+      "loss": 1.0015,
+      "step": 3265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7657249077571855e-05,
+      "loss": 0.946,
+      "step": 3266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7655810551419525e-05,
+      "loss": 0.9216,
+      "step": 3267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765437164238786e-05,
+      "loss": 0.948,
+      "step": 3268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765293235054883e-05,
+      "loss": 1.0305,
+      "step": 3269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.765149267597441e-05,
+      "loss": 0.9793,
+      "step": 3270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7650052618736596e-05,
+      "loss": 0.9062,
+      "step": 3271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7648612178907414e-05,
+      "loss": 0.9574,
+      "step": 3272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7647171356558902e-05,
+      "loss": 0.9676,
+      "step": 3273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.764573015176312e-05,
+      "loss": 1.0923,
+      "step": 3274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.764428856459214e-05,
+      "loss": 1.0092,
+      "step": 3275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7642846595118057e-05,
+      "loss": 1.0185,
+      "step": 3276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7641404243412994e-05,
+      "loss": 1.0031,
+      "step": 3277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763996150954908e-05,
+      "loss": 1.0401,
+      "step": 3278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7638518393598463e-05,
+      "loss": 0.6583,
+      "step": 3279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763707489563333e-05,
+      "loss": 0.9467,
+      "step": 3280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763563101572586e-05,
+      "loss": 1.0238,
+      "step": 3281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.763418675394827e-05,
+      "loss": 1.1477,
+      "step": 3282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7632742110372787e-05,
+      "loss": 1.0184,
+      "step": 3283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7631297085071663e-05,
+      "loss": 0.9632,
+      "step": 3284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.762985167811716e-05,
+      "loss": 0.8358,
+      "step": 3285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7628405889581573e-05,
+      "loss": 0.9843,
+      "step": 3286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7626959719537203e-05,
+      "loss": 1.0911,
+      "step": 3287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7625513168056373e-05,
+      "loss": 1.0692,
+      "step": 3288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7624066235211437e-05,
+      "loss": 1.0442,
+      "step": 3289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.762261892107475e-05,
+      "loss": 1.0798,
+      "step": 3290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7621171225718692e-05,
+      "loss": 0.7048,
+      "step": 3291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.761972314921567e-05,
+      "loss": 1.1024,
+      "step": 3292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.761827469163811e-05,
+      "loss": 1.0602,
+      "step": 3293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7616825853058444e-05,
+      "loss": 0.9799,
+      "step": 3294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.761537663354913e-05,
+      "loss": 0.9833,
+      "step": 3295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.761392703318265e-05,
+      "loss": 1.0312,
+      "step": 3296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7612477052031497e-05,
+      "loss": 1.0012,
+      "step": 3297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.761102669016819e-05,
+      "loss": 1.0727,
+      "step": 3298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7609575947665262e-05,
+      "loss": 0.9344,
+      "step": 3299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7608124824595272e-05,
+      "loss": 1.0179,
+      "step": 3300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7606673321030784e-05,
+      "loss": 1.0186,
+      "step": 3301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7605221437044396e-05,
+      "loss": 0.9275,
+      "step": 3302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7603769172708718e-05,
+      "loss": 0.6766,
+      "step": 3303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7602316528096382e-05,
+      "loss": 0.9759,
+      "step": 3304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.760086350328003e-05,
+      "loss": 1.0335,
+      "step": 3305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759941009833234e-05,
+      "loss": 0.9933,
+      "step": 3306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759795631332599e-05,
+      "loss": 0.9842,
+      "step": 3307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7596502148333693e-05,
+      "loss": 1.0244,
+      "step": 3308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759504760342817e-05,
+      "loss": 0.9774,
+      "step": 3309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7593592678682166e-05,
+      "loss": 0.9331,
+      "step": 3310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7592137374168444e-05,
+      "loss": 0.978,
+      "step": 3311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759068168995979e-05,
+      "loss": 0.6945,
+      "step": 3312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7589225626129e-05,
+      "loss": 0.9569,
+      "step": 3313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7587769182748896e-05,
+      "loss": 1.0424,
+      "step": 3314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7586312359892316e-05,
+      "loss": 0.9198,
+      "step": 3315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7584855157632122e-05,
+      "loss": 0.9813,
+      "step": 3316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7583397576041185e-05,
+      "loss": 0.9962,
+      "step": 3317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.75819396151924e-05,
+      "loss": 0.9059,
+      "step": 3318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7580481275158697e-05,
+      "loss": 1.0105,
+      "step": 3319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757902255601299e-05,
+      "loss": 0.9768,
+      "step": 3320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7577563457828245e-05,
+      "loss": 1.042,
+      "step": 3321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7576103980677427e-05,
+      "loss": 0.9801,
+      "step": 3322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7574644124633532e-05,
+      "loss": 1.0019,
+      "step": 3323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7573183889769564e-05,
+      "loss": 0.7001,
+      "step": 3324
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7571723276158554e-05,
+      "loss": 0.931,
+      "step": 3325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757026228387355e-05,
+      "loss": 0.9541,
+      "step": 3326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7568800912987625e-05,
+      "loss": 1.0403,
+      "step": 3327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7567339163573853e-05,
+      "loss": 0.9346,
+      "step": 3328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7565877035705347e-05,
+      "loss": 1.0324,
+      "step": 3329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7564414529455226e-05,
+      "loss": 0.9833,
+      "step": 3330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7562951644896632e-05,
+      "loss": 1.0324,
+      "step": 3331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756148838210273e-05,
+      "loss": 1.1099,
+      "step": 3332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7560024741146697e-05,
+      "loss": 0.9701,
+      "step": 3333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7558560722101732e-05,
+      "loss": 0.9454,
+      "step": 3334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7557096325041058e-05,
+      "loss": 0.9782,
+      "step": 3335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7555631550037903e-05,
+      "loss": 0.9881,
+      "step": 3336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7554166397165525e-05,
+      "loss": 1.0395,
+      "step": 3337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7552700866497208e-05,
+      "loss": 1.0217,
+      "step": 3338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7551234958106236e-05,
+      "loss": 0.9745,
+      "step": 3339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.754976867206592e-05,
+      "loss": 0.7085,
+      "step": 3340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.75483020084496e-05,
+      "loss": 0.9756,
+      "step": 3341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.754683496733062e-05,
+      "loss": 0.6951,
+      "step": 3342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.754536754878235e-05,
+      "loss": 0.9595,
+      "step": 3343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7543899752878178e-05,
+      "loss": 0.6833,
+      "step": 3344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7542431579691512e-05,
+      "loss": 0.9759,
+      "step": 3345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7540963029295775e-05,
+      "loss": 1.1388,
+      "step": 3346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.753949410176442e-05,
+      "loss": 1.0152,
+      "step": 3347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.75380247971709e-05,
+      "loss": 0.9109,
+      "step": 3348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.75365551155887e-05,
+      "loss": 0.9737,
+      "step": 3349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7535085057091325e-05,
+      "loss": 0.9191,
+      "step": 3350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.753361462175229e-05,
+      "loss": 1.0524,
+      "step": 3351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7532143809645135e-05,
+      "loss": 0.9144,
+      "step": 3352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7530672620843422e-05,
+      "loss": 1.0821,
+      "step": 3353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7529201055420723e-05,
+      "loss": 0.9116,
+      "step": 3354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.752772911345063e-05,
+      "loss": 0.9967,
+      "step": 3355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.752625679500677e-05,
+      "loss": 1.0709,
+      "step": 3356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7524784100162762e-05,
+      "loss": 0.9988,
+      "step": 3357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.752331102899227e-05,
+      "loss": 0.9766,
+      "step": 3358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7521837581568952e-05,
+      "loss": 1.0411,
+      "step": 3359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.752036375796651e-05,
+      "loss": 1.0417,
+      "step": 3360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.751888955825864e-05,
+      "loss": 0.952,
+      "step": 3361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7517414982519073e-05,
+      "loss": 0.9454,
+      "step": 3362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7515940030821563e-05,
+      "loss": 0.8992,
+      "step": 3363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.751446470323987e-05,
+      "loss": 0.9769,
+      "step": 3364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7512988999847774e-05,
+      "loss": 1.0208,
+      "step": 3365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7511512920719077e-05,
+      "loss": 0.7078,
+      "step": 3366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.75100364659276e-05,
+      "loss": 0.9337,
+      "step": 3367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7508559635547192e-05,
+      "loss": 0.959,
+      "step": 3368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7507082429651704e-05,
+      "loss": 0.9877,
+      "step": 3369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7505604848315014e-05,
+      "loss": 0.9889,
+      "step": 3370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.750412689161102e-05,
+      "loss": 0.9014,
+      "step": 3371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7502648559613633e-05,
+      "loss": 1.0053,
+      "step": 3372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.750116985239679e-05,
+      "loss": 0.9933,
+      "step": 3373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7499690770034444e-05,
+      "loss": 0.9492,
+      "step": 3374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7498211312600563e-05,
+      "loss": 1.0254,
+      "step": 3375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.749673148016914e-05,
+      "loss": 1.0151,
+      "step": 3376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7495251272814185e-05,
+      "loss": 0.9299,
+      "step": 3377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7493770690609723e-05,
+      "loss": 1.0406,
+      "step": 3378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7492289733629803e-05,
+      "loss": 0.9678,
+      "step": 3379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.749080840194848e-05,
+      "loss": 0.9952,
+      "step": 3380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7489326695639855e-05,
+      "loss": 0.9463,
+      "step": 3381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7487844614778017e-05,
+      "loss": 1.0106,
+      "step": 3382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.748636215943709e-05,
+      "loss": 1.0257,
+      "step": 3383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.748487932969122e-05,
+      "loss": 1.0544,
+      "step": 3384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7483396125614554e-05,
+      "loss": 0.962,
+      "step": 3385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7481912547281285e-05,
+      "loss": 1.0583,
+      "step": 3386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7480428594765596e-05,
+      "loss": 1.0394,
+      "step": 3387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7478944268141707e-05,
+      "loss": 0.9767,
+      "step": 3388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.747745956748385e-05,
+      "loss": 0.9795,
+      "step": 3389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.747597449286628e-05,
+      "loss": 0.9283,
+      "step": 3390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7474489044363266e-05,
+      "loss": 0.9868,
+      "step": 3391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7473003222049096e-05,
+      "loss": 0.9729,
+      "step": 3392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.747151702599808e-05,
+      "loss": 0.6746,
+      "step": 3393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.747003045628455e-05,
+      "loss": 0.9349,
+      "step": 3394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.746854351298284e-05,
+      "loss": 1.0258,
+      "step": 3395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7467056196167325e-05,
+      "loss": 0.9498,
+      "step": 3396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7465568505912386e-05,
+      "loss": 0.9872,
+      "step": 3397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7464080442292422e-05,
+      "loss": 0.8847,
+      "step": 3398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7462592005381857e-05,
+      "loss": 1.0711,
+      "step": 3399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7461103195255124e-05,
+      "loss": 0.7115,
+      "step": 3400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7459614011986685e-05,
+      "loss": 0.8551,
+      "step": 3401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7458124455651017e-05,
+      "loss": 0.9395,
+      "step": 3402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7456634526322615e-05,
+      "loss": 1.0238,
+      "step": 3403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7455144224075992e-05,
+      "loss": 0.929,
+      "step": 3404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7453653548985675e-05,
+      "loss": 1.0045,
+      "step": 3405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7452162501126228e-05,
+      "loss": 1.0551,
+      "step": 3406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.745067108057221e-05,
+      "loss": 1.0834,
+      "step": 3407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7449179287398213e-05,
+      "loss": 0.9264,
+      "step": 3408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.744768712167884e-05,
+      "loss": 0.9302,
+      "step": 3409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.744619458348872e-05,
+      "loss": 0.941,
+      "step": 3410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7444701672902504e-05,
+      "loss": 1.0393,
+      "step": 3411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7443208389994844e-05,
+      "loss": 1.0081,
+      "step": 3412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7441714734840422e-05,
+      "loss": 1.0451,
+      "step": 3413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7440220707513943e-05,
+      "loss": 1.0011,
+      "step": 3414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.743872630809013e-05,
+      "loss": 0.9494,
+      "step": 3415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.743723153664371e-05,
+      "loss": 0.9602,
+      "step": 3416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.743573639324944e-05,
+      "loss": 0.977,
+      "step": 3417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7434240877982104e-05,
+      "loss": 1.0214,
+      "step": 3418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7432744990916486e-05,
+      "loss": 1.0322,
+      "step": 3419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.74312487321274e-05,
+      "loss": 1.0109,
+      "step": 3420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.742975210168968e-05,
+      "loss": 0.9396,
+      "step": 3421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7428255099678167e-05,
+      "loss": 0.987,
+      "step": 3422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.742675772616774e-05,
+      "loss": 1.1289,
+      "step": 3423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7425259981233275e-05,
+      "loss": 0.9829,
+      "step": 3424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7423761864949677e-05,
+      "loss": 0.9195,
+      "step": 3425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7422263377391877e-05,
+      "loss": 1.058,
+      "step": 3426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.742076451863481e-05,
+      "loss": 1.0285,
+      "step": 3427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.741926528875344e-05,
+      "loss": 0.9734,
+      "step": 3428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7417765687822743e-05,
+      "loss": 1.0866,
+      "step": 3429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7416265715917718e-05,
+      "loss": 0.9676,
+      "step": 3430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7414765373113377e-05,
+      "loss": 0.988,
+      "step": 3431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.741326465948476e-05,
+      "loss": 0.9639,
+      "step": 3432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7411763575106925e-05,
+      "loss": 1.0136,
+      "step": 3433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.741026212005493e-05,
+      "loss": 0.9516,
+      "step": 3434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7408760294403877e-05,
+      "loss": 1.1687,
+      "step": 3435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7407258098228866e-05,
+      "loss": 0.6573,
+      "step": 3436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7405755531605025e-05,
+      "loss": 0.9708,
+      "step": 3437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.740425259460751e-05,
+      "loss": 0.9248,
+      "step": 3438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7402749287311474e-05,
+      "loss": 1.0589,
+      "step": 3439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7401245609792106e-05,
+      "loss": 0.9147,
+      "step": 3440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73997415621246e-05,
+      "loss": 0.9029,
+      "step": 3441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7398237144384186e-05,
+      "loss": 0.945,
+      "step": 3442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73967323566461e-05,
+      "loss": 1.0207,
+      "step": 3443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7395227198985593e-05,
+      "loss": 1.0826,
+      "step": 3444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7393721671477942e-05,
+      "loss": 0.962,
+      "step": 3445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.739221577419844e-05,
+      "loss": 1.0419,
+      "step": 3446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7390709507222405e-05,
+      "loss": 0.9586,
+      "step": 3447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7389202870625168e-05,
+      "loss": 0.9854,
+      "step": 3448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.738769586448207e-05,
+      "loss": 0.9073,
+      "step": 3449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7386188488868484e-05,
+      "loss": 0.9968,
+      "step": 3450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73846807438598e-05,
+      "loss": 1.0533,
+      "step": 3451
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7383172629531414e-05,
+      "loss": 0.9341,
+      "step": 3452
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7381664145958757e-05,
+      "loss": 0.9718,
+      "step": 3453
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7380155293217264e-05,
+      "loss": 0.9866,
+      "step": 3454
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73786460713824e-05,
+      "loss": 0.6985,
+      "step": 3455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7377136480529647e-05,
+      "loss": 1.1159,
+      "step": 3456
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.737562652073449e-05,
+      "loss": 1.0398,
+      "step": 3457
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7374116192072457e-05,
+      "loss": 0.9917,
+      "step": 3458
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7372605494619076e-05,
+      "loss": 1.0425,
+      "step": 3459
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73710944284499e-05,
+      "loss": 0.9938,
+      "step": 3460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7369582993640498e-05,
+      "loss": 1.0759,
+      "step": 3461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736807119026646e-05,
+      "loss": 1.0219,
+      "step": 3462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73665590184034e-05,
+      "loss": 1.0073,
+      "step": 3463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7365046478126934e-05,
+      "loss": 0.9431,
+      "step": 3464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7363533569512717e-05,
+      "loss": 0.6294,
+      "step": 3465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7362020292636404e-05,
+      "loss": 1.0411,
+      "step": 3466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7360506647573677e-05,
+      "loss": 1.0125,
+      "step": 3467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735899263440024e-05,
+      "loss": 1.0315,
+      "step": 3468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735747825319181e-05,
+      "loss": 1.0987,
+      "step": 3469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7355963504024123e-05,
+      "loss": 0.852,
+      "step": 3470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735444838697293e-05,
+      "loss": 0.9269,
+      "step": 3471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735293290211401e-05,
+      "loss": 1.0247,
+      "step": 3472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7351417049523155e-05,
+      "loss": 0.9815,
+      "step": 3473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734990082927617e-05,
+      "loss": 0.9892,
+      "step": 3474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734838424144889e-05,
+      "loss": 0.9452,
+      "step": 3475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734686728611716e-05,
+      "loss": 1.0021,
+      "step": 3476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.734534996335684e-05,
+      "loss": 1.0098,
+      "step": 3477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.734383227324382e-05,
+      "loss": 0.9431,
+      "step": 3478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7342314215854e-05,
+      "loss": 1.0372,
+      "step": 3479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.73407957912633e-05,
+      "loss": 0.9533,
+      "step": 3480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.733927699954766e-05,
+      "loss": 0.9863,
+      "step": 3481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7337757840783034e-05,
+      "loss": 0.6741,
+      "step": 3482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7336238315045403e-05,
+      "loss": 1.0174,
+      "step": 3483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7334718422410758e-05,
+      "loss": 0.9819,
+      "step": 3484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7333198162955112e-05,
+      "loss": 1.0749,
+      "step": 3485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.733167753675449e-05,
+      "loss": 1.0277,
+      "step": 3486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.733015654388495e-05,
+      "loss": 1.03,
+      "step": 3487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7328635184422557e-05,
+      "loss": 1.043,
+      "step": 3488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7327113458443387e-05,
+      "loss": 0.9864,
+      "step": 3489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.732559136602356e-05,
+      "loss": 0.9527,
+      "step": 3490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7324068907239183e-05,
+      "loss": 1.0265,
+      "step": 3491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.732254608216641e-05,
+      "loss": 0.9575,
+      "step": 3492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7321022890881392e-05,
+      "loss": 1.0105,
+      "step": 3493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7319499333460303e-05,
+      "loss": 0.9454,
+      "step": 3494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.731797540997935e-05,
+      "loss": 1.0156,
+      "step": 3495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7316451120514735e-05,
+      "loss": 1.0007,
+      "step": 3496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7314926465142695e-05,
+      "loss": 0.8991,
+      "step": 3497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7313401443939484e-05,
+      "loss": 0.9293,
+      "step": 3498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7311876056981364e-05,
+      "loss": 0.91,
+      "step": 3499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.731035030434463e-05,
+      "loss": 0.996,
+      "step": 3500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.730882418610558e-05,
+      "loss": 0.8702,
+      "step": 3501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.730729770234054e-05,
+      "loss": 1.0005,
+      "step": 3502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7305770853125853e-05,
+      "loss": 0.9737,
+      "step": 3503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7304243638537878e-05,
+      "loss": 0.67,
+      "step": 3504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7302716058652996e-05,
+      "loss": 1.1197,
+      "step": 3505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.73011881135476e-05,
+      "loss": 0.6832,
+      "step": 3506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7299659803298106e-05,
+      "loss": 1.0527,
+      "step": 3507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.729813112798095e-05,
+      "loss": 1.0175,
+      "step": 3508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.729660208767258e-05,
+      "loss": 1.019,
+      "step": 3509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.729507268244947e-05,
+      "loss": 1.0108,
+      "step": 3510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7293542912388103e-05,
+      "loss": 1.0423,
+      "step": 3511
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7292012777564986e-05,
+      "loss": 0.9976,
+      "step": 3512
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7290482278056645e-05,
+      "loss": 0.9654,
+      "step": 3513
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7288951413939624e-05,
+      "loss": 1.0264,
+      "step": 3514
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.728742018529048e-05,
+      "loss": 0.9177,
+      "step": 3515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.72858885921858e-05,
+      "loss": 1.0642,
+      "step": 3516
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7284356634702173e-05,
+      "loss": 1.0435,
+      "step": 3517
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7282824312916216e-05,
+      "loss": 1.011,
+      "step": 3518
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7281291626904567e-05,
+      "loss": 0.9767,
+      "step": 3519
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7279758576743875e-05,
+      "loss": 1.0084,
+      "step": 3520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7278225162510815e-05,
+      "loss": 1.0838,
+      "step": 3521
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7276691384282066e-05,
+      "loss": 0.9512,
+      "step": 3522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7275157242134342e-05,
+      "loss": 0.9177,
+      "step": 3523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7273622736144368e-05,
+      "loss": 1.0363,
+      "step": 3524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.727208786638888e-05,
+      "loss": 0.9547,
+      "step": 3525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.727055263294465e-05,
+      "loss": 0.8703,
+      "step": 3526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.726901703588845e-05,
+      "loss": 1.0578,
+      "step": 3527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.726748107529708e-05,
+      "loss": 1.0868,
+      "step": 3528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7265944751247355e-05,
+      "loss": 1.0119,
+      "step": 3529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.726440806381611e-05,
+      "loss": 1.0152,
+      "step": 3530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.72628710130802e-05,
+      "loss": 1.0406,
+      "step": 3531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7261333599116487e-05,
+      "loss": 1.0231,
+      "step": 3532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7259795822001867e-05,
+      "loss": 0.7134,
+      "step": 3533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7258257681813244e-05,
+      "loss": 0.6549,
+      "step": 3534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7256719178627545e-05,
+      "loss": 0.9644,
+      "step": 3535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.725518031252171e-05,
+      "loss": 1.0879,
+      "step": 3536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7253641083572698e-05,
+      "loss": 0.9958,
+      "step": 3537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7252101491857497e-05,
+      "loss": 1.0165,
+      "step": 3538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7250561537453095e-05,
+      "loss": 0.996,
+      "step": 3539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7249021220436514e-05,
+      "loss": 1.0656,
+      "step": 3540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7247480540884787e-05,
+      "loss": 1.0678,
+      "step": 3541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7245939498874962e-05,
+      "loss": 1.0064,
+      "step": 3542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7244398094484108e-05,
+      "loss": 1.0792,
+      "step": 3543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.724285632778932e-05,
+      "loss": 1.0247,
+      "step": 3544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7241314198867702e-05,
+      "loss": 1.1006,
+      "step": 3545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.723977170779638e-05,
+      "loss": 0.9636,
+      "step": 3546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7238228854652487e-05,
+      "loss": 1.1036,
+      "step": 3547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.723668563951319e-05,
+      "loss": 1.0299,
+      "step": 3548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7235142062455668e-05,
+      "loss": 1.1001,
+      "step": 3549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.723359812355712e-05,
+      "loss": 1.0882,
+      "step": 3550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7232053822894757e-05,
+      "loss": 1.1108,
+      "step": 3551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7230509160545813e-05,
+      "loss": 0.9949,
+      "step": 3552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7228964136587538e-05,
+      "loss": 1.0743,
+      "step": 3553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7227418751097206e-05,
+      "loss": 1.0287,
+      "step": 3554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7225873004152095e-05,
+      "loss": 0.9936,
+      "step": 3555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7224326895829517e-05,
+      "loss": 1.0023,
+      "step": 3556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.72227804262068e-05,
+      "loss": 0.9219,
+      "step": 3557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7221233595361272e-05,
+      "loss": 0.928,
+      "step": 3558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7219686403370305e-05,
+      "loss": 0.7055,
+      "step": 3559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7218138850311267e-05,
+      "loss": 0.942,
+      "step": 3560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7216590936261563e-05,
+      "loss": 0.8952,
+      "step": 3561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.72150426612986e-05,
+      "loss": 1.0805,
+      "step": 3562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.721349402549981e-05,
+      "loss": 0.9488,
+      "step": 3563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.721194502894265e-05,
+      "loss": 1.0623,
+      "step": 3564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.721039567170457e-05,
+      "loss": 0.9637,
+      "step": 3565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7208845953863077e-05,
+      "loss": 0.984,
+      "step": 3566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7207295875495667e-05,
+      "loss": 0.8591,
+      "step": 3567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.720574543667986e-05,
+      "loss": 0.9325,
+      "step": 3568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7204194637493192e-05,
+      "loss": 1.0252,
+      "step": 3569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7202643478013226e-05,
+      "loss": 0.9832,
+      "step": 3570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.720109195831754e-05,
+      "loss": 0.8961,
+      "step": 3571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7199540078483728e-05,
+      "loss": 1.0763,
+      "step": 3572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7197987838589396e-05,
+      "loss": 0.95,
+      "step": 3573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7196435238712176e-05,
+      "loss": 1.0059,
+      "step": 3574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7194882278929723e-05,
+      "loss": 0.9317,
+      "step": 3575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7193328959319694e-05,
+      "loss": 0.9391,
+      "step": 3576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7191775279959773e-05,
+      "loss": 0.919,
+      "step": 3577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.719022124092767e-05,
+      "loss": 1.0155,
+      "step": 3578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7188666842301103e-05,
+      "loss": 0.9216,
+      "step": 3579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7187112084157802e-05,
+      "loss": 0.8729,
+      "step": 3580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.718555696657553e-05,
+      "loss": 1.0846,
+      "step": 3581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.718400148963206e-05,
+      "loss": 1.0784,
+      "step": 3582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7182445653405183e-05,
+      "loss": 0.9935,
+      "step": 3583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.718088945797271e-05,
+      "loss": 0.9144,
+      "step": 3584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7179332903412467e-05,
+      "loss": 1.0628,
+      "step": 3585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71777759898023e-05,
+      "loss": 0.8847,
+      "step": 3586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7176218717220075e-05,
+      "loss": 0.9206,
+      "step": 3587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7174661085743668e-05,
+      "loss": 1.0731,
+      "step": 3588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7173103095450986e-05,
+      "loss": 0.9515,
+      "step": 3589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7171544746419945e-05,
+      "loss": 0.9763,
+      "step": 3590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7169986038728476e-05,
+      "loss": 0.9821,
+      "step": 3591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7168426972454536e-05,
+      "loss": 1.0439,
+      "step": 3592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7166867547676097e-05,
+      "loss": 0.9281,
+      "step": 3593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7165307764471144e-05,
+      "loss": 0.9637,
+      "step": 3594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716374762291769e-05,
+      "loss": 1.0984,
+      "step": 3595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7162187123093756e-05,
+      "loss": 1.0005,
+      "step": 3596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7160626265077386e-05,
+      "loss": 0.8501,
+      "step": 3597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7159065048946643e-05,
+      "loss": 0.9963,
+      "step": 3598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7157503474779607e-05,
+      "loss": 0.9485,
+      "step": 3599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715594154265437e-05,
+      "loss": 1.0349,
+      "step": 3600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7154379252649046e-05,
+      "loss": 0.9909,
+      "step": 3601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7152816604841772e-05,
+      "loss": 1.0084,
+      "step": 3602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71512535993107e-05,
+      "loss": 1.1938,
+      "step": 3603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7149690236133994e-05,
+      "loss": 0.8856,
+      "step": 3604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7148126515389842e-05,
+      "loss": 1.0133,
+      "step": 3605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7146562437156447e-05,
+      "loss": 1.0316,
+      "step": 3606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7144998001512032e-05,
+      "loss": 0.6638,
+      "step": 3607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714343320853484e-05,
+      "loss": 0.9744,
+      "step": 3608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7141868058303123e-05,
+      "loss": 1.1092,
+      "step": 3609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714030255089516e-05,
+      "loss": 1.0461,
+      "step": 3610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7138736686389243e-05,
+      "loss": 0.9417,
+      "step": 3611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7137170464863686e-05,
+      "loss": 0.9154,
+      "step": 3612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7135603886396815e-05,
+      "loss": 1.0137,
+      "step": 3613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713403695106698e-05,
+      "loss": 0.9692,
+      "step": 3614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7132469658952545e-05,
+      "loss": 0.68,
+      "step": 3615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7130902010131894e-05,
+      "loss": 1.0132,
+      "step": 3616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7129334004683423e-05,
+      "loss": 0.978,
+      "step": 3617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7127765642685554e-05,
+      "loss": 1.03,
+      "step": 3618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7126196924216725e-05,
+      "loss": 0.9489,
+      "step": 3619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7124627849355385e-05,
+      "loss": 1.0288,
+      "step": 3620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.712305841818001e-05,
+      "loss": 1.0528,
+      "step": 3621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.712148863076909e-05,
+      "loss": 0.9043,
+      "step": 3622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.711991848720113e-05,
+      "loss": 0.9709,
+      "step": 3623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.711834798755466e-05,
+      "loss": 1.0524,
+      "step": 3624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7116777131908216e-05,
+      "loss": 1.0345,
+      "step": 3625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7115205920340364e-05,
+      "loss": 1.0125,
+      "step": 3626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.711363435292968e-05,
+      "loss": 0.8992,
+      "step": 3627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7112062429754764e-05,
+      "loss": 1.0695,
+      "step": 3628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7110490150894228e-05,
+      "loss": 0.9931,
+      "step": 3629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7108917516426706e-05,
+      "loss": 1.0157,
+      "step": 3630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7107344526430846e-05,
+      "loss": 0.9627,
+      "step": 3631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.710577118098532e-05,
+      "loss": 1.031,
+      "step": 3632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.71041974801688e-05,
+      "loss": 1.0147,
+      "step": 3633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.710262342406001e-05,
+      "loss": 1.0319,
+      "step": 3634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7101049012737653e-05,
+      "loss": 1.0058,
+      "step": 3635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.709947424628048e-05,
+      "loss": 0.9629,
+      "step": 3636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7097899124767243e-05,
+      "loss": 0.9141,
+      "step": 3637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7096323648276714e-05,
+      "loss": 1.0276,
+      "step": 3638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7094747816887688e-05,
+      "loss": 1.0597,
+      "step": 3639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7093171630678976e-05,
+      "loss": 1.0895,
+      "step": 3640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.70915950897294e-05,
+      "loss": 1.1427,
+      "step": 3641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.709001819411781e-05,
+      "loss": 0.9344,
+      "step": 3642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.708844094392307e-05,
+      "loss": 1.1518,
+      "step": 3643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7086863339224057e-05,
+      "loss": 0.9262,
+      "step": 3644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.708528538009967e-05,
+      "loss": 0.9879,
+      "step": 3645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.708370706662883e-05,
+      "loss": 0.6936,
+      "step": 3646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7082128398890468e-05,
+      "loss": 0.9374,
+      "step": 3647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7080549376963537e-05,
+      "loss": 1.0943,
+      "step": 3648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7078970000927e-05,
+      "loss": 0.6667,
+      "step": 3649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7077390270859852e-05,
+      "loss": 1.0322,
+      "step": 3650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7075810186841093e-05,
+      "loss": 0.8764,
+      "step": 3651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.707422974894975e-05,
+      "loss": 1.0618,
+      "step": 3652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7072648957264857e-05,
+      "loss": 1.0652,
+      "step": 3653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.9764,
+      "step": 3654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7069486312830682e-05,
+      "loss": 1.1373,
+      "step": 3655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7067904460239574e-05,
+      "loss": 0.9737,
+      "step": 3656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.706632225417125e-05,
+      "loss": 0.8575,
+      "step": 3657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.706473969470485e-05,
+      "loss": 1.0112,
+      "step": 3658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7063156781919518e-05,
+      "loss": 1.0545,
+      "step": 3659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.706157351589441e-05,
+      "loss": 0.9867,
+      "step": 3660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7059989896708718e-05,
+      "loss": 0.909,
+      "step": 3661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7058405924441638e-05,
+      "loss": 0.9733,
+      "step": 3662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7056821599172384e-05,
+      "loss": 1.0406,
+      "step": 3663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7055236920980195e-05,
+      "loss": 0.9547,
+      "step": 3664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.705365188994432e-05,
+      "loss": 0.9935,
+      "step": 3665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.705206650614403e-05,
+      "loss": 1.0268,
+      "step": 3666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7050480769658616e-05,
+      "loss": 0.9572,
+      "step": 3667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7048894680567377e-05,
+      "loss": 1.138,
+      "step": 3668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7047308238949635e-05,
+      "loss": 0.9581,
+      "step": 3669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.704572144488474e-05,
+      "loss": 1.0061,
+      "step": 3670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7044134298452043e-05,
+      "loss": 1.1021,
+      "step": 3671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.704254679973092e-05,
+      "loss": 0.9301,
+      "step": 3672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.704095894880077e-05,
+      "loss": 0.9573,
+      "step": 3673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7039370745740996e-05,
+      "loss": 1.0296,
+      "step": 3674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7037782190631028e-05,
+      "loss": 1.0971,
+      "step": 3675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.703619328355032e-05,
+      "loss": 0.9883,
+      "step": 3676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7034604024578323e-05,
+      "loss": 0.9561,
+      "step": 3677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7033014413794527e-05,
+      "loss": 1.0206,
+      "step": 3678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7031424451278432e-05,
+      "loss": 0.9587,
+      "step": 3679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7029834137109552e-05,
+      "loss": 0.9672,
+      "step": 3680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.702824347136742e-05,
+      "loss": 1.0681,
+      "step": 3681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7026652454131588e-05,
+      "loss": 0.9901,
+      "step": 3682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.702506108548163e-05,
+      "loss": 1.0558,
+      "step": 3683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7023469365497125e-05,
+      "loss": 0.9399,
+      "step": 3684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.702187729425768e-05,
+      "loss": 0.9757,
+      "step": 3685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7020284871842925e-05,
+      "loss": 0.8676,
+      "step": 3686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.701869209833249e-05,
+      "loss": 0.9542,
+      "step": 3687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7017098973806035e-05,
+      "loss": 0.9699,
+      "step": 3688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7015505498343232e-05,
+      "loss": 0.9715,
+      "step": 3689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.701391167202378e-05,
+      "loss": 1.0596,
+      "step": 3690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7012317494927385e-05,
+      "loss": 1.0419,
+      "step": 3691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7010722967133772e-05,
+      "loss": 0.9424,
+      "step": 3692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.700912808872269e-05,
+      "loss": 0.9351,
+      "step": 3693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7007532859773902e-05,
+      "loss": 0.9041,
+      "step": 3694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.700593728036718e-05,
+      "loss": 1.0622,
+      "step": 3695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.700434135058233e-05,
+      "loss": 0.9568,
+      "step": 3696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7002745070499164e-05,
+      "loss": 0.9226,
+      "step": 3697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7001148440197517e-05,
+      "loss": 0.6826,
+      "step": 3698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6999551459757234e-05,
+      "loss": 0.961,
+      "step": 3699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6997954129258186e-05,
+      "loss": 0.9547,
+      "step": 3700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6996356448780256e-05,
+      "loss": 1.0395,
+      "step": 3701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.699475841840335e-05,
+      "loss": 0.8604,
+      "step": 3702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6993160038207385e-05,
+      "loss": 0.8758,
+      "step": 3703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.69915613082723e-05,
+      "loss": 0.6549,
+      "step": 3704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.698996222867805e-05,
+      "loss": 0.6823,
+      "step": 3705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6988362799504608e-05,
+      "loss": 0.9975,
+      "step": 3706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.698676302083196e-05,
+      "loss": 0.9462,
+      "step": 3707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6985162892740116e-05,
+      "loss": 0.9543,
+      "step": 3708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6983562415309104e-05,
+      "loss": 0.9742,
+      "step": 3709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.698196158861896e-05,
+      "loss": 0.9996,
+      "step": 3710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.698036041274975e-05,
+      "loss": 1.0523,
+      "step": 3711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6978758887781554e-05,
+      "loss": 0.9604,
+      "step": 3712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6977157013794456e-05,
+      "loss": 0.9747,
+      "step": 3713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.697555479086858e-05,
+      "loss": 1.1362,
+      "step": 3714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6973952219084044e-05,
+      "loss": 1.007,
+      "step": 3715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6972349298521e-05,
+      "loss": 0.8236,
+      "step": 3716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.697074602925962e-05,
+      "loss": 0.9948,
+      "step": 3717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.696914241138008e-05,
+      "loss": 1.0725,
+      "step": 3718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6967538444962576e-05,
+      "loss": 0.9832,
+      "step": 3719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.696593413008733e-05,
+      "loss": 0.9154,
+      "step": 3720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6964329466834574e-05,
+      "loss": 1.021,
+      "step": 3721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.696272445528456e-05,
+      "loss": 0.9716,
+      "step": 3722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.696111909551756e-05,
+      "loss": 1.0268,
+      "step": 3723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6959513387613852e-05,
+      "loss": 0.6974,
+      "step": 3724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695790733165375e-05,
+      "loss": 0.984,
+      "step": 3725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6956300927717574e-05,
+      "loss": 1.0044,
+      "step": 3726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695469417588566e-05,
+      "loss": 1.0571,
+      "step": 3727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6953087076238362e-05,
+      "loss": 1.008,
+      "step": 3728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695147962885606e-05,
+      "loss": 1.0555,
+      "step": 3729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6949871833819138e-05,
+      "loss": 1.0488,
+      "step": 3730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.694826369120801e-05,
+      "loss": 0.9702,
+      "step": 3731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6946655201103096e-05,
+      "loss": 0.9332,
+      "step": 3732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6945046363584844e-05,
+      "loss": 1.0084,
+      "step": 3733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6943437178733715e-05,
+      "loss": 1.0073,
+      "step": 3734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6941827646630188e-05,
+      "loss": 0.9784,
+      "step": 3735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.694021776735475e-05,
+      "loss": 0.9482,
+      "step": 3736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6938607540987922e-05,
+      "loss": 0.9918,
+      "step": 3737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693699696761023e-05,
+      "loss": 1.0081,
+      "step": 3738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693538604730223e-05,
+      "loss": 0.9546,
+      "step": 3739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693377478014447e-05,
+      "loss": 0.8975,
+      "step": 3740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693216316621754e-05,
+      "loss": 1.0145,
+      "step": 3741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6930551205602045e-05,
+      "loss": 0.8991,
+      "step": 3742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6928938898378598e-05,
+      "loss": 1.0583,
+      "step": 3743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692732624462783e-05,
+      "loss": 1.0719,
+      "step": 3744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6925713244430393e-05,
+      "loss": 0.9668,
+      "step": 3745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6924099897866962e-05,
+      "loss": 0.9711,
+      "step": 3746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6922486205018217e-05,
+      "loss": 1.026,
+      "step": 3747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692087216596486e-05,
+      "loss": 0.9999,
+      "step": 3748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.691925778078762e-05,
+      "loss": 1.07,
+      "step": 3749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6917643049567226e-05,
+      "loss": 0.6856,
+      "step": 3750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6916027972384437e-05,
+      "loss": 1.0639,
+      "step": 3751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6914412549320028e-05,
+      "loss": 1.0126,
+      "step": 3752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6912796780454783e-05,
+      "loss": 0.9655,
+      "step": 3753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6911180665869517e-05,
+      "loss": 1.1034,
+      "step": 3754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6909564205645047e-05,
+      "loss": 1.0727,
+      "step": 3755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6907947399862214e-05,
+      "loss": 0.9746,
+      "step": 3756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6906330248601886e-05,
+      "loss": 0.9382,
+      "step": 3757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690471275194493e-05,
+      "loss": 1.0094,
+      "step": 3758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690309490997225e-05,
+      "loss": 0.9735,
+      "step": 3759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6901476722764745e-05,
+      "loss": 1.0234,
+      "step": 3760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6899858190403353e-05,
+      "loss": 1.0498,
+      "step": 3761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.689823931296901e-05,
+      "loss": 0.9973,
+      "step": 3762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6896620090542687e-05,
+      "loss": 1.0064,
+      "step": 3763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6895000523205363e-05,
+      "loss": 0.9671,
+      "step": 3764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6893380611038027e-05,
+      "loss": 0.9984,
+      "step": 3765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6891760354121704e-05,
+      "loss": 1.0212,
+      "step": 3766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.689013975253742e-05,
+      "loss": 1.0436,
+      "step": 3767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6888518806366227e-05,
+      "loss": 1.018,
+      "step": 3768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6886897515689184e-05,
+      "loss": 0.9474,
+      "step": 3769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.688527588058738e-05,
+      "loss": 1.0215,
+      "step": 3770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6883653901141917e-05,
+      "loss": 0.9983,
+      "step": 3771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6882031577433907e-05,
+      "loss": 0.9565,
+      "step": 3772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.688040890954449e-05,
+      "loss": 1.0083,
+      "step": 3773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6878785897554818e-05,
+      "loss": 1.0231,
+      "step": 3774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.687716254154606e-05,
+      "loss": 1.0012,
+      "step": 3775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6875538841599402e-05,
+      "loss": 0.9,
+      "step": 3776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6873914797796046e-05,
+      "loss": 1.0455,
+      "step": 3777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6872290410217216e-05,
+      "loss": 1.0606,
+      "step": 3778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6870665678944146e-05,
+      "loss": 1.1032,
+      "step": 3779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6869040604058097e-05,
+      "loss": 0.941,
+      "step": 3780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.686741518564034e-05,
+      "loss": 0.9991,
+      "step": 3781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.686578942377216e-05,
+      "loss": 1.1069,
+      "step": 3782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.686416331853487e-05,
+      "loss": 0.9315,
+      "step": 3783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6862536870009792e-05,
+      "loss": 0.9385,
+      "step": 3784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.686091007827827e-05,
+      "loss": 1.0431,
+      "step": 3785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6859282943421655e-05,
+      "loss": 0.9972,
+      "step": 3786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6857655465521325e-05,
+      "loss": 0.9358,
+      "step": 3787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.685602764465868e-05,
+      "loss": 0.9135,
+      "step": 3788
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6854399480915122e-05,
+      "loss": 0.6766,
+      "step": 3789
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.685277097437208e-05,
+      "loss": 0.9797,
+      "step": 3790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6851142125111e-05,
+      "loss": 1.0986,
+      "step": 3791
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6849512933213343e-05,
+      "loss": 1.0253,
+      "step": 3792
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.684788339876059e-05,
+      "loss": 1.0837,
+      "step": 3793
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.684625352183423e-05,
+      "loss": 0.8989,
+      "step": 3794
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6844623302515782e-05,
+      "loss": 0.9383,
+      "step": 3795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.684299274088677e-05,
+      "loss": 0.9728,
+      "step": 3796
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6841361837028744e-05,
+      "loss": 1.0384,
+      "step": 3797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6839730591023273e-05,
+      "loss": 0.6985,
+      "step": 3798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6838099002951928e-05,
+      "loss": 0.6744,
+      "step": 3799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.683646707289632e-05,
+      "loss": 0.9766,
+      "step": 3800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.683483480093805e-05,
+      "loss": 1.0351,
+      "step": 3801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6833202187158762e-05,
+      "loss": 1.0351,
+      "step": 3802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6831569231640098e-05,
+      "loss": 0.9952,
+      "step": 3803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.682993593446373e-05,
+      "loss": 1.0208,
+      "step": 3804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.682830229571134e-05,
+      "loss": 0.9045,
+      "step": 3805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6826668315464633e-05,
+      "loss": 1.0214,
+      "step": 3806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6825033993805315e-05,
+      "loss": 0.9388,
+      "step": 3807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6823399330815134e-05,
+      "loss": 0.8522,
+      "step": 3808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6821764326575837e-05,
+      "loss": 0.9434,
+      "step": 3809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.682012898116919e-05,
+      "loss": 0.9193,
+      "step": 3810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6818493294676988e-05,
+      "loss": 1.0185,
+      "step": 3811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6816857267181023e-05,
+      "loss": 0.9216,
+      "step": 3812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6815220898763126e-05,
+      "loss": 1.0362,
+      "step": 3813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6813584189505127e-05,
+      "loss": 0.9659,
+      "step": 3814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6811947139488884e-05,
+      "loss": 1.0568,
+      "step": 3815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6810309748796263e-05,
+      "loss": 0.9886,
+      "step": 3816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.680867201750916e-05,
+      "loss": 1.0569,
+      "step": 3817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6807033945709477e-05,
+      "loss": 0.7081,
+      "step": 3818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6805395533479138e-05,
+      "loss": 1.0667,
+      "step": 3819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.680375678090008e-05,
+      "loss": 1.0402,
+      "step": 3820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6802117688054265e-05,
+      "loss": 0.9916,
+      "step": 3821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.680047825502366e-05,
+      "loss": 0.9676,
+      "step": 3822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6798838481890257e-05,
+      "loss": 1.1393,
+      "step": 3823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6797198368736068e-05,
+      "loss": 0.8777,
+      "step": 3824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6795557915643115e-05,
+      "loss": 0.9252,
+      "step": 3825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6793917122693442e-05,
+      "loss": 0.9846,
+      "step": 3826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6792275989969103e-05,
+      "loss": 0.9305,
+      "step": 3827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6790634517552175e-05,
+      "loss": 1.0278,
+      "step": 3828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6788992705524753e-05,
+      "loss": 0.9912,
+      "step": 3829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6787350553968947e-05,
+      "loss": 1.0058,
+      "step": 3830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.678570806296688e-05,
+      "loss": 1.0068,
+      "step": 3831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6784065232600697e-05,
+      "loss": 1.0823,
+      "step": 3832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.678242206295256e-05,
+      "loss": 0.9972,
+      "step": 3833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6780778554104644e-05,
+      "loss": 0.9189,
+      "step": 3834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.677913470613915e-05,
+      "loss": 1.0434,
+      "step": 3835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.677749051913828e-05,
+      "loss": 0.9653,
+      "step": 3836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6775845993184267e-05,
+      "loss": 1.0001,
+      "step": 3837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6774201128359358e-05,
+      "loss": 0.9103,
+      "step": 3838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6772555924745814e-05,
+      "loss": 0.7043,
+      "step": 3839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6770910382425912e-05,
+      "loss": 0.9355,
+      "step": 3840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6769264501481957e-05,
+      "loss": 1.1247,
+      "step": 3841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.676761828199625e-05,
+      "loss": 1.0534,
+      "step": 3842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6765971724051128e-05,
+      "loss": 0.9001,
+      "step": 3843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6764324827728935e-05,
+      "loss": 0.9756,
+      "step": 3844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6762677593112034e-05,
+      "loss": 1.0369,
+      "step": 3845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6761030020282813e-05,
+      "loss": 1.0167,
+      "step": 3846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6759382109323664e-05,
+      "loss": 1.0468,
+      "step": 3847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6757733860317e-05,
+      "loss": 1.0625,
+      "step": 3848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.675608527334526e-05,
+      "loss": 1.0636,
+      "step": 3849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6754436348490885e-05,
+      "loss": 1.0133,
+      "step": 3850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.675278708583634e-05,
+      "loss": 0.9694,
+      "step": 3851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6751137485464114e-05,
+      "loss": 1.0139,
+      "step": 3852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6749487547456703e-05,
+      "loss": 1.0125,
+      "step": 3853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.674783727189662e-05,
+      "loss": 1.0544,
+      "step": 3854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6746186658866405e-05,
+      "loss": 1.0492,
+      "step": 3855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6744535708448597e-05,
+      "loss": 0.9968,
+      "step": 3856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6742884420725773e-05,
+      "loss": 1.0705,
+      "step": 3857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.674123279578051e-05,
+      "loss": 0.9764,
+      "step": 3858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6739580833695414e-05,
+      "loss": 0.9687,
+      "step": 3859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6737928534553095e-05,
+      "loss": 0.9542,
+      "step": 3860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6736275898436195e-05,
+      "loss": 0.6901,
+      "step": 3861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6734622925427362e-05,
+      "loss": 0.9455,
+      "step": 3862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6732969615609257e-05,
+      "loss": 1.0078,
+      "step": 3863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.673131596906458e-05,
+      "loss": 1.0437,
+      "step": 3864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6729661985876016e-05,
+      "loss": 1.0813,
+      "step": 3865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6728007666126292e-05,
+      "loss": 1.0368,
+      "step": 3866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6726353009898144e-05,
+      "loss": 0.9516,
+      "step": 3867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6724698017274316e-05,
+      "loss": 1.088,
+      "step": 3868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6723042688337588e-05,
+      "loss": 0.7964,
+      "step": 3869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.672138702317074e-05,
+      "loss": 0.9644,
+      "step": 3870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6719731021856574e-05,
+      "loss": 1.0208,
+      "step": 3871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6718074684477906e-05,
+      "loss": 0.9825,
+      "step": 3872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6716418011117583e-05,
+      "loss": 0.9638,
+      "step": 3873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6714761001858446e-05,
+      "loss": 1.0628,
+      "step": 3874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6713103656783373e-05,
+      "loss": 0.9795,
+      "step": 3875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6711445975975242e-05,
+      "loss": 0.9353,
+      "step": 3876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6709787959516966e-05,
+      "loss": 1.0164,
+      "step": 3877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670812960749146e-05,
+      "loss": 1.0767,
+      "step": 3878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670647091998166e-05,
+      "loss": 1.0696,
+      "step": 3879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670481189707052e-05,
+      "loss": 1.0708,
+      "step": 3880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6703152538841014e-05,
+      "loss": 0.9561,
+      "step": 3881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670149284537612e-05,
+      "loss": 1.0645,
+      "step": 3882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6699832816758858e-05,
+      "loss": 0.8967,
+      "step": 3883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6698172453072234e-05,
+      "loss": 1.0337,
+      "step": 3884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6696511754399288e-05,
+      "loss": 1.0587,
+      "step": 3885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669485072082308e-05,
+      "loss": 1.0609,
+      "step": 3886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6693189352426677e-05,
+      "loss": 0.9115,
+      "step": 3887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6691527649293168e-05,
+      "loss": 0.8636,
+      "step": 3888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6689865611505657e-05,
+      "loss": 1.0792,
+      "step": 3889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6688203239147266e-05,
+      "loss": 0.9611,
+      "step": 3890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6686540532301128e-05,
+      "loss": 0.9934,
+      "step": 3891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66848774910504e-05,
+      "loss": 0.9648,
+      "step": 3892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668321411547826e-05,
+      "loss": 0.9672,
+      "step": 3893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668155040566789e-05,
+      "loss": 1.0222,
+      "step": 3894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6679886361702494e-05,
+      "loss": 0.9955,
+      "step": 3895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6678221983665297e-05,
+      "loss": 0.9895,
+      "step": 3896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6676557271639534e-05,
+      "loss": 0.6651,
+      "step": 3897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667489222570846e-05,
+      "loss": 1.0397,
+      "step": 3898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667322684595535e-05,
+      "loss": 1.0656,
+      "step": 3899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667156113246349e-05,
+      "loss": 1.0387,
+      "step": 3900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6669895085316185e-05,
+      "loss": 0.922,
+      "step": 3901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6668228704596755e-05,
+      "loss": 0.9374,
+      "step": 3902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6666561990388544e-05,
+      "loss": 0.9751,
+      "step": 3903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.66648949427749e-05,
+      "loss": 1.0218,
+      "step": 3904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.66632275618392e-05,
+      "loss": 0.9817,
+      "step": 3905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.666155984766483e-05,
+      "loss": 0.9484,
+      "step": 3906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6659891800335196e-05,
+      "loss": 0.9371,
+      "step": 3907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.665822341993372e-05,
+      "loss": 0.9952,
+      "step": 3908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6656554706543836e-05,
+      "loss": 0.6902,
+      "step": 3909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6654885660249004e-05,
+      "loss": 0.9758,
+      "step": 3910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6653216281132694e-05,
+      "loss": 0.9454,
+      "step": 3911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6651546569278396e-05,
+      "loss": 1.0121,
+      "step": 3912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6649876524769612e-05,
+      "loss": 0.8916,
+      "step": 3913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6648206147689865e-05,
+      "loss": 1.0719,
+      "step": 3914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6646535438122696e-05,
+      "loss": 0.9595,
+      "step": 3915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6644864396151653e-05,
+      "loss": 1.0311,
+      "step": 3916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6643193021860312e-05,
+      "loss": 0.8761,
+      "step": 3917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6641521315332264e-05,
+      "loss": 0.9668,
+      "step": 3918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.663984927665111e-05,
+      "loss": 0.911,
+      "step": 3919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.663817690590047e-05,
+      "loss": 1.0233,
+      "step": 3920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6636504203163986e-05,
+      "loss": 0.657,
+      "step": 3921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6634831168525313e-05,
+      "loss": 1.0367,
+      "step": 3922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6633157802068113e-05,
+      "loss": 1.0319,
+      "step": 3923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6631484103876083e-05,
+      "loss": 1.0538,
+      "step": 3924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6629810074032925e-05,
+      "loss": 0.6427,
+      "step": 3925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6628135712622357e-05,
+      "loss": 1.1008,
+      "step": 3926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6626461019728123e-05,
+      "loss": 1.0009,
+      "step": 3927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.662478599543397e-05,
+      "loss": 0.9241,
+      "step": 3928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.662311063982367e-05,
+      "loss": 1.1021,
+      "step": 3929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6621434952981017e-05,
+      "loss": 0.9417,
+      "step": 3930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6619758934989805e-05,
+      "loss": 0.9027,
+      "step": 3931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6618082585933862e-05,
+      "loss": 1.0107,
+      "step": 3932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.661640590589702e-05,
+      "loss": 1.0468,
+      "step": 3933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6614728894963135e-05,
+      "loss": 0.9994,
+      "step": 3934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6613051553216077e-05,
+      "loss": 0.985,
+      "step": 3935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6611373880739733e-05,
+      "loss": 0.9706,
+      "step": 3936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6609695877618e-05,
+      "loss": 0.8803,
+      "step": 3937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6608017543934808e-05,
+      "loss": 0.9212,
+      "step": 3938
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6606338879774085e-05,
+      "loss": 1.0566,
+      "step": 3939
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6604659885219786e-05,
+      "loss": 0.9934,
+      "step": 3940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6602980560355885e-05,
+      "loss": 0.6585,
+      "step": 3941
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6601300905266357e-05,
+      "loss": 1.0127,
+      "step": 3942
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6599620920035214e-05,
+      "loss": 1.0044,
+      "step": 3943
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.659794060474647e-05,
+      "loss": 0.9897,
+      "step": 3944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.659625995948416e-05,
+      "loss": 1.0408,
+      "step": 3945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6594578984332338e-05,
+      "loss": 1.0811,
+      "step": 3946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6592897679375074e-05,
+      "loss": 0.9097,
+      "step": 3947
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.659121604469645e-05,
+      "loss": 0.9923,
+      "step": 3948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6589534080380568e-05,
+      "loss": 1.041,
+      "step": 3949
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6587851786511545e-05,
+      "loss": 1.0605,
+      "step": 3950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6586169163173514e-05,
+      "loss": 0.9622,
+      "step": 3951
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6584486210450627e-05,
+      "loss": 0.6967,
+      "step": 3952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6582802928427052e-05,
+      "loss": 0.8945,
+      "step": 3953
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6581119317186972e-05,
+      "loss": 1.0257,
+      "step": 3954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.657943537681459e-05,
+      "loss": 0.868,
+      "step": 3955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6577751107394116e-05,
+      "loss": 1.0691,
+      "step": 3956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6576066509009788e-05,
+      "loss": 1.0351,
+      "step": 3957
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6574381581745853e-05,
+      "loss": 1.0316,
+      "step": 3958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6572696325686578e-05,
+      "loss": 0.9536,
+      "step": 3959
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6571010740916245e-05,
+      "loss": 0.9983,
+      "step": 3960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6569324827519154e-05,
+      "loss": 0.9057,
+      "step": 3961
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.656763858557962e-05,
+      "loss": 1.015,
+      "step": 3962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6565952015181974e-05,
+      "loss": 0.9341,
+      "step": 3963
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6564265116410565e-05,
+      "loss": 0.9435,
+      "step": 3964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6562577889349754e-05,
+      "loss": 0.9092,
+      "step": 3965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6560890334083925e-05,
+      "loss": 1.0594,
+      "step": 3966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6559202450697477e-05,
+      "loss": 0.9977,
+      "step": 3967
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.655751423927482e-05,
+      "loss": 1.0208,
+      "step": 3968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6555825699900387e-05,
+      "loss": 0.9979,
+      "step": 3969
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6554136832658622e-05,
+      "loss": 0.6827,
+      "step": 3970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.655244763763399e-05,
+      "loss": 1.0037,
+      "step": 3971
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.655075811491097e-05,
+      "loss": 0.9435,
+      "step": 3972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6549068264574054e-05,
+      "loss": 1.04,
+      "step": 3973
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.654737808670776e-05,
+      "loss": 1.0047,
+      "step": 3974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6545687581396616e-05,
+      "loss": 0.9552,
+      "step": 3975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.654399674872516e-05,
+      "loss": 0.9859,
+      "step": 3976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.654230558877796e-05,
+      "loss": 1.0044,
+      "step": 3977
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6540614101639594e-05,
+      "loss": 0.8955,
+      "step": 3978
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6538922287394652e-05,
+      "loss": 1.1014,
+      "step": 3979
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6537230146127746e-05,
+      "loss": 0.9359,
+      "step": 3980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6535537677923503e-05,
+      "loss": 0.9577,
+      "step": 3981
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.653384488286657e-05,
+      "loss": 0.9567,
+      "step": 3982
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6532151761041593e-05,
+      "loss": 1.0337,
+      "step": 3983
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6530458312533266e-05,
+      "loss": 1.0564,
+      "step": 3984
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.652876453742627e-05,
+      "loss": 1.0086,
+      "step": 3985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6527070435805307e-05,
+      "loss": 0.9995,
+      "step": 3986
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.652537600775512e-05,
+      "loss": 0.6589,
+      "step": 3987
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6523681253360437e-05,
+      "loss": 0.9676,
+      "step": 3988
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6521986172706017e-05,
+      "loss": 0.7067,
+      "step": 3989
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6520290765876638e-05,
+      "loss": 0.9555,
+      "step": 3990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6518595032957084e-05,
+      "loss": 0.9951,
+      "step": 3991
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6516898974032164e-05,
+      "loss": 0.7351,
+      "step": 3992
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6515202589186702e-05,
+      "loss": 0.9141,
+      "step": 3993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6513505878505538e-05,
+      "loss": 0.9304,
+      "step": 3994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6511808842073524e-05,
+      "loss": 0.9748,
+      "step": 3995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6510111479975536e-05,
+      "loss": 0.974,
+      "step": 3996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6508413792296453e-05,
+      "loss": 0.9415,
+      "step": 3997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6506715779121186e-05,
+      "loss": 0.7188,
+      "step": 3998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.650501744053466e-05,
+      "loss": 1.1053,
+      "step": 3999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.65033187766218e-05,
+      "loss": 1.0936,
+      "step": 4000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6501619787467564e-05,
+      "loss": 0.9621,
+      "step": 4001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6499920473156927e-05,
+      "loss": 0.9205,
+      "step": 4002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6498220833774864e-05,
+      "loss": 1.0665,
+      "step": 4003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6496520869406382e-05,
+      "loss": 0.9056,
+      "step": 4004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.64948205801365e-05,
+      "loss": 1.0134,
+      "step": 4005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6493119966050253e-05,
+      "loss": 0.9256,
+      "step": 4006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6491419027232683e-05,
+      "loss": 0.965,
+      "step": 4007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648971776376887e-05,
+      "loss": 0.6614,
+      "step": 4008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648801617574389e-05,
+      "loss": 0.972,
+      "step": 4009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6486314263242834e-05,
+      "loss": 0.9477,
+      "step": 4010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6484612026350833e-05,
+      "loss": 0.9043,
+      "step": 4011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648290946515301e-05,
+      "loss": 1.0522,
+      "step": 4012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6481206579734513e-05,
+      "loss": 1.0291,
+      "step": 4013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647950337018051e-05,
+      "loss": 0.6797,
+      "step": 4014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647779983657617e-05,
+      "loss": 0.8114,
+      "step": 4015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6476095979006704e-05,
+      "loss": 0.936,
+      "step": 4016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6474391797557317e-05,
+      "loss": 1.006,
+      "step": 4017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6472687292313242e-05,
+      "loss": 0.8827,
+      "step": 4018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6470982463359717e-05,
+      "loss": 0.9948,
+      "step": 4019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6469277310782008e-05,
+      "loss": 0.9122,
+      "step": 4020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646757183466539e-05,
+      "loss": 0.9604,
+      "step": 4021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6465866035095165e-05,
+      "loss": 0.9698,
+      "step": 4022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6464159912156627e-05,
+      "loss": 0.9481,
+      "step": 4023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6462453465935116e-05,
+      "loss": 1.0098,
+      "step": 4024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646074669651597e-05,
+      "loss": 1.0273,
+      "step": 4025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6459039603984544e-05,
+      "loss": 0.9735,
+      "step": 4026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6457332188426213e-05,
+      "loss": 0.9094,
+      "step": 4027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6455624449926376e-05,
+      "loss": 0.9773,
+      "step": 4028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645391638857043e-05,
+      "loss": 0.9999,
+      "step": 4029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.64522080044438e-05,
+      "loss": 0.929,
+      "step": 4030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6450499297631926e-05,
+      "loss": 1.0504,
+      "step": 4031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6448790268220264e-05,
+      "loss": 1.0136,
+      "step": 4032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644708091629428e-05,
+      "loss": 0.8949,
+      "step": 4033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6445371241939475e-05,
+      "loss": 1.0608,
+      "step": 4034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6443661245241337e-05,
+      "loss": 1.0204,
+      "step": 4035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644195092628539e-05,
+      "loss": 0.8444,
+      "step": 4036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6440240285157177e-05,
+      "loss": 1.0278,
+      "step": 4037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6438529321942246e-05,
+      "loss": 0.9468,
+      "step": 4038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643681803672616e-05,
+      "loss": 1.0495,
+      "step": 4039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643510642959451e-05,
+      "loss": 1.1031,
+      "step": 4040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6433394500632885e-05,
+      "loss": 0.9772,
+      "step": 4041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643168224992692e-05,
+      "loss": 0.674,
+      "step": 4042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6429969677562233e-05,
+      "loss": 0.9928,
+      "step": 4043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6428256783624474e-05,
+      "loss": 0.9455,
+      "step": 4044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.642654356819931e-05,
+      "loss": 0.9926,
+      "step": 4045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6424830031372424e-05,
+      "loss": 0.9945,
+      "step": 4046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.642311617322951e-05,
+      "loss": 0.9492,
+      "step": 4047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.642140199385628e-05,
+      "loss": 0.6505,
+      "step": 4048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6419687493338466e-05,
+      "loss": 0.9118,
+      "step": 4049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6417972671761808e-05,
+      "loss": 0.9478,
+      "step": 4050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.641625752921207e-05,
+      "loss": 0.6712,
+      "step": 4051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.641454206577503e-05,
+      "loss": 0.972,
+      "step": 4052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6412826281536483e-05,
+      "loss": 0.9053,
+      "step": 4053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.641111017658223e-05,
+      "loss": 1.0641,
+      "step": 4054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6409393750998103e-05,
+      "loss": 0.9734,
+      "step": 4055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.640767700486994e-05,
+      "loss": 1.0214,
+      "step": 4056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6405959938283603e-05,
+      "loss": 1.0414,
+      "step": 4057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.640424255132496e-05,
+      "loss": 0.9586,
+      "step": 4058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6402524844079902e-05,
+      "loss": 1.0126,
+      "step": 4059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.640080681663433e-05,
+      "loss": 0.6512,
+      "step": 4060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.639908846907418e-05,
+      "loss": 0.8922,
+      "step": 4061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.639736980148537e-05,
+      "loss": 0.97,
+      "step": 4062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6395650813953864e-05,
+      "loss": 0.9422,
+      "step": 4063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.639393150656563e-05,
+      "loss": 0.9395,
+      "step": 4064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6392211879406654e-05,
+      "loss": 0.9983,
+      "step": 4065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6390491932562936e-05,
+      "loss": 0.9415,
+      "step": 4066
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6388771666120495e-05,
+      "loss": 0.9244,
+      "step": 4067
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.638705108016536e-05,
+      "loss": 0.6383,
+      "step": 4068
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.638533017478358e-05,
+      "loss": 1.0222,
+      "step": 4069
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6383608950061233e-05,
+      "loss": 0.997,
+      "step": 4070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6381887406084384e-05,
+      "loss": 0.9781,
+      "step": 4071
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6380165542939134e-05,
+      "loss": 1.0303,
+      "step": 4072
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6378443360711603e-05,
+      "loss": 1.0262,
+      "step": 4073
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6376720859487916e-05,
+      "loss": 1.018,
+      "step": 4074
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6374998039354217e-05,
+      "loss": 0.9299,
+      "step": 4075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6373274900396667e-05,
+      "loss": 0.9763,
+      "step": 4076
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6371551442701448e-05,
+      "loss": 0.9976,
+      "step": 4077
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6369827666354746e-05,
+      "loss": 0.9657,
+      "step": 4078
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.636810357144277e-05,
+      "loss": 1.0515,
+      "step": 4079
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.636637915805175e-05,
+      "loss": 1.0233,
+      "step": 4080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.636465442626792e-05,
+      "loss": 0.94,
+      "step": 4081
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.636292937617755e-05,
+      "loss": 0.9733,
+      "step": 4082
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6361204007866892e-05,
+      "loss": 0.9307,
+      "step": 4083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6359478321422252e-05,
+      "loss": 0.9984,
+      "step": 4084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6357752316929927e-05,
+      "loss": 0.9754,
+      "step": 4085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6356025994476236e-05,
+      "loss": 1.0226,
+      "step": 4086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.635429935414752e-05,
+      "loss": 0.864,
+      "step": 4087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6352572396030126e-05,
+      "loss": 0.6646,
+      "step": 4088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6350845120210424e-05,
+      "loss": 1.078,
+      "step": 4089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.63491175267748e-05,
+      "loss": 1.0213,
+      "step": 4090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.634738961580965e-05,
+      "loss": 0.9816,
+      "step": 4091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.634566138740139e-05,
+      "loss": 1.0423,
+      "step": 4092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6343932841636455e-05,
+      "loss": 0.918,
+      "step": 4093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.634220397860129e-05,
+      "loss": 0.944,
+      "step": 4094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.634047479838236e-05,
+      "loss": 1.0155,
+      "step": 4095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6338745301066138e-05,
+      "loss": 0.9135,
+      "step": 4096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.633701548673913e-05,
+      "loss": 0.8828,
+      "step": 4097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6335285355487834e-05,
+      "loss": 1.0213,
+      "step": 4098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6333554907398786e-05,
+      "loss": 0.8643,
+      "step": 4099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6331824142558526e-05,
+      "loss": 0.9204,
+      "step": 4100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6330093061053605e-05,
+      "loss": 1.0365,
+      "step": 4101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.632836166297061e-05,
+      "loss": 0.9689,
+      "step": 4102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6326629948396123e-05,
+      "loss": 0.8687,
+      "step": 4103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6324897917416752e-05,
+      "loss": 1.0097,
+      "step": 4104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6323165570119117e-05,
+      "loss": 1.0808,
+      "step": 4105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6321432906589857e-05,
+      "loss": 0.9077,
+      "step": 4106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6319699926915628e-05,
+      "loss": 1.033,
+      "step": 4107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6317966631183093e-05,
+      "loss": 0.9315,
+      "step": 4108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6316233019478937e-05,
+      "loss": 0.9999,
+      "step": 4109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.631449909188987e-05,
+      "loss": 1.0226,
+      "step": 4110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.63127648485026e-05,
+      "loss": 0.8911,
+      "step": 4111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.631103028940386e-05,
+      "loss": 1.0063,
+      "step": 4112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6309295414680398e-05,
+      "loss": 0.9039,
+      "step": 4113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.630756022441898e-05,
+      "loss": 1.0172,
+      "step": 4114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6305824718706383e-05,
+      "loss": 0.9923,
+      "step": 4115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.630408889762941e-05,
+      "loss": 0.865,
+      "step": 4116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6302352761274863e-05,
+      "loss": 0.9519,
+      "step": 4117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6300616309729572e-05,
+      "loss": 0.9442,
+      "step": 4118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.629887954308038e-05,
+      "loss": 0.9776,
+      "step": 4119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6297142461414146e-05,
+      "loss": 0.9741,
+      "step": 4120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6295405064817743e-05,
+      "loss": 0.9391,
+      "step": 4121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6293667353378066e-05,
+      "loss": 0.9612,
+      "step": 4122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6291929327182016e-05,
+      "loss": 0.9677,
+      "step": 4123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6290190986316512e-05,
+      "loss": 0.9292,
+      "step": 4124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6288452330868495e-05,
+      "loss": 0.93,
+      "step": 4125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6286713360924916e-05,
+      "loss": 0.9907,
+      "step": 4126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6284974076572747e-05,
+      "loss": 0.958,
+      "step": 4127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6283234477898974e-05,
+      "loss": 0.6689,
+      "step": 4128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6281494564990586e-05,
+      "loss": 0.9589,
+      "step": 4129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.627975433793461e-05,
+      "loss": 1.0014,
+      "step": 4130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6278013796818074e-05,
+      "loss": 1.0695,
+      "step": 4131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6276272941728024e-05,
+      "loss": 0.9723,
+      "step": 4132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6274531772751522e-05,
+      "loss": 1.0641,
+      "step": 4133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.627279028997565e-05,
+      "loss": 0.9842,
+      "step": 4134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6271048493487502e-05,
+      "loss": 0.9042,
+      "step": 4135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.626930638337419e-05,
+      "loss": 0.9415,
+      "step": 4136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6267563959722833e-05,
+      "loss": 0.6815,
+      "step": 4137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6265821222620577e-05,
+      "loss": 0.9635,
+      "step": 4138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6264078172154577e-05,
+      "loss": 0.9186,
+      "step": 4139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6262334808412007e-05,
+      "loss": 0.9054,
+      "step": 4140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6260591131480056e-05,
+      "loss": 0.9625,
+      "step": 4141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.625884714144593e-05,
+      "loss": 0.9832,
+      "step": 4142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6257102838396843e-05,
+      "loss": 0.9985,
+      "step": 4143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6255358222420032e-05,
+      "loss": 1.0472,
+      "step": 4144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6253613293602753e-05,
+      "loss": 1.0122,
+      "step": 4145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6251868052032267e-05,
+      "loss": 0.9252,
+      "step": 4146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6250122497795858e-05,
+      "loss": 1.0664,
+      "step": 4147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6248376630980825e-05,
+      "loss": 1.0536,
+      "step": 4148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6246630451674485e-05,
+      "loss": 0.993,
+      "step": 4149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6244883959964158e-05,
+      "loss": 1.0223,
+      "step": 4150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6243137155937195e-05,
+      "loss": 0.975,
+      "step": 4151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624139003968096e-05,
+      "loss": 0.9759,
+      "step": 4152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6239642611282822e-05,
+      "loss": 0.9999,
+      "step": 4153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6237894870830175e-05,
+      "loss": 1.039,
+      "step": 4154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6236146818410424e-05,
+      "loss": 1.0606,
+      "step": 4155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6234398454110998e-05,
+      "loss": 0.9134,
+      "step": 4156
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623264977801933e-05,
+      "loss": 0.9077,
+      "step": 4157
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623090079022288e-05,
+      "loss": 0.9648,
+      "step": 4158
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622915149080911e-05,
+      "loss": 0.8978,
+      "step": 4159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6227401879865512e-05,
+      "loss": 1.0133,
+      "step": 4160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622565195747958e-05,
+      "loss": 0.8607,
+      "step": 4161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622390172373884e-05,
+      "loss": 0.9798,
+      "step": 4162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6222151178730814e-05,
+      "loss": 0.9214,
+      "step": 4163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6220400322543056e-05,
+      "loss": 1.0119,
+      "step": 4164
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6218649155263126e-05,
+      "loss": 0.9322,
+      "step": 4165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6216897676978604e-05,
+      "loss": 1.071,
+      "step": 4166
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6215145887777085e-05,
+      "loss": 0.9811,
+      "step": 4167
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6213393787746174e-05,
+      "loss": 0.9535,
+      "step": 4168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6211641376973505e-05,
+      "loss": 1.0032,
+      "step": 4169
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620988865554671e-05,
+      "loss": 0.9555,
+      "step": 4170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6208135623553452e-05,
+      "loss": 0.9779,
+      "step": 4171
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6206382281081397e-05,
+      "loss": 1.0073,
+      "step": 4172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620462862821824e-05,
+      "loss": 0.9137,
+      "step": 4173
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6202874665051675e-05,
+      "loss": 0.9636,
+      "step": 4174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620112039166943e-05,
+      "loss": 0.8842,
+      "step": 4175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619936580815923e-05,
+      "loss": 1.0177,
+      "step": 4176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619761091460883e-05,
+      "loss": 0.9722,
+      "step": 4177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6195855711105998e-05,
+      "loss": 0.9457,
+      "step": 4178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6194100197738505e-05,
+      "loss": 1.001,
+      "step": 4179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6192344374594152e-05,
+      "loss": 1.0323,
+      "step": 4180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6190588241760755e-05,
+      "loss": 0.9612,
+      "step": 4181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6188831799326132e-05,
+      "loss": 0.6716,
+      "step": 4182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6187075047378135e-05,
+      "loss": 0.6954,
+      "step": 4183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6185317986004615e-05,
+      "loss": 0.6628,
+      "step": 4184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6183560615293446e-05,
+      "loss": 0.9302,
+      "step": 4185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6181802935332524e-05,
+      "loss": 0.9772,
+      "step": 4186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6180044946209746e-05,
+      "loss": 1.0729,
+      "step": 4187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.617828664801303e-05,
+      "loss": 0.9779,
+      "step": 4188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.617652804083032e-05,
+      "loss": 1.0559,
+      "step": 4189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.617476912474956e-05,
+      "loss": 1.0787,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6173009899858717e-05,
+      "loss": 0.8704,
+      "step": 4191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.617125036624578e-05,
+      "loss": 0.9972,
+      "step": 4192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6169490523998733e-05,
+      "loss": 0.9529,
+      "step": 4193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.61677303732056e-05,
+      "loss": 0.9522,
+      "step": 4194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.61659699139544e-05,
+      "loss": 0.9139,
+      "step": 4195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6164209146333183e-05,
+      "loss": 0.8455,
+      "step": 4196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6162448070430004e-05,
+      "loss": 0.9643,
+      "step": 4197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.616068668633294e-05,
+      "loss": 1.0285,
+      "step": 4198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.615892499413008e-05,
+      "loss": 1.0188,
+      "step": 4199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6157162993909532e-05,
+      "loss": 0.9839,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6155400685759407e-05,
+      "loss": 0.9404,
+      "step": 4201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6153638069767852e-05,
+      "loss": 0.9415,
+      "step": 4202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.615187514602301e-05,
+      "loss": 0.9739,
+      "step": 4203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.615011191461305e-05,
+      "loss": 1.033,
+      "step": 4204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.614834837562616e-05,
+      "loss": 1.027,
+      "step": 4205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6146584529150527e-05,
+      "loss": 1.1175,
+      "step": 4206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.614482037527437e-05,
+      "loss": 0.9682,
+      "step": 4207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6143055914085917e-05,
+      "loss": 1.0492,
+      "step": 4208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.614129114567341e-05,
+      "loss": 0.8712,
+      "step": 4209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.613952607012511e-05,
+      "loss": 0.8579,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6137760687529285e-05,
+      "loss": 0.9094,
+      "step": 4211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6135994997974235e-05,
+      "loss": 0.9583,
+      "step": 4212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6134229001548256e-05,
+      "loss": 1.0463,
+      "step": 4213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6132462698339672e-05,
+      "loss": 0.9526,
+      "step": 4214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.613069608843682e-05,
+      "loss": 0.8997,
+      "step": 4215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.612892917192805e-05,
+      "loss": 1.0409,
+      "step": 4216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6127161948901724e-05,
+      "loss": 0.9877,
+      "step": 4217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6125394419446226e-05,
+      "loss": 1.0253,
+      "step": 4218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.612362658364996e-05,
+      "loss": 0.9644,
+      "step": 4219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.612185844160133e-05,
+      "loss": 0.954,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6120089993388767e-05,
+      "loss": 1.0181,
+      "step": 4221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6118321239100713e-05,
+      "loss": 1.0207,
+      "step": 4222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6116552178825622e-05,
+      "loss": 0.9847,
+      "step": 4223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.611478281265198e-05,
+      "loss": 1.0678,
+      "step": 4224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.611301314066826e-05,
+      "loss": 0.9514,
+      "step": 4225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6111243162962976e-05,
+      "loss": 0.6678,
+      "step": 4226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6109472879624646e-05,
+      "loss": 0.9127,
+      "step": 4227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6107702290741803e-05,
+      "loss": 0.944,
+      "step": 4228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6105931396403e-05,
+      "loss": 0.9782,
+      "step": 4229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6104160196696796e-05,
+      "loss": 0.9345,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.610238869171178e-05,
+      "loss": 0.912,
+      "step": 4231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6100616881536543e-05,
+      "loss": 1.0316,
+      "step": 4232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6098844766259695e-05,
+      "loss": 0.9512,
+      "step": 4233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6097072345969864e-05,
+      "loss": 1.0759,
+      "step": 4234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6095299620755694e-05,
+      "loss": 1.0095,
+      "step": 4235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6093526590705835e-05,
+      "loss": 0.8912,
+      "step": 4236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6091753255908967e-05,
+      "loss": 0.6555,
+      "step": 4237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.608997961645377e-05,
+      "loss": 0.9874,
+      "step": 4238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.608820567242895e-05,
+      "loss": 0.9277,
+      "step": 4239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.608643142392323e-05,
+      "loss": 1.0072,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6084656871025333e-05,
+      "loss": 0.9455,
+      "step": 4241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6082882013824014e-05,
+      "loss": 0.9376,
+      "step": 4242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6081106852408036e-05,
+      "loss": 1.0372,
+      "step": 4243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6079331386866176e-05,
+      "loss": 0.9002,
+      "step": 4244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6077555617287227e-05,
+      "loss": 1.0609,
+      "step": 4245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.607577954376e-05,
+      "loss": 0.9901,
+      "step": 4246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6074003166373318e-05,
+      "loss": 0.9552,
+      "step": 4247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.607222648521602e-05,
+      "loss": 0.8903,
+      "step": 4248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6070449500376963e-05,
+      "loss": 0.8995,
+      "step": 4249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6068672211945015e-05,
+      "loss": 0.9601,
+      "step": 4250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6066894620009063e-05,
+      "loss": 0.9686,
+      "step": 4251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6065116724658005e-05,
+      "loss": 0.6856,
+      "step": 4252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6063338525980755e-05,
+      "loss": 0.9423,
+      "step": 4253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.606156002406625e-05,
+      "loss": 0.971,
+      "step": 4254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6059781219003428e-05,
+      "loss": 0.9288,
+      "step": 4255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6058002110881254e-05,
+      "loss": 0.9976,
+      "step": 4256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6056222699788705e-05,
+      "loss": 1.0674,
+      "step": 4257
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6054442985814767e-05,
+      "loss": 0.9959,
+      "step": 4258
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6052662969048452e-05,
+      "loss": 0.8887,
+      "step": 4259
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.605088264957878e-05,
+      "loss": 0.6882,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.604910202749479e-05,
+      "loss": 1.0198,
+      "step": 4261
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6047321102885523e-05,
+      "loss": 0.6876,
+      "step": 4262
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.604553987584006e-05,
+      "loss": 1.0497,
+      "step": 4263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6043758346447476e-05,
+      "loss": 0.9169,
+      "step": 4264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6041976514796865e-05,
+      "loss": 1.0064,
+      "step": 4265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6040194380977347e-05,
+      "loss": 1.019,
+      "step": 4266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6038411945078047e-05,
+      "loss": 1.0418,
+      "step": 4267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6036629207188102e-05,
+      "loss": 1.0204,
+      "step": 4268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6034846167396675e-05,
+      "loss": 1.0166,
+      "step": 4269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6033062825792935e-05,
+      "loss": 0.6835,
+      "step": 4270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6031279182466075e-05,
+      "loss": 0.6903,
+      "step": 4271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6029495237505293e-05,
+      "loss": 0.9478,
+      "step": 4272
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.602771099099981e-05,
+      "loss": 0.9855,
+      "step": 4273
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6025926443038855e-05,
+      "loss": 0.9833,
+      "step": 4274
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6024141593711678e-05,
+      "loss": 0.9588,
+      "step": 4275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.602235644310755e-05,
+      "loss": 0.9436,
+      "step": 4276
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6020570991315735e-05,
+      "loss": 1.064,
+      "step": 4277
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.601878523842554e-05,
+      "loss": 0.9499,
+      "step": 4278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6016999184526263e-05,
+      "loss": 0.8086,
+      "step": 4279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6015212829707232e-05,
+      "loss": 0.9006,
+      "step": 4280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6013426174057786e-05,
+      "loss": 1.0045,
+      "step": 4281
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6011639217667275e-05,
+      "loss": 1.0271,
+      "step": 4282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.600985196062507e-05,
+      "loss": 1.0477,
+      "step": 4283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6008064403020555e-05,
+      "loss": 1.0108,
+      "step": 4284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.600627654494313e-05,
+      "loss": 0.8664,
+      "step": 4285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6004488386482204e-05,
+      "loss": 1.0479,
+      "step": 4286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.600269992772721e-05,
+      "loss": 0.6398,
+      "step": 4287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.600091116876759e-05,
+      "loss": 0.9346,
+      "step": 4288
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59991221096928e-05,
+      "loss": 0.6243,
+      "step": 4289
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5997332750592318e-05,
+      "loss": 1.0096,
+      "step": 4290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5995543091555633e-05,
+      "loss": 1.0462,
+      "step": 4291
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5993753132672245e-05,
+      "loss": 0.9062,
+      "step": 4292
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5991962874031673e-05,
+      "loss": 1.0169,
+      "step": 4293
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599017231572346e-05,
+      "loss": 0.9566,
+      "step": 4294
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598838145783714e-05,
+      "loss": 0.8937,
+      "step": 4295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5986590300462285e-05,
+      "loss": 0.6725,
+      "step": 4296
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598479884368847e-05,
+      "loss": 0.8494,
+      "step": 4297
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5983007087605292e-05,
+      "loss": 0.9289,
+      "step": 4298
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598121503230236e-05,
+      "loss": 1.0309,
+      "step": 4299
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59794226778693e-05,
+      "loss": 0.9364,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597763002439574e-05,
+      "loss": 0.6628,
+      "step": 4301
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597583707197134e-05,
+      "loss": 0.9957,
+      "step": 4302
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597404382068577e-05,
+      "loss": 0.8703,
+      "step": 4303
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5972250270628717e-05,
+      "loss": 1.0087,
+      "step": 4304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5970456421889867e-05,
+      "loss": 0.9753,
+      "step": 4305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596866227455894e-05,
+      "loss": 0.9508,
+      "step": 4306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596686782872567e-05,
+      "loss": 0.6958,
+      "step": 4307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596507308447979e-05,
+      "loss": 0.9174,
+      "step": 4308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5963278041911065e-05,
+      "loss": 0.8941,
+      "step": 4309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5961482701109266e-05,
+      "loss": 1.0163,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.595968706216418e-05,
+      "loss": 1.0569,
+      "step": 4311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.595789112516561e-05,
+      "loss": 0.944,
+      "step": 4312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5956094890203372e-05,
+      "loss": 0.8821,
+      "step": 4313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5954298357367305e-05,
+      "loss": 1.0422,
+      "step": 4314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5952501526747245e-05,
+      "loss": 0.9968,
+      "step": 4315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.595070439843307e-05,
+      "loss": 0.9683,
+      "step": 4316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5948906972514643e-05,
+      "loss": 0.9296,
+      "step": 4317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5947109249081863e-05,
+      "loss": 0.9148,
+      "step": 4318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5945311228224637e-05,
+      "loss": 1.0074,
+      "step": 4319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5943512910032883e-05,
+      "loss": 0.9639,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5941714294596543e-05,
+      "loss": 0.93,
+      "step": 4321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5939915382005565e-05,
+      "loss": 1.0305,
+      "step": 4322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5938116172349917e-05,
+      "loss": 1.0176,
+      "step": 4323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5936316665719577e-05,
+      "loss": 0.9724,
+      "step": 4324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593451686220455e-05,
+      "loss": 0.9467,
+      "step": 4325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5932716761894838e-05,
+      "loss": 0.9708,
+      "step": 4326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593091636488047e-05,
+      "loss": 1.101,
+      "step": 4327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5929115671251487e-05,
+      "loss": 1.0155,
+      "step": 4328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5927314681097943e-05,
+      "loss": 0.9866,
+      "step": 4329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5925513394509912e-05,
+      "loss": 0.9416,
+      "step": 4330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5923711811577474e-05,
+      "loss": 0.9626,
+      "step": 4331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5921909932390736e-05,
+      "loss": 1.016,
+      "step": 4332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.592010775703981e-05,
+      "loss": 0.953,
+      "step": 4333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.591830528561482e-05,
+      "loss": 0.9641,
+      "step": 4334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5916502518205915e-05,
+      "loss": 0.9974,
+      "step": 4335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.591469945490326e-05,
+      "loss": 1.0216,
+      "step": 4336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5912896095797018e-05,
+      "loss": 1.0331,
+      "step": 4337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5911092440977386e-05,
+      "loss": 1.0806,
+      "step": 4338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5909288490534565e-05,
+      "loss": 0.9259,
+      "step": 4339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.590748424455877e-05,
+      "loss": 0.977,
+      "step": 4340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5905679703140242e-05,
+      "loss": 0.6275,
+      "step": 4341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5903874866369223e-05,
+      "loss": 1.0378,
+      "step": 4342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5902069734335976e-05,
+      "loss": 0.993,
+      "step": 4343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.590026430713078e-05,
+      "loss": 0.9004,
+      "step": 4344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.589845858484393e-05,
+      "loss": 0.9819,
+      "step": 4345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.589665256756573e-05,
+      "loss": 0.9157,
+      "step": 4346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5894846255386503e-05,
+      "loss": 0.9426,
+      "step": 4347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.589303964839658e-05,
+      "loss": 1.0435,
+      "step": 4348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.589123274668632e-05,
+      "loss": 0.9451,
+      "step": 4349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.588942555034609e-05,
+      "loss": 0.9745,
+      "step": 4350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5887618059466263e-05,
+      "loss": 0.8769,
+      "step": 4351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5885810274137233e-05,
+      "loss": 0.9975,
+      "step": 4352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5884002194449425e-05,
+      "loss": 0.9355,
+      "step": 4353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.588219382049325e-05,
+      "loss": 1.0351,
+      "step": 4354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.588038515235915e-05,
+      "loss": 1.0189,
+      "step": 4355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5878576190137584e-05,
+      "loss": 1.0663,
+      "step": 4356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5876766933919023e-05,
+      "loss": 1.0324,
+      "step": 4357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.587495738379394e-05,
+      "loss": 0.9935,
+      "step": 4358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.587314753985284e-05,
+      "loss": 0.9599,
+      "step": 4359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.587133740218624e-05,
+      "loss": 0.9622,
+      "step": 4360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5869526970884664e-05,
+      "loss": 0.9837,
+      "step": 4361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5867716246038654e-05,
+      "loss": 1.0647,
+      "step": 4362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5865905227738767e-05,
+      "loss": 0.8846,
+      "step": 4363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.586409391607558e-05,
+      "loss": 0.9873,
+      "step": 4364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.586228231113967e-05,
+      "loss": 0.947,
+      "step": 4365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5860470413021644e-05,
+      "loss": 0.9794,
+      "step": 4366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5858658221812118e-05,
+      "loss": 1.0118,
+      "step": 4367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5856845737601722e-05,
+      "loss": 0.907,
+      "step": 4368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5855032960481102e-05,
+      "loss": 1.0083,
+      "step": 4369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5853219890540918e-05,
+      "loss": 1.0033,
+      "step": 4370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5851406527871846e-05,
+      "loss": 0.9644,
+      "step": 4371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5849592872564566e-05,
+      "loss": 0.9644,
+      "step": 4372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5847778924709796e-05,
+      "loss": 0.9445,
+      "step": 4373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5845964684398242e-05,
+      "loss": 1.0103,
+      "step": 4374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5844150151720646e-05,
+      "loss": 1.01,
+      "step": 4375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5842335326767752e-05,
+      "loss": 1.0319,
+      "step": 4376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.584052020963032e-05,
+      "loss": 1.0433,
+      "step": 4377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.583870480039913e-05,
+      "loss": 0.9491,
+      "step": 4378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5836889099164974e-05,
+      "loss": 0.9805,
+      "step": 4379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5835073106018655e-05,
+      "loss": 1.0255,
+      "step": 4380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5833256821050994e-05,
+      "loss": 0.8964,
+      "step": 4381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5831440244352832e-05,
+      "loss": 0.9657,
+      "step": 4382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5829623376015013e-05,
+      "loss": 0.9754,
+      "step": 4383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.58278062161284e-05,
+      "loss": 0.9783,
+      "step": 4384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.582598876478388e-05,
+      "loss": 0.9562,
+      "step": 4385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.582417102207234e-05,
+      "loss": 1.0136,
+      "step": 4386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.582235298808469e-05,
+      "loss": 0.643,
+      "step": 4387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5820534662911856e-05,
+      "loss": 0.9925,
+      "step": 4388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5818716046644768e-05,
+      "loss": 0.6837,
+      "step": 4389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5816897139374385e-05,
+      "loss": 0.9817,
+      "step": 4390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.581507794119167e-05,
+      "loss": 0.9468,
+      "step": 4391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5813258452187605e-05,
+      "loss": 0.9463,
+      "step": 4392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5811438672453182e-05,
+      "loss": 1.0838,
+      "step": 4393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5809618602079423e-05,
+      "loss": 1.0344,
+      "step": 4394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5807798241157337e-05,
+      "loss": 0.9689,
+      "step": 4395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.580597758977797e-05,
+      "loss": 1.0369,
+      "step": 4396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.580415664803238e-05,
+      "loss": 1.0621,
+      "step": 4397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5802335416011625e-05,
+      "loss": 0.9859,
+      "step": 4398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5800513893806796e-05,
+      "loss": 0.9757,
+      "step": 4399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5798692081508988e-05,
+      "loss": 0.9288,
+      "step": 4400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5796869979209314e-05,
+      "loss": 0.889,
+      "step": 4401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5795047586998903e-05,
+      "loss": 0.9226,
+      "step": 4402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5793224904968885e-05,
+      "loss": 1.0527,
+      "step": 4403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5791401933210427e-05,
+      "loss": 0.9903,
+      "step": 4404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.578957867181469e-05,
+      "loss": 0.9492,
+      "step": 4405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5787755120872864e-05,
+      "loss": 0.8647,
+      "step": 4406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5785931280476145e-05,
+      "loss": 0.9105,
+      "step": 4407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5784107150715746e-05,
+      "loss": 0.6812,
+      "step": 4408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5782282731682896e-05,
+      "loss": 1.035,
+      "step": 4409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5780458023468838e-05,
+      "loss": 0.9197,
+      "step": 4410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5778633026164826e-05,
+      "loss": 0.6502,
+      "step": 4411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5776807739862134e-05,
+      "loss": 1.0208,
+      "step": 4412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5774982164652042e-05,
+      "loss": 0.9757,
+      "step": 4413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5773156300625857e-05,
+      "loss": 0.9462,
+      "step": 4414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5771330147874888e-05,
+      "loss": 1.0255,
+      "step": 4415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5769503706490466e-05,
+      "loss": 1.0314,
+      "step": 4416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5767676976563935e-05,
+      "loss": 1.0982,
+      "step": 4417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5765849958186653e-05,
+      "loss": 0.9604,
+      "step": 4418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5764022651449987e-05,
+      "loss": 1.0017,
+      "step": 4419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5762195056445333e-05,
+      "loss": 0.8585,
+      "step": 4420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5760367173264085e-05,
+      "loss": 0.8899,
+      "step": 4421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5758539001997654e-05,
+      "loss": 0.8967,
+      "step": 4422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.575671054273748e-05,
+      "loss": 0.9953,
+      "step": 4423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5754881795575006e-05,
+      "loss": 0.6693,
+      "step": 4424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.575305276060168e-05,
+      "loss": 1.0874,
+      "step": 4425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.575122343790899e-05,
+      "loss": 1.032,
+      "step": 4426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5749393827588408e-05,
+      "loss": 0.9445,
+      "step": 4427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5747563929731447e-05,
+      "loss": 0.9524,
+      "step": 4428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574573374442962e-05,
+      "loss": 1.0485,
+      "step": 4429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5743903271774453e-05,
+      "loss": 0.9259,
+      "step": 4430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.57420725118575e-05,
+      "loss": 0.8861,
+      "step": 4431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5740241464770313e-05,
+      "loss": 1.0136,
+      "step": 4432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5738410130604466e-05,
+      "loss": 0.9954,
+      "step": 4433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573657850945155e-05,
+      "loss": 1.0617,
+      "step": 4434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5734746601403167e-05,
+      "loss": 1.0075,
+      "step": 4435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5732914406550934e-05,
+      "loss": 0.9518,
+      "step": 4436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5731081924986478e-05,
+      "loss": 0.895,
+      "step": 4437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5729249156801444e-05,
+      "loss": 1.0125,
+      "step": 4438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.57274161020875e-05,
+      "loss": 0.9868,
+      "step": 4439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5725582760936312e-05,
+      "loss": 0.9673,
+      "step": 4440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.572374913343957e-05,
+      "loss": 0.9443,
+      "step": 4441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5721915219688976e-05,
+      "loss": 1.0571,
+      "step": 4442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5720081019776254e-05,
+      "loss": 0.6462,
+      "step": 4443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5718246533793123e-05,
+      "loss": 1.0168,
+      "step": 4444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571641176183134e-05,
+      "loss": 0.9692,
+      "step": 4445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571457670398266e-05,
+      "loss": 0.9243,
+      "step": 4446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5712741360338856e-05,
+      "loss": 0.9626,
+      "step": 4447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5710905730991716e-05,
+      "loss": 0.9194,
+      "step": 4448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570906981603305e-05,
+      "loss": 0.992,
+      "step": 4449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5707233615554663e-05,
+      "loss": 1.0305,
+      "step": 4450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.57053971296484e-05,
+      "loss": 0.9811,
+      "step": 4451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5703560358406095e-05,
+      "loss": 0.9674,
+      "step": 4452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5701723301919614e-05,
+      "loss": 1.0438,
+      "step": 4453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5699885960280826e-05,
+      "loss": 0.9377,
+      "step": 4454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5698048333581625e-05,
+      "loss": 1.031,
+      "step": 4455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5696210421913916e-05,
+      "loss": 0.6499,
+      "step": 4456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5694372225369606e-05,
+      "loss": 0.985,
+      "step": 4457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5692533744040635e-05,
+      "loss": 0.9697,
+      "step": 4458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5690694978018944e-05,
+      "loss": 0.9765,
+      "step": 4459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5688855927396494e-05,
+      "loss": 1.0756,
+      "step": 4460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568701659226526e-05,
+      "loss": 0.9745,
+      "step": 4461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5685176972717222e-05,
+      "loss": 0.6737,
+      "step": 4462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5683337068844396e-05,
+      "loss": 1.0738,
+      "step": 4463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568149688073879e-05,
+      "loss": 0.9255,
+      "step": 4464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.567965640849243e-05,
+      "loss": 0.9757,
+      "step": 4465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5677815652197374e-05,
+      "loss": 0.9442,
+      "step": 4466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5675974611945674e-05,
+      "loss": 0.9473,
+      "step": 4467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56741332878294e-05,
+      "loss": 1.0523,
+      "step": 4468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5672291679940646e-05,
+      "loss": 0.892,
+      "step": 4469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5670449788371517e-05,
+      "loss": 0.967,
+      "step": 4470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5668607613214114e-05,
+      "loss": 0.6432,
+      "step": 4471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.566676515456058e-05,
+      "loss": 0.9751,
+      "step": 4472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5664922412503053e-05,
+      "loss": 0.9665,
+      "step": 4473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5663079387133696e-05,
+      "loss": 0.923,
+      "step": 4474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5661236078544678e-05,
+      "loss": 0.899,
+      "step": 4475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.565939248682819e-05,
+      "loss": 0.886,
+      "step": 4476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5657548612076428e-05,
+      "loss": 0.9661,
+      "step": 4477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.565570445438161e-05,
+      "loss": 0.8649,
+      "step": 4478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5653860013835968e-05,
+      "loss": 0.9178,
+      "step": 4479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5652015290531738e-05,
+      "loss": 0.9697,
+      "step": 4480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5650170284561182e-05,
+      "loss": 1.0467,
+      "step": 4481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5648324996016572e-05,
+      "loss": 0.9813,
+      "step": 4482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.564647942499019e-05,
+      "loss": 1.0428,
+      "step": 4483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.564463357157435e-05,
+      "loss": 0.6911,
+      "step": 4484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5642787435861345e-05,
+      "loss": 0.9978,
+      "step": 4485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5640941017943516e-05,
+      "loss": 0.9453,
+      "step": 4486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5639094317913204e-05,
+      "loss": 1.0254,
+      "step": 4487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5637247335862765e-05,
+      "loss": 0.9786,
+      "step": 4488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.563540007188457e-05,
+      "loss": 0.8844,
+      "step": 4489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5633552526070995e-05,
+      "loss": 0.9555,
+      "step": 4490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5631704698514454e-05,
+      "loss": 1.0241,
+      "step": 4491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5629856589307348e-05,
+      "loss": 0.9541,
+      "step": 4492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5628008198542106e-05,
+      "loss": 1.0324,
+      "step": 4493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5626159526311173e-05,
+      "loss": 0.9592,
+      "step": 4494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5624310572707005e-05,
+      "loss": 0.8686,
+      "step": 4495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5622461337822064e-05,
+      "loss": 1.051,
+      "step": 4496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.562061182174884e-05,
+      "loss": 1.0076,
+      "step": 4497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5618762024579825e-05,
+      "loss": 0.9113,
+      "step": 4498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5616911946407537e-05,
+      "loss": 0.9833,
+      "step": 4499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5615061587324493e-05,
+      "loss": 0.8883,
+      "step": 4500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5613210947423233e-05,
+      "loss": 0.9305,
+      "step": 4501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.561136002679632e-05,
+      "loss": 0.9908,
+      "step": 4502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5609508825536314e-05,
+      "loss": 0.9501,
+      "step": 4503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.56076573437358e-05,
+      "loss": 0.9213,
+      "step": 4504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.560580558148737e-05,
+      "loss": 0.9087,
+      "step": 4505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5603953538883635e-05,
+      "loss": 0.8894,
+      "step": 4506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5602101216017217e-05,
+      "loss": 1.0316,
+      "step": 4507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.560024861298076e-05,
+      "loss": 0.8899,
+      "step": 4508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5598395729866904e-05,
+      "loss": 1.0211,
+      "step": 4509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5596542566768326e-05,
+      "loss": 0.9933,
+      "step": 4510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5594689123777702e-05,
+      "loss": 1.0348,
+      "step": 4511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5592835400987725e-05,
+      "loss": 0.9588,
+      "step": 4512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5590981398491104e-05,
+      "loss": 0.9328,
+      "step": 4513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5589127116380555e-05,
+      "loss": 0.9782,
+      "step": 4514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.558727255474882e-05,
+      "loss": 1.0977,
+      "step": 4515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5585417713688648e-05,
+      "loss": 1.0397,
+      "step": 4516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5583562593292797e-05,
+      "loss": 1.0428,
+      "step": 4517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5581707193654056e-05,
+      "loss": 1.0491,
+      "step": 4518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.55798515148652e-05,
+      "loss": 0.9809,
+      "step": 4519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5577995557019053e-05,
+      "loss": 0.9918,
+      "step": 4520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.557613932020842e-05,
+      "loss": 0.8506,
+      "step": 4521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.557428280452614e-05,
+      "loss": 0.9458,
+      "step": 4522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5572426010065057e-05,
+      "loss": 0.9103,
+      "step": 4523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.557056893691804e-05,
+      "loss": 1.0468,
+      "step": 4524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5568711585177952e-05,
+      "loss": 0.8443,
+      "step": 4525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5566853954937694e-05,
+      "loss": 0.8535,
+      "step": 4526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5564996046290164e-05,
+      "loss": 0.9975,
+      "step": 4527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5563137859328282e-05,
+      "loss": 0.8663,
+      "step": 4528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5561279394144975e-05,
+      "loss": 0.9189,
+      "step": 4529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5559420650833188e-05,
+      "loss": 0.6397,
+      "step": 4530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.555756162948588e-05,
+      "loss": 0.8896,
+      "step": 4531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 0.9205,
+      "step": 4532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.555384275305661e-05,
+      "loss": 0.9047,
+      "step": 4533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5551982898160632e-05,
+      "loss": 0.9506,
+      "step": 4534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5550122765601106e-05,
+      "loss": 1.0093,
+      "step": 4535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5548262355471065e-05,
+      "loss": 0.8661,
+      "step": 4536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.554640166786355e-05,
+      "loss": 0.9346,
+      "step": 4537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5544540702871612e-05,
+      "loss": 1.0289,
+      "step": 4538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5542679460588326e-05,
+      "loss": 0.9472,
+      "step": 4539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5540817941106773e-05,
+      "loss": 0.8985,
+      "step": 4540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5538956144520048e-05,
+      "loss": 1.0168,
+      "step": 4541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5537094070921267e-05,
+      "loss": 1.0255,
+      "step": 4542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5535231720403556e-05,
+      "loss": 0.9099,
+      "step": 4543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5533369093060052e-05,
+      "loss": 1.0028,
+      "step": 4544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5531506188983904e-05,
+      "loss": 1.0336,
+      "step": 4545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5529643008268287e-05,
+      "loss": 1.0078,
+      "step": 4546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5527779551006374e-05,
+      "loss": 0.944,
+      "step": 4547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5525915817291365e-05,
+      "loss": 0.8921,
+      "step": 4548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5524051807216465e-05,
+      "loss": 0.9619,
+      "step": 4549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5522187520874898e-05,
+      "loss": 0.9618,
+      "step": 4550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5520322958359898e-05,
+      "loss": 0.9552,
+      "step": 4551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5518458119764718e-05,
+      "loss": 0.9029,
+      "step": 4552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5516593005182616e-05,
+      "loss": 0.9044,
+      "step": 4553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5514727614706876e-05,
+      "loss": 0.9841,
+      "step": 4554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5512861948430785e-05,
+      "loss": 0.9731,
+      "step": 4555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.551099600644765e-05,
+      "loss": 1.0222,
+      "step": 4556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5509129788850786e-05,
+      "loss": 0.974,
+      "step": 4557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.550726329573353e-05,
+      "loss": 0.9142,
+      "step": 4558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5505396527189225e-05,
+      "loss": 0.8643,
+      "step": 4559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5503529483311234e-05,
+      "loss": 0.9034,
+      "step": 4560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5501662164192933e-05,
+      "loss": 0.9875,
+      "step": 4561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.54997945699277e-05,
+      "loss": 0.9147,
+      "step": 4562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5497926700608944e-05,
+      "loss": 0.6645,
+      "step": 4563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5496058556330082e-05,
+      "loss": 0.9818,
+      "step": 4564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5494190137184535e-05,
+      "loss": 0.9713,
+      "step": 4565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5492321443265753e-05,
+      "loss": 0.9864,
+      "step": 4566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.549045247466719e-05,
+      "loss": 1.0262,
+      "step": 4567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5488583231482313e-05,
+      "loss": 0.9778,
+      "step": 4568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5486713713804613e-05,
+      "loss": 0.9415,
+      "step": 4569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5484843921727578e-05,
+      "loss": 0.9402,
+      "step": 4570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5482973855344726e-05,
+      "loss": 0.9166,
+      "step": 4571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.548110351474958e-05,
+      "loss": 0.6113,
+      "step": 4572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5479232900035678e-05,
+      "loss": 0.9087,
+      "step": 4573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5477362011296578e-05,
+      "loss": 0.989,
+      "step": 4574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5475490848625837e-05,
+      "loss": 0.9683,
+      "step": 4575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5473619412117042e-05,
+      "loss": 0.9669,
+      "step": 4576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5471747701863784e-05,
+      "loss": 1.0683,
+      "step": 4577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546987571795967e-05,
+      "loss": 1.073,
+      "step": 4578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5468003460498318e-05,
+      "loss": 0.8936,
+      "step": 4579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546613092957337e-05,
+      "loss": 0.9869,
+      "step": 4580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5464258125278465e-05,
+      "loss": 1.0034,
+      "step": 4581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5462385047707276e-05,
+      "loss": 1.11,
+      "step": 4582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546051169695347e-05,
+      "loss": 0.8494,
+      "step": 4583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.545863807311074e-05,
+      "loss": 1.1009,
+      "step": 4584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5456764176272787e-05,
+      "loss": 1.0322,
+      "step": 4585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5454890006533324e-05,
+      "loss": 1.0364,
+      "step": 4586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.545301556398609e-05,
+      "loss": 1.0301,
+      "step": 4587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5451140848724822e-05,
+      "loss": 0.9233,
+      "step": 4588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544926586084328e-05,
+      "loss": 0.9536,
+      "step": 4589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544739060043524e-05,
+      "loss": 1.0164,
+      "step": 4590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5445515067594477e-05,
+      "loss": 0.9736,
+      "step": 4591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5443639262414797e-05,
+      "loss": 1.0249,
+      "step": 4592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5441763184990007e-05,
+      "loss": 1.0627,
+      "step": 4593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5439886835413935e-05,
+      "loss": 1.0411,
+      "step": 4594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.543801021378042e-05,
+      "loss": 1.0255,
+      "step": 4595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5436133320183316e-05,
+      "loss": 0.9494,
+      "step": 4596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.543425615471649e-05,
+      "loss": 1.0027,
+      "step": 4597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5432378717473818e-05,
+      "loss": 1.0708,
+      "step": 4598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.54305010085492e-05,
+      "loss": 0.9221,
+      "step": 4599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5428623028036533e-05,
+      "loss": 1.0622,
+      "step": 4600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5426744776029746e-05,
+      "loss": 1.0287,
+      "step": 4601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5424866252622775e-05,
+      "loss": 1.0415,
+      "step": 4602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542298745790956e-05,
+      "loss": 0.8817,
+      "step": 4603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542110839198407e-05,
+      "loss": 0.6837,
+      "step": 4604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5419229054940274e-05,
+      "loss": 0.9813,
+      "step": 4605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5417349446872167e-05,
+      "loss": 1.0331,
+      "step": 4606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5415469567873748e-05,
+      "loss": 1.0802,
+      "step": 4607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.541358941803903e-05,
+      "loss": 0.8527,
+      "step": 4608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.541170899746205e-05,
+      "loss": 0.9222,
+      "step": 4609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5409828306236843e-05,
+      "loss": 1.0393,
+      "step": 4610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5407947344457465e-05,
+      "loss": 1.0004,
+      "step": 4611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5406066112217993e-05,
+      "loss": 0.914,
+      "step": 4612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5404184609612506e-05,
+      "loss": 0.7853,
+      "step": 4613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5402302836735102e-05,
+      "loss": 0.9551,
+      "step": 4614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5400420793679887e-05,
+      "loss": 0.9111,
+      "step": 4615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5398538480540994e-05,
+      "loss": 1.0511,
+      "step": 4616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5396655897412554e-05,
+      "loss": 0.9041,
+      "step": 4617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.539477304438872e-05,
+      "loss": 0.676,
+      "step": 4618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5392889921563654e-05,
+      "loss": 0.9865,
+      "step": 4619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.539100652903154e-05,
+      "loss": 0.9884,
+      "step": 4620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5389122866886558e-05,
+      "loss": 0.9925,
+      "step": 4621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.538723893522293e-05,
+      "loss": 0.8817,
+      "step": 4622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5385354734134857e-05,
+      "loss": 0.9013,
+      "step": 4623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5383470263716584e-05,
+      "loss": 0.8804,
+      "step": 4624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5381585524062345e-05,
+      "loss": 0.9438,
+      "step": 4625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.537970051526641e-05,
+      "loss": 0.9713,
+      "step": 4626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5377815237423045e-05,
+      "loss": 0.9454,
+      "step": 4627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5375929690626537e-05,
+      "loss": 1.0068,
+      "step": 4628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5374043874971184e-05,
+      "loss": 0.9348,
+      "step": 4629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.53721577905513e-05,
+      "loss": 1.0647,
+      "step": 4630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5370271437461216e-05,
+      "loss": 0.8735,
+      "step": 4631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.536838481579526e-05,
+      "loss": 0.9888,
+      "step": 4632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5366497925647794e-05,
+      "loss": 0.6381,
+      "step": 4633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5364610767113182e-05,
+      "loss": 0.9862,
+      "step": 4634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5362723340285804e-05,
+      "loss": 0.9863,
+      "step": 4635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5360835645260053e-05,
+      "loss": 0.9211,
+      "step": 4636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5358947682130335e-05,
+      "loss": 0.9735,
+      "step": 4637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.535705945099107e-05,
+      "loss": 0.9813,
+      "step": 4638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5355170951936693e-05,
+      "loss": 0.9946,
+      "step": 4639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.535328218506165e-05,
+      "loss": 1.0027,
+      "step": 4640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.53513931504604e-05,
+      "loss": 1.0528,
+      "step": 4641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.534950384822742e-05,
+      "loss": 0.9409,
+      "step": 4642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5347614278457196e-05,
+      "loss": 1.0074,
+      "step": 4643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5345724441244226e-05,
+      "loss": 1.0915,
+      "step": 4644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5343834336683023e-05,
+      "loss": 0.9407,
+      "step": 4645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.534194396486812e-05,
+      "loss": 0.8972,
+      "step": 4646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5340053325894047e-05,
+      "loss": 0.9798,
+      "step": 4647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5338162419855368e-05,
+      "loss": 0.9609,
+      "step": 4648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.533627124684665e-05,
+      "loss": 0.9713,
+      "step": 4649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5334379806962467e-05,
+      "loss": 1.0145,
+      "step": 4650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.533248810029742e-05,
+      "loss": 1.0268,
+      "step": 4651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.533059612694611e-05,
+      "loss": 0.9104,
+      "step": 4652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.532870388700316e-05,
+      "loss": 1.027,
+      "step": 4653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5326811380563203e-05,
+      "loss": 1.0255,
+      "step": 4654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.532491860772089e-05,
+      "loss": 0.9508,
+      "step": 4655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.532302556857088e-05,
+      "loss": 0.9763,
+      "step": 4656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5321132263207844e-05,
+      "loss": 0.997,
+      "step": 4657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5319238691726473e-05,
+      "loss": 0.932,
+      "step": 4658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5317344854221462e-05,
+      "loss": 1.036,
+      "step": 4659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.531545075078753e-05,
+      "loss": 0.9164,
+      "step": 4660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.531355638151941e-05,
+      "loss": 0.976,
+      "step": 4661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5311661746511825e-05,
+      "loss": 1.0397,
+      "step": 4662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5309766845859543e-05,
+      "loss": 0.8245,
+      "step": 4663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.530787167965733e-05,
+      "loss": 1.0354,
+      "step": 4664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5305976247999958e-05,
+      "loss": 1.0794,
+      "step": 4665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5304080550982227e-05,
+      "loss": 1.0309,
+      "step": 4666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5302184588698944e-05,
+      "loss": 0.903,
+      "step": 4667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5300288361244926e-05,
+      "loss": 1.0393,
+      "step": 4668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.529839186871501e-05,
+      "loss": 1.0045,
+      "step": 4669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.529649511120404e-05,
+      "loss": 1.1353,
+      "step": 4670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.529459808880688e-05,
+      "loss": 0.8981,
+      "step": 4671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5292700801618395e-05,
+      "loss": 0.8416,
+      "step": 4672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.529080324973348e-05,
+      "loss": 1.072,
+      "step": 4673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5288905433247028e-05,
+      "loss": 0.9442,
+      "step": 4674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5287007352253956e-05,
+      "loss": 0.9728,
+      "step": 4675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5285109006849185e-05,
+      "loss": 0.9525,
+      "step": 4676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.528321039712766e-05,
+      "loss": 0.9079,
+      "step": 4677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5281311523184336e-05,
+      "loss": 0.9626,
+      "step": 4678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5279412385114166e-05,
+      "loss": 0.9586,
+      "step": 4679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5277512983012144e-05,
+      "loss": 0.9475,
+      "step": 4680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5275613316973253e-05,
+      "loss": 0.9747,
+      "step": 4681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5273713387092496e-05,
+      "loss": 0.9695,
+      "step": 4682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5271813193464904e-05,
+      "loss": 0.9799,
+      "step": 4683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5269912736185497e-05,
+      "loss": 0.9763,
+      "step": 4684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5268012015349324e-05,
+      "loss": 1.0159,
+      "step": 4685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.526611103105144e-05,
+      "loss": 1.0589,
+      "step": 4686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5264209783386922e-05,
+      "loss": 0.9221,
+      "step": 4687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5262308272450854e-05,
+      "loss": 0.9949,
+      "step": 4688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5260406498338326e-05,
+      "loss": 0.646,
+      "step": 4689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5258504461144456e-05,
+      "loss": 0.899,
+      "step": 4690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5256602160964369e-05,
+      "loss": 0.91,
+      "step": 4691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5254699597893194e-05,
+      "loss": 0.9411,
+      "step": 4692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.525279677202609e-05,
+      "loss": 0.9817,
+      "step": 4693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5250893683458214e-05,
+      "loss": 0.888,
+      "step": 4694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5248990332284745e-05,
+      "loss": 1.0465,
+      "step": 4695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5247086718600874e-05,
+      "loss": 0.7868,
+      "step": 4696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.52451828425018e-05,
+      "loss": 0.8816,
+      "step": 4697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5243278704082744e-05,
+      "loss": 1.0135,
+      "step": 4698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5241374303438932e-05,
+      "loss": 1.0212,
+      "step": 4699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5239469640665605e-05,
+      "loss": 0.6434,
+      "step": 4700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.523756471585802e-05,
+      "loss": 1.0567,
+      "step": 4701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5235659529111445e-05,
+      "loss": 1.0385,
+      "step": 4702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5233754080521162e-05,
+      "loss": 0.9892,
+      "step": 4703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5231848370182463e-05,
+      "loss": 0.9174,
+      "step": 4704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5229942398190657e-05,
+      "loss": 0.9341,
+      "step": 4705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5228036164641071e-05,
+      "loss": 0.9302,
+      "step": 4706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5226129669629028e-05,
+      "loss": 0.9801,
+      "step": 4707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5224222913249883e-05,
+      "loss": 1.0773,
+      "step": 4708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5222315895598988e-05,
+      "loss": 0.9187,
+      "step": 4709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5220408616771727e-05,
+      "loss": 0.9981,
+      "step": 4710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5218501076863477e-05,
+      "loss": 0.9478,
+      "step": 4711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.521659327596964e-05,
+      "loss": 1.1259,
+      "step": 4712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5214685214185627e-05,
+      "loss": 0.9963,
+      "step": 4713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5212776891606864e-05,
+      "loss": 1.0797,
+      "step": 4714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5210868308328793e-05,
+      "loss": 0.9876,
+      "step": 4715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5208959464446861e-05,
+      "loss": 0.9521,
+      "step": 4716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5207050360056532e-05,
+      "loss": 0.9498,
+      "step": 4717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5205140995253285e-05,
+      "loss": 0.8766,
+      "step": 4718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5203231370132612e-05,
+      "loss": 0.9153,
+      "step": 4719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5201321484790012e-05,
+      "loss": 0.6258,
+      "step": 4720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5199411339321008e-05,
+      "loss": 0.9422,
+      "step": 4721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5197500933821122e-05,
+      "loss": 0.9492,
+      "step": 4722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5195590268385898e-05,
+      "loss": 0.9569,
+      "step": 4723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5193679343110896e-05,
+      "loss": 0.8902,
+      "step": 4724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5191768158091683e-05,
+      "loss": 1.0932,
+      "step": 4725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5189856713423833e-05,
+      "loss": 1.0305,
+      "step": 4726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5187945009202954e-05,
+      "loss": 1.0025,
+      "step": 4727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5186033045524641e-05,
+      "loss": 0.884,
+      "step": 4728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.518412082248452e-05,
+      "loss": 0.9681,
+      "step": 4729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5182208340178225e-05,
+      "loss": 0.9906,
+      "step": 4730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5180295598701402e-05,
+      "loss": 1.0266,
+      "step": 4731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5178382598149708e-05,
+      "loss": 0.9223,
+      "step": 4732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5176469338618815e-05,
+      "loss": 0.9919,
+      "step": 4733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5174555820204409e-05,
+      "loss": 0.9127,
+      "step": 4734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.517264204300219e-05,
+      "loss": 0.9088,
+      "step": 4735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5170728007107869e-05,
+      "loss": 0.9191,
+      "step": 4736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5168813712617167e-05,
+      "loss": 1.009,
+      "step": 4737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5166899159625823e-05,
+      "loss": 0.9408,
+      "step": 4738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5164984348229586e-05,
+      "loss": 0.8797,
+      "step": 4739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5163069278524222e-05,
+      "loss": 0.9936,
+      "step": 4740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5161153950605501e-05,
+      "loss": 1.0718,
+      "step": 4741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5159238364569216e-05,
+      "loss": 1.0707,
+      "step": 4742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5157322520511168e-05,
+      "loss": 0.9316,
+      "step": 4743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5155406418527165e-05,
+      "loss": 0.948,
+      "step": 4744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5153490058713045e-05,
+      "loss": 0.9459,
+      "step": 4745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5151573441164642e-05,
+      "loss": 1.0276,
+      "step": 4746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5149656565977808e-05,
+      "loss": 0.9048,
+      "step": 4747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5147739433248414e-05,
+      "loss": 0.9529,
+      "step": 4748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5145822043072335e-05,
+      "loss": 0.8752,
+      "step": 4749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5143904395545465e-05,
+      "loss": 0.9768,
+      "step": 4750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5141986490763705e-05,
+      "loss": 1.0994,
+      "step": 4751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.514006832882298e-05,
+      "loss": 0.9408,
+      "step": 4752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5138149909819211e-05,
+      "loss": 0.8795,
+      "step": 4753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5136231233848343e-05,
+      "loss": 0.6657,
+      "step": 4754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5134312301006339e-05,
+      "loss": 0.8841,
+      "step": 4755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5132393111389163e-05,
+      "loss": 1.0574,
+      "step": 4756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5130473665092797e-05,
+      "loss": 0.9174,
+      "step": 4757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5128553962213235e-05,
+      "loss": 0.9737,
+      "step": 4758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5126634002846485e-05,
+      "loss": 0.923,
+      "step": 4759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5124713787088564e-05,
+      "loss": 0.9289,
+      "step": 4760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5122793315035513e-05,
+      "loss": 1.0234,
+      "step": 4761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5120872586783373e-05,
+      "loss": 1.0116,
+      "step": 4762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5118951602428204e-05,
+      "loss": 0.9202,
+      "step": 4763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5117030362066075e-05,
+      "loss": 1.0017,
+      "step": 4764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.511510886579307e-05,
+      "loss": 1.0236,
+      "step": 4765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.511318711370529e-05,
+      "loss": 0.9666,
+      "step": 4766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5111265105898842e-05,
+      "loss": 1.0127,
+      "step": 4767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5109342842469851e-05,
+      "loss": 0.989,
+      "step": 4768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5107420323514448e-05,
+      "loss": 0.9299,
+      "step": 4769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5105497549128786e-05,
+      "loss": 0.8822,
+      "step": 4770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5103574519409026e-05,
+      "loss": 1.0643,
+      "step": 4771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.510165123445134e-05,
+      "loss": 0.9611,
+      "step": 4772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.509972769435191e-05,
+      "loss": 0.9659,
+      "step": 4773
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5097803899206946e-05,
+      "loss": 1.0369,
+      "step": 4774
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5095879849112652e-05,
+      "loss": 0.9044,
+      "step": 4775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5093955544165248e-05,
+      "loss": 0.9831,
+      "step": 4776
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5092030984460988e-05,
+      "loss": 0.9595,
+      "step": 4777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5090106170096108e-05,
+      "loss": 0.9734,
+      "step": 4778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5088181101166876e-05,
+      "loss": 1.0676,
+      "step": 4779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5086255777769569e-05,
+      "loss": 0.9435,
+      "step": 4780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5084330200000472e-05,
+      "loss": 1.0318,
+      "step": 4781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.508240436795589e-05,
+      "loss": 0.941,
+      "step": 4782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5080478281732136e-05,
+      "loss": 0.8738,
+      "step": 4783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5078551941425534e-05,
+      "loss": 1.0441,
+      "step": 4784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5076625347132425e-05,
+      "loss": 1.0135,
+      "step": 4785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5074698498949162e-05,
+      "loss": 1.1015,
+      "step": 4786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.507277139697211e-05,
+      "loss": 0.8705,
+      "step": 4787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5070844041297644e-05,
+      "loss": 0.8699,
+      "step": 4788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5068916432022156e-05,
+      "loss": 1.0135,
+      "step": 4789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5066988569242047e-05,
+      "loss": 1.0445,
+      "step": 4790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5065060453053732e-05,
+      "loss": 1.0215,
+      "step": 4791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5063132083553646e-05,
+      "loss": 0.9519,
+      "step": 4792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5061203460838225e-05,
+      "loss": 0.9906,
+      "step": 4793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.505927458500392e-05,
+      "loss": 0.9729,
+      "step": 4794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.50573454561472e-05,
+      "loss": 0.642,
+      "step": 4795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5055416074364545e-05,
+      "loss": 0.9593,
+      "step": 4796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5053486439752444e-05,
+      "loss": 0.9368,
+      "step": 4797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.50515565524074e-05,
+      "loss": 0.9446,
+      "step": 4798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5049626412425934e-05,
+      "loss": 0.9678,
+      "step": 4799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5047696019904577e-05,
+      "loss": 0.9473,
+      "step": 4800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5045765374939861e-05,
+      "loss": 1.0039,
+      "step": 4801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5043834477628348e-05,
+      "loss": 1.0635,
+      "step": 4802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5041903328066605e-05,
+      "loss": 0.9808,
+      "step": 4803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.503997192635121e-05,
+      "loss": 0.921,
+      "step": 4804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.503804027257876e-05,
+      "loss": 0.9936,
+      "step": 4805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5036108366845855e-05,
+      "loss": 0.8389,
+      "step": 4806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.503417620924911e-05,
+      "loss": 0.9919,
+      "step": 4807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5032243799885163e-05,
+      "loss": 0.9459,
+      "step": 4808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5030311138850653e-05,
+      "loss": 0.9277,
+      "step": 4809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5028378226242236e-05,
+      "loss": 1.0368,
+      "step": 4810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.502644506215658e-05,
+      "loss": 1.0223,
+      "step": 4811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5024511646690365e-05,
+      "loss": 1.0166,
+      "step": 4812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5022577979940285e-05,
+      "loss": 1.0101,
+      "step": 4813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5020644062003047e-05,
+      "loss": 0.9133,
+      "step": 4814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5018709892975368e-05,
+      "loss": 0.9117,
+      "step": 4815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5016775472953977e-05,
+      "loss": 0.992,
+      "step": 4816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5014840802035621e-05,
+      "loss": 1.0788,
+      "step": 4817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5012905880317053e-05,
+      "loss": 0.9595,
+      "step": 4818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5010970707895043e-05,
+      "loss": 1.0316,
+      "step": 4819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.500903528486637e-05,
+      "loss": 0.9235,
+      "step": 4820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5007099611327832e-05,
+      "loss": 1.0112,
+      "step": 4821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5005163687376233e-05,
+      "loss": 0.9597,
+      "step": 4822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5003227513108386e-05,
+      "loss": 0.9389,
+      "step": 4823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.500129108862113e-05,
+      "loss": 0.9131,
+      "step": 4824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4999354414011309e-05,
+      "loss": 0.9442,
+      "step": 4825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4997417489375773e-05,
+      "loss": 1.0549,
+      "step": 4826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4995480314811397e-05,
+      "loss": 0.9538,
+      "step": 4827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4993542890415056e-05,
+      "loss": 0.9019,
+      "step": 4828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4991605216283648e-05,
+      "loss": 0.6597,
+      "step": 4829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4989667292514081e-05,
+      "loss": 1.0229,
+      "step": 4830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4987729119203267e-05,
+      "loss": 1.0768,
+      "step": 4831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4985790696448142e-05,
+      "loss": 0.9862,
+      "step": 4832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4983852024345646e-05,
+      "loss": 1.0132,
+      "step": 4833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4981913102992742e-05,
+      "loss": 0.9751,
+      "step": 4834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4979973932486391e-05,
+      "loss": 0.9206,
+      "step": 4835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4978034512923579e-05,
+      "loss": 1.049,
+      "step": 4836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4976094844401296e-05,
+      "loss": 0.9258,
+      "step": 4837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.497415492701655e-05,
+      "loss": 0.6405,
+      "step": 4838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4972214760866358e-05,
+      "loss": 1.0708,
+      "step": 4839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4970274346047754e-05,
+      "loss": 0.9185,
+      "step": 4840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4968333682657776e-05,
+      "loss": 1.0405,
+      "step": 4841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4966392770793485e-05,
+      "loss": 0.9252,
+      "step": 4842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4964451610551946e-05,
+      "loss": 0.9442,
+      "step": 4843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.496251020203024e-05,
+      "loss": 1.0781,
+      "step": 4844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.496056854532546e-05,
+      "loss": 1.006,
+      "step": 4845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4958626640534713e-05,
+      "loss": 1.1142,
+      "step": 4846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4956684487755115e-05,
+      "loss": 0.9301,
+      "step": 4847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4954742087083795e-05,
+      "loss": 0.9824,
+      "step": 4848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.49527994386179e-05,
+      "loss": 0.8935,
+      "step": 4849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.495085654245458e-05,
+      "loss": 0.8688,
+      "step": 4850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4948913398691003e-05,
+      "loss": 0.9558,
+      "step": 4851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4946970007424353e-05,
+      "loss": 1.0366,
+      "step": 4852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4945026368751815e-05,
+      "loss": 0.9638,
+      "step": 4853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.49430824827706e-05,
+      "loss": 0.97,
+      "step": 4854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4941138349577923e-05,
+      "loss": 0.9396,
+      "step": 4855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4939193969271013e-05,
+      "loss": 0.989,
+      "step": 4856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4937249341947113e-05,
+      "loss": 1.0333,
+      "step": 4857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.493530446770347e-05,
+      "loss": 1.0368,
+      "step": 4858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4933359346637362e-05,
+      "loss": 0.9361,
+      "step": 4859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4931413978846056e-05,
+      "loss": 0.981,
+      "step": 4860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4929468364426851e-05,
+      "loss": 1.0041,
+      "step": 4861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4927522503477047e-05,
+      "loss": 0.9912,
+      "step": 4862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4925576396093959e-05,
+      "loss": 0.9135,
+      "step": 4863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4923630042374916e-05,
+      "loss": 1.0272,
+      "step": 4864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4921683442417258e-05,
+      "loss": 0.9615,
+      "step": 4865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.491973659631834e-05,
+      "loss": 1.0806,
+      "step": 4866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4917789504175521e-05,
+      "loss": 0.9993,
+      "step": 4867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4915842166086185e-05,
+      "loss": 1.0003,
+      "step": 4868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4913894582147718e-05,
+      "loss": 0.9964,
+      "step": 4869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4911946752457519e-05,
+      "loss": 1.1463,
+      "step": 4870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4909998677113007e-05,
+      "loss": 0.9579,
+      "step": 4871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4908050356211607e-05,
+      "loss": 0.9636,
+      "step": 4872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4906101789850758e-05,
+      "loss": 0.9062,
+      "step": 4873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.490415297812791e-05,
+      "loss": 0.8875,
+      "step": 4874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4902203921140523e-05,
+      "loss": 1.0024,
+      "step": 4875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4900254618986082e-05,
+      "loss": 1.0397,
+      "step": 4876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4898305071762064e-05,
+      "loss": 0.9463,
+      "step": 4877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4896355279565974e-05,
+      "loss": 0.8916,
+      "step": 4878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4894405242495324e-05,
+      "loss": 0.9226,
+      "step": 4879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4892454960647639e-05,
+      "loss": 0.882,
+      "step": 4880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4890504434120455e-05,
+      "loss": 1.0357,
+      "step": 4881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.488855366301132e-05,
+      "loss": 0.903,
+      "step": 4882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4886602647417796e-05,
+      "loss": 0.9753,
+      "step": 4883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4884651387437459e-05,
+      "loss": 0.9256,
+      "step": 4884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4882699883167886e-05,
+      "loss": 0.8843,
+      "step": 4885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4880748134706686e-05,
+      "loss": 0.6142,
+      "step": 4886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4878796142151464e-05,
+      "loss": 0.9492,
+      "step": 4887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.487684390559984e-05,
+      "loss": 0.8221,
+      "step": 4888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4874891425149454e-05,
+      "loss": 1.1334,
+      "step": 4889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4872938700897946e-05,
+      "loss": 0.9856,
+      "step": 4890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4870985732942979e-05,
+      "loss": 0.8763,
+      "step": 4891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4869032521382224e-05,
+      "loss": 0.9284,
+      "step": 4892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4867079066313364e-05,
+      "loss": 1.0313,
+      "step": 4893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4865125367834092e-05,
+      "loss": 0.9643,
+      "step": 4894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4863171426042118e-05,
+      "loss": 0.6627,
+      "step": 4895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4861217241035164e-05,
+      "loss": 0.8514,
+      "step": 4896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4859262812910958e-05,
+      "loss": 1.0496,
+      "step": 4897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4857308141767243e-05,
+      "loss": 0.8871,
+      "step": 4898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.485535322770178e-05,
+      "loss": 0.9927,
+      "step": 4899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4853398070812334e-05,
+      "loss": 1.01,
+      "step": 4900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4851442671196688e-05,
+      "loss": 1.0718,
+      "step": 4901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.484948702895263e-05,
+      "loss": 0.8811,
+      "step": 4902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4847531144177972e-05,
+      "loss": 0.9571,
+      "step": 4903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4845575016970524e-05,
+      "loss": 0.8807,
+      "step": 4904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4843618647428118e-05,
+      "loss": 0.974,
+      "step": 4905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4841662035648594e-05,
+      "loss": 0.9168,
+      "step": 4906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4839705181729809e-05,
+      "loss": 0.9926,
+      "step": 4907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4837748085769623e-05,
+      "loss": 0.9362,
+      "step": 4908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4835790747865918e-05,
+      "loss": 0.9024,
+      "step": 4909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4833833168116582e-05,
+      "loss": 1.0819,
+      "step": 4910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4831875346619517e-05,
+      "loss": 1.0357,
+      "step": 4911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4829917283472635e-05,
+      "loss": 0.9292,
+      "step": 4912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4827958978773865e-05,
+      "loss": 0.9577,
+      "step": 4913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4826000432621141e-05,
+      "loss": 0.9957,
+      "step": 4914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4824041645112417e-05,
+      "loss": 0.6842,
+      "step": 4915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4822082616345653e-05,
+      "loss": 1.0551,
+      "step": 4916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.482012334641882e-05,
+      "loss": 1.012,
+      "step": 4917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.481816383542991e-05,
+      "loss": 0.9388,
+      "step": 4918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4816204083476921e-05,
+      "loss": 0.9836,
+      "step": 4919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4814244090657858e-05,
+      "loss": 0.9932,
+      "step": 4920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4812283857070749e-05,
+      "loss": 0.9547,
+      "step": 4921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4810323382813626e-05,
+      "loss": 0.9435,
+      "step": 4922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4808362667984532e-05,
+      "loss": 0.9379,
+      "step": 4923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4806401712681533e-05,
+      "loss": 0.926,
+      "step": 4924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4804440517002693e-05,
+      "loss": 0.93,
+      "step": 4925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4802479081046101e-05,
+      "loss": 0.9546,
+      "step": 4926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4800517404909841e-05,
+      "loss": 1.0064,
+      "step": 4927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4798555488692033e-05,
+      "loss": 0.9247,
+      "step": 4928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4796593332490785e-05,
+      "loss": 0.8349,
+      "step": 4929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4794630936404233e-05,
+      "loss": 1.0056,
+      "step": 4930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4792668300530518e-05,
+      "loss": 0.6506,
+      "step": 4931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4790705424967788e-05,
+      "loss": 0.9884,
+      "step": 4932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4788742309814223e-05,
+      "loss": 1.0472,
+      "step": 4933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4786778955167992e-05,
+      "loss": 0.9943,
+      "step": 4934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.478481536112729e-05,
+      "loss": 0.989,
+      "step": 4935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4782851527790317e-05,
+      "loss": 0.8462,
+      "step": 4936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4780887455255287e-05,
+      "loss": 0.9523,
+      "step": 4937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4778923143620424e-05,
+      "loss": 1.1297,
+      "step": 4938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4776958592983974e-05,
+      "loss": 0.9699,
+      "step": 4939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4774993803444182e-05,
+      "loss": 0.8834,
+      "step": 4940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4773028775099313e-05,
+      "loss": 0.8408,
+      "step": 4941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4771063508047636e-05,
+      "loss": 1.0742,
+      "step": 4942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4769098002387444e-05,
+      "loss": 0.9342,
+      "step": 4943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4767132258217029e-05,
+      "loss": 1.0022,
+      "step": 4944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4765166275634707e-05,
+      "loss": 1.0492,
+      "step": 4945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4763200054738796e-05,
+      "loss": 0.9611,
+      "step": 4946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4761233595627626e-05,
+      "loss": 0.9346,
+      "step": 4947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4759266898399549e-05,
+      "loss": 0.9775,
+      "step": 4948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4757299963152922e-05,
+      "loss": 0.9956,
+      "step": 4949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4755332789986113e-05,
+      "loss": 0.866,
+      "step": 4950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4753365378997509e-05,
+      "loss": 0.963,
+      "step": 4951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4751397730285492e-05,
+      "loss": 0.9805,
+      "step": 4952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4749429843948478e-05,
+      "loss": 0.9822,
+      "step": 4953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4747461720084878e-05,
+      "loss": 1.0443,
+      "step": 4954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4745493358793121e-05,
+      "loss": 0.9502,
+      "step": 4955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4743524760171652e-05,
+      "loss": 0.6555,
+      "step": 4956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.474155592431892e-05,
+      "loss": 1.0455,
+      "step": 4957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4739586851333391e-05,
+      "loss": 0.8657,
+      "step": 4958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4737617541313543e-05,
+      "loss": 1.0716,
+      "step": 4959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4735647994357862e-05,
+      "loss": 0.9961,
+      "step": 4960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.473367821056485e-05,
+      "loss": 0.9883,
+      "step": 4961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4731708190033018e-05,
+      "loss": 1.0145,
+      "step": 4962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.472973793286089e-05,
+      "loss": 0.9552,
+      "step": 4963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4727767439146997e-05,
+      "loss": 0.6137,
+      "step": 4964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4725796708989899e-05,
+      "loss": 0.9901,
+      "step": 4965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4723825742488144e-05,
+      "loss": 1.0006,
+      "step": 4966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4721854539740308e-05,
+      "loss": 0.8825,
+      "step": 4967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4719883100844974e-05,
+      "loss": 0.6559,
+      "step": 4968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4717911425900733e-05,
+      "loss": 1.0493,
+      "step": 4969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4715939515006194e-05,
+      "loss": 0.9083,
+      "step": 4970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4713967368259981e-05,
+      "loss": 0.9116,
+      "step": 4971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.471199498576071e-05,
+      "loss": 1.0029,
+      "step": 4972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4710022367607039e-05,
+      "loss": 0.9632,
+      "step": 4973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4708049513897611e-05,
+      "loss": 0.9502,
+      "step": 4974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.47060764247311e-05,
+      "loss": 0.968,
+      "step": 4975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4704103100206173e-05,
+      "loss": 0.6727,
+      "step": 4976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4702129540421529e-05,
+      "loss": 0.9953,
+      "step": 4977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4700155745475864e-05,
+      "loss": 0.9879,
+      "step": 4978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4698181715467886e-05,
+      "loss": 0.9864,
+      "step": 4979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4696207450496329e-05,
+      "loss": 0.9506,
+      "step": 4980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4694232950659925e-05,
+      "loss": 0.9693,
+      "step": 4981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4692258216057422e-05,
+      "loss": 0.8412,
+      "step": 4982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.469028324678758e-05,
+      "loss": 1.021,
+      "step": 4983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4688308042949166e-05,
+      "loss": 0.8402,
+      "step": 4984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4686332604640972e-05,
+      "loss": 0.9513,
+      "step": 4985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4684356931961782e-05,
+      "loss": 1.0869,
+      "step": 4986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4682381025010413e-05,
+      "loss": 0.9895,
+      "step": 4987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4680404883885677e-05,
+      "loss": 1.0345,
+      "step": 4988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4678428508686404e-05,
+      "loss": 0.6412,
+      "step": 4989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.467645189951144e-05,
+      "loss": 0.955,
+      "step": 4990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4674475056459633e-05,
+      "loss": 0.9727,
+      "step": 4991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4672497979629852e-05,
+      "loss": 1.0,
+      "step": 4992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4670520669120974e-05,
+      "loss": 0.9566,
+      "step": 4993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4668543125031881e-05,
+      "loss": 0.9872,
+      "step": 4994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4666565347461479e-05,
+      "loss": 0.9637,
+      "step": 4995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.466458733650868e-05,
+      "loss": 0.9105,
+      "step": 4996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4662609092272409e-05,
+      "loss": 0.8905,
+      "step": 4997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4660630614851596e-05,
+      "loss": 0.8544,
+      "step": 4998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4658651904345189e-05,
+      "loss": 0.6312,
+      "step": 4999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4656672960852152e-05,
+      "loss": 0.961,
+      "step": 5000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4654693784471446e-05,
+      "loss": 1.0359,
+      "step": 5001
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4652714375302062e-05,
+      "loss": 0.9955,
+      "step": 5002
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4650734733442988e-05,
+      "loss": 0.9362,
+      "step": 5003
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4648754858993229e-05,
+      "loss": 1.0134,
+      "step": 5004
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4646774752051803e-05,
+      "loss": 0.9654,
+      "step": 5005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4644794412717738e-05,
+      "loss": 0.9584,
+      "step": 5006
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4642813841090073e-05,
+      "loss": 0.9445,
+      "step": 5007
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4640833037267861e-05,
+      "loss": 0.9416,
+      "step": 5008
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4638852001350167e-05,
+      "loss": 0.9406,
+      "step": 5009
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4636870733436064e-05,
+      "loss": 1.0426,
+      "step": 5010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4634889233624634e-05,
+      "loss": 1.0026,
+      "step": 5011
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4632907502014982e-05,
+      "loss": 1.0869,
+      "step": 5012
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4630925538706213e-05,
+      "loss": 1.0179,
+      "step": 5013
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.462894334379745e-05,
+      "loss": 1.0452,
+      "step": 5014
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4626960917387827e-05,
+      "loss": 0.9697,
+      "step": 5015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4624978259576485e-05,
+      "loss": 0.997,
+      "step": 5016
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4622995370462582e-05,
+      "loss": 1.0387,
+      "step": 5017
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4621012250145286e-05,
+      "loss": 0.9739,
+      "step": 5018
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4619028898723775e-05,
+      "loss": 0.6944,
+      "step": 5019
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4617045316297238e-05,
+      "loss": 0.8743,
+      "step": 5020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.461506150296488e-05,
+      "loss": 0.9645,
+      "step": 5021
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4613077458825916e-05,
+      "loss": 1.0009,
+      "step": 5022
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4611093183979563e-05,
+      "loss": 0.9243,
+      "step": 5023
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4609108678525067e-05,
+      "loss": 1.0247,
+      "step": 5024
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4607123942561674e-05,
+      "loss": 1.0524,
+      "step": 5025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4605138976188637e-05,
+      "loss": 0.9478,
+      "step": 5026
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4603153779505237e-05,
+      "loss": 0.974,
+      "step": 5027
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4601168352610754e-05,
+      "loss": 0.9781,
+      "step": 5028
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.459918269560448e-05,
+      "loss": 1.0044,
+      "step": 5029
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.459719680858572e-05,
+      "loss": 0.8938,
+      "step": 5030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4595210691653794e-05,
+      "loss": 1.0738,
+      "step": 5031
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4593224344908032e-05,
+      "loss": 0.987,
+      "step": 5032
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4591237768447773e-05,
+      "loss": 0.9652,
+      "step": 5033
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4589250962372368e-05,
+      "loss": 0.8123,
+      "step": 5034
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4587263926781181e-05,
+      "loss": 0.9693,
+      "step": 5035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4585276661773584e-05,
+      "loss": 0.8753,
+      "step": 5036
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.458328916744897e-05,
+      "loss": 1.009,
+      "step": 5037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.458130144390673e-05,
+      "loss": 1.1306,
+      "step": 5038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4579313491246278e-05,
+      "loss": 0.9596,
+      "step": 5039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4577325309567032e-05,
+      "loss": 0.9835,
+      "step": 5040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4575336898968424e-05,
+      "loss": 0.9845,
+      "step": 5041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4573348259549897e-05,
+      "loss": 0.9635,
+      "step": 5042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4571359391410911e-05,
+      "loss": 0.8981,
+      "step": 5043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4569370294650929e-05,
+      "loss": 0.9025,
+      "step": 5044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4567380969369427e-05,
+      "loss": 0.9804,
+      "step": 5045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4565391415665894e-05,
+      "loss": 0.9558,
+      "step": 5046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4563401633639838e-05,
+      "loss": 0.9678,
+      "step": 5047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4561411623390765e-05,
+      "loss": 0.8524,
+      "step": 5048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4559421385018198e-05,
+      "loss": 0.9614,
+      "step": 5049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4557430918621676e-05,
+      "loss": 0.9586,
+      "step": 5050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.455544022430074e-05,
+      "loss": 0.9975,
+      "step": 5051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4553449302154954e-05,
+      "loss": 0.9746,
+      "step": 5052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4551458152283882e-05,
+      "loss": 0.971,
+      "step": 5053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4549466774787109e-05,
+      "loss": 0.9316,
+      "step": 5054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4547475169764223e-05,
+      "loss": 0.6566,
+      "step": 5055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.454548333731483e-05,
+      "loss": 0.902,
+      "step": 5056
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4543491277538544e-05,
+      "loss": 1.0004,
+      "step": 5057
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4541498990534988e-05,
+      "loss": 0.9913,
+      "step": 5058
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4539506476403805e-05,
+      "loss": 0.8646,
+      "step": 5059
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4537513735244642e-05,
+      "loss": 1.0221,
+      "step": 5060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4535520767157156e-05,
+      "loss": 0.8947,
+      "step": 5061
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4533527572241024e-05,
+      "loss": 1.0379,
+      "step": 5062
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4531534150595923e-05,
+      "loss": 1.0197,
+      "step": 5063
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.452954050232155e-05,
+      "loss": 0.9122,
+      "step": 5064
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4527546627517612e-05,
+      "loss": 0.7847,
+      "step": 5065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4525552526283826e-05,
+      "loss": 0.949,
+      "step": 5066
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4523558198719916e-05,
+      "loss": 0.9965,
+      "step": 5067
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4521563644925625e-05,
+      "loss": 1.0235,
+      "step": 5068
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4519568865000701e-05,
+      "loss": 0.6757,
+      "step": 5069
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4517573859044909e-05,
+      "loss": 1.0094,
+      "step": 5070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4515578627158024e-05,
+      "loss": 0.6438,
+      "step": 5071
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4513583169439827e-05,
+      "loss": 0.9437,
+      "step": 5072
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4511587485990113e-05,
+      "loss": 1.0154,
+      "step": 5073
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4509591576908692e-05,
+      "loss": 0.9209,
+      "step": 5074
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4507595442295385e-05,
+      "loss": 1.0158,
+      "step": 5075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4505599082250017e-05,
+      "loss": 0.96,
+      "step": 5076
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4503602496872434e-05,
+      "loss": 1.0414,
+      "step": 5077
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4501605686262481e-05,
+      "loss": 1.0218,
+      "step": 5078
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4499608650520029e-05,
+      "loss": 0.9673,
+      "step": 5079
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.449761138974495e-05,
+      "loss": 0.9455,
+      "step": 5080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.449561390403713e-05,
+      "loss": 1.044,
+      "step": 5081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4493616193496469e-05,
+      "loss": 1.0636,
+      "step": 5082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4491618258222872e-05,
+      "loss": 0.8845,
+      "step": 5083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4489620098316262e-05,
+      "loss": 0.9731,
+      "step": 5084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4487621713876567e-05,
+      "loss": 0.8948,
+      "step": 5085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4485623105003732e-05,
+      "loss": 1.0618,
+      "step": 5086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4483624271797711e-05,
+      "loss": 0.9452,
+      "step": 5087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4481625214358466e-05,
+      "loss": 0.9423,
+      "step": 5088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4479625932785974e-05,
+      "loss": 1.0443,
+      "step": 5089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4477626427180223e-05,
+      "loss": 0.9642,
+      "step": 5090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4475626697641211e-05,
+      "loss": 1.0328,
+      "step": 5091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4473626744268954e-05,
+      "loss": 0.9839,
+      "step": 5092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.447162656716346e-05,
+      "loss": 1.0766,
+      "step": 5093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4469626166424773e-05,
+      "loss": 0.9417,
+      "step": 5094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4467625542152928e-05,
+      "loss": 0.9967,
+      "step": 5095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4465624694447983e-05,
+      "loss": 0.9286,
+      "step": 5096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4463623623410006e-05,
+      "loss": 1.0179,
+      "step": 5097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4461622329139069e-05,
+      "loss": 0.9512,
+      "step": 5098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4459620811735263e-05,
+      "loss": 1.0237,
+      "step": 5099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4457619071298684e-05,
+      "loss": 0.9674,
+      "step": 5100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4455617107929447e-05,
+      "loss": 0.922,
+      "step": 5101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4453614921727668e-05,
+      "loss": 0.9818,
+      "step": 5102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4451612512793486e-05,
+      "loss": 0.9529,
+      "step": 5103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4449609881227038e-05,
+      "loss": 0.9302,
+      "step": 5104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4447607027128481e-05,
+      "loss": 1.1155,
+      "step": 5105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4445603950597984e-05,
+      "loss": 0.9254,
+      "step": 5106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.444360065173572e-05,
+      "loss": 0.9854,
+      "step": 5107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.444159713064188e-05,
+      "loss": 0.671,
+      "step": 5108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4439593387416662e-05,
+      "loss": 1.0282,
+      "step": 5109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4437589422160276e-05,
+      "loss": 0.9309,
+      "step": 5110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4435585234972947e-05,
+      "loss": 0.9526,
+      "step": 5111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.44335808259549e-05,
+      "loss": 0.9513,
+      "step": 5112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4431576195206387e-05,
+      "loss": 0.9912,
+      "step": 5113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.442957134282766e-05,
+      "loss": 0.8794,
+      "step": 5114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.442756626891898e-05,
+      "loss": 1.0098,
+      "step": 5115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4425560973580632e-05,
+      "loss": 1.0593,
+      "step": 5116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4423555456912896e-05,
+      "loss": 1.1242,
+      "step": 5117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4421549719016082e-05,
+      "loss": 0.9601,
+      "step": 5118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4419543759990489e-05,
+      "loss": 0.9063,
+      "step": 5119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4417537579936439e-05,
+      "loss": 0.9253,
+      "step": 5120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4415531178954271e-05,
+      "loss": 0.9633,
+      "step": 5121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4413524557144328e-05,
+      "loss": 1.0185,
+      "step": 5122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4411517714606958e-05,
+      "loss": 1.0876,
+      "step": 5123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4409510651442534e-05,
+      "loss": 0.9619,
+      "step": 5124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4407503367751423e-05,
+      "loss": 1.1292,
+      "step": 5125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.440549586363402e-05,
+      "loss": 0.9453,
+      "step": 5126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4403488139190723e-05,
+      "loss": 1.0424,
+      "step": 5127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4401480194521938e-05,
+      "loss": 0.6252,
+      "step": 5128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.439947202972809e-05,
+      "loss": 0.6786,
+      "step": 5129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4397463644909605e-05,
+      "loss": 0.9881,
+      "step": 5130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4395455040166927e-05,
+      "loss": 1.0158,
+      "step": 5131
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4393446215600513e-05,
+      "loss": 0.968,
+      "step": 5132
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4391437171310827e-05,
+      "loss": 0.878,
+      "step": 5133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4389427907398343e-05,
+      "loss": 0.9933,
+      "step": 5134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4387418423963546e-05,
+      "loss": 1.122,
+      "step": 5135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4385408721106936e-05,
+      "loss": 1.0514,
+      "step": 5136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4383398798929018e-05,
+      "loss": 0.9502,
+      "step": 5137
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4381388657530315e-05,
+      "loss": 1.0075,
+      "step": 5138
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4379378297011359e-05,
+      "loss": 1.0061,
+      "step": 5139
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4377367717472685e-05,
+      "loss": 1.0793,
+      "step": 5140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4375356919014852e-05,
+      "loss": 0.9442,
+      "step": 5141
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4373345901738418e-05,
+      "loss": 0.8674,
+      "step": 5142
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4371334665743962e-05,
+      "loss": 1.0042,
+      "step": 5143
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4369323211132065e-05,
+      "loss": 0.8504,
+      "step": 5144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4367311538003323e-05,
+      "loss": 0.9145,
+      "step": 5145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4365299646458346e-05,
+      "loss": 0.9091,
+      "step": 5146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.436328753659775e-05,
+      "loss": 0.7021,
+      "step": 5147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4361275208522165e-05,
+      "loss": 1.0013,
+      "step": 5148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.435926266233223e-05,
+      "loss": 1.0023,
+      "step": 5149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4357249898128593e-05,
+      "loss": 0.9695,
+      "step": 5150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4355236916011924e-05,
+      "loss": 0.9623,
+      "step": 5151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4353223716082883e-05,
+      "loss": 1.0021,
+      "step": 5152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4351210298442164e-05,
+      "loss": 1.0145,
+      "step": 5153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4349196663190457e-05,
+      "loss": 0.9721,
+      "step": 5154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4347182810428465e-05,
+      "loss": 1.0945,
+      "step": 5155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.434516874025691e-05,
+      "loss": 1.0335,
+      "step": 5156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4343154452776513e-05,
+      "loss": 1.0413,
+      "step": 5157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4341139948088017e-05,
+      "loss": 1.0552,
+      "step": 5158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4339125226292163e-05,
+      "loss": 0.9917,
+      "step": 5159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4337110287489719e-05,
+      "loss": 0.974,
+      "step": 5160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4335095131781451e-05,
+      "loss": 1.0421,
+      "step": 5161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4333079759268138e-05,
+      "loss": 0.6621,
+      "step": 5162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4331064170050577e-05,
+      "loss": 1.0153,
+      "step": 5163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4329048364229565e-05,
+      "loss": 0.9409,
+      "step": 5164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4327032341905924e-05,
+      "loss": 0.9318,
+      "step": 5165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4325016103180471e-05,
+      "loss": 0.9581,
+      "step": 5166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4322999648154042e-05,
+      "loss": 0.977,
+      "step": 5167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4320982976927487e-05,
+      "loss": 1.0186,
+      "step": 5168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4318966089601662e-05,
+      "loss": 0.9402,
+      "step": 5169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4316948986277432e-05,
+      "loss": 0.8598,
+      "step": 5170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.431493166705568e-05,
+      "loss": 0.9105,
+      "step": 5171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.431291413203729e-05,
+      "loss": 0.955,
+      "step": 5172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4310896381323165e-05,
+      "loss": 1.0012,
+      "step": 5173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4308878415014215e-05,
+      "loss": 0.9386,
+      "step": 5174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4306860233211366e-05,
+      "loss": 0.9619,
+      "step": 5175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4304841836015547e-05,
+      "loss": 0.9342,
+      "step": 5176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4302823223527698e-05,
+      "loss": 1.0293,
+      "step": 5177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4300804395848777e-05,
+      "loss": 0.8082,
+      "step": 5178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.429878535307975e-05,
+      "loss": 0.8896,
+      "step": 5179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4296766095321588e-05,
+      "loss": 0.9879,
+      "step": 5180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4294746622675285e-05,
+      "loss": 0.6368,
+      "step": 5181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4292726935241827e-05,
+      "loss": 0.97,
+      "step": 5182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4290707033122228e-05,
+      "loss": 0.97,
+      "step": 5183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.428868691641751e-05,
+      "loss": 0.9684,
+      "step": 5184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4286666585228697e-05,
+      "loss": 0.9565,
+      "step": 5185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.428464603965683e-05,
+      "loss": 0.9862,
+      "step": 5186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4282625279802961e-05,
+      "loss": 0.953,
+      "step": 5187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4280604305768153e-05,
+      "loss": 0.9765,
+      "step": 5188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4278583117653474e-05,
+      "loss": 1.0123,
+      "step": 5189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4276561715560008e-05,
+      "loss": 1.0225,
+      "step": 5190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4274540099588852e-05,
+      "loss": 0.9892,
+      "step": 5191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4272518269841104e-05,
+      "loss": 0.9326,
+      "step": 5192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4270496226417888e-05,
+      "loss": 0.6706,
+      "step": 5193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4268473969420318e-05,
+      "loss": 1.0242,
+      "step": 5194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4266451498949539e-05,
+      "loss": 0.6628,
+      "step": 5195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4264428815106695e-05,
+      "loss": 1.0501,
+      "step": 5196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4262405917992946e-05,
+      "loss": 0.8559,
+      "step": 5197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4260382807709457e-05,
+      "loss": 0.9036,
+      "step": 5198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4258359484357405e-05,
+      "loss": 0.9935,
+      "step": 5199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4256335948037987e-05,
+      "loss": 0.9466,
+      "step": 5200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.42543121988524e-05,
+      "loss": 0.943,
+      "step": 5201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4252288236901852e-05,
+      "loss": 1.0058,
+      "step": 5202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4250264062287569e-05,
+      "loss": 0.9937,
+      "step": 5203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.424823967511078e-05,
+      "loss": 0.8735,
+      "step": 5204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.424621507547273e-05,
+      "loss": 0.9198,
+      "step": 5205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.424419026347467e-05,
+      "loss": 0.8932,
+      "step": 5206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4242165239217864e-05,
+      "loss": 1.0221,
+      "step": 5207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4240140002803591e-05,
+      "loss": 0.983,
+      "step": 5208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4238114554333131e-05,
+      "loss": 0.9803,
+      "step": 5209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4236088893907784e-05,
+      "loss": 0.9091,
+      "step": 5210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4234063021628854e-05,
+      "loss": 1.0969,
+      "step": 5211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.423203693759766e-05,
+      "loss": 0.9824,
+      "step": 5212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4230010641915528e-05,
+      "loss": 1.0086,
+      "step": 5213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4227984134683798e-05,
+      "loss": 0.9065,
+      "step": 5214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4225957416003815e-05,
+      "loss": 0.9255,
+      "step": 5215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4223930485976944e-05,
+      "loss": 0.9702,
+      "step": 5216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.422190334470455e-05,
+      "loss": 1.0286,
+      "step": 5217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.421987599228802e-05,
+      "loss": 0.9104,
+      "step": 5218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4217848428828738e-05,
+      "loss": 0.9622,
+      "step": 5219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.421582065442811e-05,
+      "loss": 0.6676,
+      "step": 5220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4213792669187546e-05,
+      "loss": 1.089,
+      "step": 5221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4211764473208471e-05,
+      "loss": 1.0135,
+      "step": 5222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4209736066592316e-05,
+      "loss": 0.8724,
+      "step": 5223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4207707449440527e-05,
+      "loss": 0.9642,
+      "step": 5224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4205678621854558e-05,
+      "loss": 0.9208,
+      "step": 5225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.420364958393587e-05,
+      "loss": 0.9447,
+      "step": 5226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4201620335785944e-05,
+      "loss": 1.0095,
+      "step": 5227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4199590877506263e-05,
+      "loss": 0.6694,
+      "step": 5228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4197561209198325e-05,
+      "loss": 0.941,
+      "step": 5229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4195531330963634e-05,
+      "loss": 0.9481,
+      "step": 5230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.419350124290371e-05,
+      "loss": 0.8695,
+      "step": 5231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4191470945120083e-05,
+      "loss": 0.9225,
+      "step": 5232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4189440437714285e-05,
+      "loss": 0.9228,
+      "step": 5233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.418740972078787e-05,
+      "loss": 1.0411,
+      "step": 5234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4185378794442399e-05,
+      "loss": 1.0444,
+      "step": 5235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4183347658779436e-05,
+      "loss": 1.0001,
+      "step": 5236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4181316313900566e-05,
+      "loss": 0.9976,
+      "step": 5237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4179284759907378e-05,
+      "loss": 0.9259,
+      "step": 5238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4177252996901474e-05,
+      "loss": 1.0012,
+      "step": 5239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4175221024984467e-05,
+      "loss": 1.0761,
+      "step": 5240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4173188844257975e-05,
+      "loss": 0.8453,
+      "step": 5241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4171156454823632e-05,
+      "loss": 0.9081,
+      "step": 5242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4169123856783082e-05,
+      "loss": 0.6628,
+      "step": 5243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4167091050237982e-05,
+      "loss": 0.9819,
+      "step": 5244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4165058035289989e-05,
+      "loss": 0.9912,
+      "step": 5245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.416302481204078e-05,
+      "loss": 0.8706,
+      "step": 5246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4160991380592045e-05,
+      "loss": 0.904,
+      "step": 5247
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4158957741045471e-05,
+      "loss": 1.0435,
+      "step": 5248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4156923893502768e-05,
+      "loss": 0.9929,
+      "step": 5249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4154889838065652e-05,
+      "loss": 0.9287,
+      "step": 5250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4152855574835847e-05,
+      "loss": 0.9626,
+      "step": 5251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4150821103915094e-05,
+      "loss": 0.9627,
+      "step": 5252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4148786425405134e-05,
+      "loss": 0.9501,
+      "step": 5253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.414675153940773e-05,
+      "loss": 1.0105,
+      "step": 5254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.414471644602465e-05,
+      "loss": 1.0737,
+      "step": 5255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4142681145357666e-05,
+      "loss": 0.6547,
+      "step": 5256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4140645637508575e-05,
+      "loss": 0.9728,
+      "step": 5257
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4138609922579169e-05,
+      "loss": 0.9345,
+      "step": 5258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4136574000671262e-05,
+      "loss": 0.637,
+      "step": 5259
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.413453787188667e-05,
+      "loss": 1.0241,
+      "step": 5260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4132501536327225e-05,
+      "loss": 0.937,
+      "step": 5261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4130464994094768e-05,
+      "loss": 1.0528,
+      "step": 5262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.412842824529115e-05,
+      "loss": 0.9883,
+      "step": 5263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4126391290018231e-05,
+      "loss": 0.8978,
+      "step": 5264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4124354128377881e-05,
+      "loss": 1.0404,
+      "step": 5265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4122316760471983e-05,
+      "loss": 1.0015,
+      "step": 5266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4120279186402433e-05,
+      "loss": 0.8302,
+      "step": 5267
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4118241406271125e-05,
+      "loss": 1.0148,
+      "step": 5268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4116203420179979e-05,
+      "loss": 0.9777,
+      "step": 5269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4114165228230916e-05,
+      "loss": 1.0206,
+      "step": 5270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4112126830525865e-05,
+      "loss": 0.9667,
+      "step": 5271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4110088227166779e-05,
+      "loss": 0.985,
+      "step": 5272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4108049418255601e-05,
+      "loss": 0.6524,
+      "step": 5273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4106010403894299e-05,
+      "loss": 1.0056,
+      "step": 5274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4103971184184853e-05,
+      "loss": 0.9453,
+      "step": 5275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4101931759229239e-05,
+      "loss": 0.9939,
+      "step": 5276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4099892129129456e-05,
+      "loss": 0.9155,
+      "step": 5277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4097852293987509e-05,
+      "loss": 0.8718,
+      "step": 5278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4095812253905415e-05,
+      "loss": 1.0227,
+      "step": 5279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4093772008985197e-05,
+      "loss": 1.0179,
+      "step": 5280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4091731559328892e-05,
+      "loss": 0.9489,
+      "step": 5281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4089690905038547e-05,
+      "loss": 0.9502,
+      "step": 5282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4087650046216217e-05,
+      "loss": 1.0306,
+      "step": 5283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.408560898296397e-05,
+      "loss": 0.8934,
+      "step": 5284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4083567715383881e-05,
+      "loss": 0.9163,
+      "step": 5285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4081526243578036e-05,
+      "loss": 0.989,
+      "step": 5286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4079484567648538e-05,
+      "loss": 1.0304,
+      "step": 5287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.407744268769749e-05,
+      "loss": 0.856,
+      "step": 5288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4075400603827007e-05,
+      "loss": 0.9845,
+      "step": 5289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4073358316139223e-05,
+      "loss": 0.9497,
+      "step": 5290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4071315824736273e-05,
+      "loss": 1.0359,
+      "step": 5291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.406927312972031e-05,
+      "loss": 0.903,
+      "step": 5292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.406723023119348e-05,
+      "loss": 0.9087,
+      "step": 5293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4065187129257965e-05,
+      "loss": 0.9463,
+      "step": 5294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4063143824015937e-05,
+      "loss": 0.9392,
+      "step": 5295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4061100315569587e-05,
+      "loss": 1.0252,
+      "step": 5296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4059056604021114e-05,
+      "loss": 1.0186,
+      "step": 5297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4057012689472727e-05,
+      "loss": 0.6957,
+      "step": 5298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4054968572026644e-05,
+      "loss": 0.7283,
+      "step": 5299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4052924251785097e-05,
+      "loss": 0.9337,
+      "step": 5300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4050879728850327e-05,
+      "loss": 0.9316,
+      "step": 5301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4048835003324576e-05,
+      "loss": 0.8863,
+      "step": 5302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4046790075310113e-05,
+      "loss": 0.9282,
+      "step": 5303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4044744944909205e-05,
+      "loss": 0.9738,
+      "step": 5304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4042699612224131e-05,
+      "loss": 0.9413,
+      "step": 5305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4040654077357183e-05,
+      "loss": 0.9438,
+      "step": 5306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.403860834041066e-05,
+      "loss": 0.8989,
+      "step": 5307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4036562401486873e-05,
+      "loss": 0.9592,
+      "step": 5308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4034516260688142e-05,
+      "loss": 0.8656,
+      "step": 5309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4032469918116803e-05,
+      "loss": 1.0002,
+      "step": 5310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4030423373875186e-05,
+      "loss": 0.9534,
+      "step": 5311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4028376628065654e-05,
+      "loss": 1.0444,
+      "step": 5312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4026329680790559e-05,
+      "loss": 0.9859,
+      "step": 5313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.402428253215228e-05,
+      "loss": 0.9467,
+      "step": 5314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.402223518225319e-05,
+      "loss": 0.9176,
+      "step": 5315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4020187631195686e-05,
+      "loss": 0.9235,
+      "step": 5316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4018139879082169e-05,
+      "loss": 0.8701,
+      "step": 5317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4016091926015043e-05,
+      "loss": 0.9688,
+      "step": 5318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4014043772096741e-05,
+      "loss": 0.9856,
+      "step": 5319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4011995417429685e-05,
+      "loss": 1.0345,
+      "step": 5320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4009946862116324e-05,
+      "loss": 0.652,
+      "step": 5321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.40078981062591e-05,
+      "loss": 0.8792,
+      "step": 5322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4005849149960486e-05,
+      "loss": 0.902,
+      "step": 5323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4003799993322944e-05,
+      "loss": 1.0248,
+      "step": 5324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4001750636448961e-05,
+      "loss": 0.9174,
+      "step": 5325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3999701079441028e-05,
+      "loss": 1.0513,
+      "step": 5326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3997651322401644e-05,
+      "loss": 0.6453,
+      "step": 5327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3995601365433322e-05,
+      "loss": 0.9213,
+      "step": 5328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3993551208638584e-05,
+      "loss": 0.9863,
+      "step": 5329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3991500852119964e-05,
+      "loss": 1.0029,
+      "step": 5330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.398945029598e-05,
+      "loss": 1.0111,
+      "step": 5331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3987399540321245e-05,
+      "loss": 1.0222,
+      "step": 5332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3985348585246257e-05,
+      "loss": 0.9721,
+      "step": 5333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3983297430857614e-05,
+      "loss": 1.0452,
+      "step": 5334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3981246077257892e-05,
+      "loss": 0.9974,
+      "step": 5335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3979194524549688e-05,
+      "loss": 0.9318,
+      "step": 5336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3977142772835599e-05,
+      "loss": 1.0116,
+      "step": 5337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3975090822218238e-05,
+      "loss": 0.9472,
+      "step": 5338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3973038672800225e-05,
+      "loss": 0.9036,
+      "step": 5339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.397098632468419e-05,
+      "loss": 0.9946,
+      "step": 5340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.396893377797278e-05,
+      "loss": 1.0622,
+      "step": 5341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3966881032768642e-05,
+      "loss": 1.0114,
+      "step": 5342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3964828089174438e-05,
+      "loss": 0.9203,
+      "step": 5343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3962774947292839e-05,
+      "loss": 0.9077,
+      "step": 5344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3960721607226525e-05,
+      "loss": 0.9483,
+      "step": 5345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3958668069078188e-05,
+      "loss": 0.985,
+      "step": 5346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3956614332950525e-05,
+      "loss": 0.9583,
+      "step": 5347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3954560398946253e-05,
+      "loss": 1.0282,
+      "step": 5348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3952506267168088e-05,
+      "loss": 0.9535,
+      "step": 5349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.395045193771876e-05,
+      "loss": 0.8662,
+      "step": 5350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3948397410701016e-05,
+      "loss": 0.6533,
+      "step": 5351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3946342686217598e-05,
+      "loss": 0.6655,
+      "step": 5352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3944287764371269e-05,
+      "loss": 0.8082,
+      "step": 5353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3942232645264801e-05,
+      "loss": 1.0144,
+      "step": 5354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3940177329000971e-05,
+      "loss": 0.9356,
+      "step": 5355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3938121815682566e-05,
+      "loss": 0.9938,
+      "step": 5356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3936066105412394e-05,
+      "loss": 0.8559,
+      "step": 5357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3934010198293257e-05,
+      "loss": 1.1212,
+      "step": 5358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3931954094427979e-05,
+      "loss": 1.0887,
+      "step": 5359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3929897793919384e-05,
+      "loss": 0.9763,
+      "step": 5360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3927841296870318e-05,
+      "loss": 0.8863,
+      "step": 5361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.392578460338362e-05,
+      "loss": 0.8269,
+      "step": 5362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3923727713562157e-05,
+      "loss": 0.9196,
+      "step": 5363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3921670627508794e-05,
+      "loss": 0.9557,
+      "step": 5364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3919613345326409e-05,
+      "loss": 0.9393,
+      "step": 5365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.391755586711789e-05,
+      "loss": 0.9441,
+      "step": 5366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3915498192986134e-05,
+      "loss": 0.9019,
+      "step": 5367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.391344032303405e-05,
+      "loss": 0.8606,
+      "step": 5368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3911382257364557e-05,
+      "loss": 1.0414,
+      "step": 5369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3909323996080582e-05,
+      "loss": 0.9718,
+      "step": 5370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3907265539285055e-05,
+      "loss": 0.8836,
+      "step": 5371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.390520688708093e-05,
+      "loss": 0.9013,
+      "step": 5372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3903148039571162e-05,
+      "loss": 0.6223,
+      "step": 5373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3901088996858719e-05,
+      "loss": 0.9029,
+      "step": 5374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.389902975904657e-05,
+      "loss": 0.9971,
+      "step": 5375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3896970326237708e-05,
+      "loss": 0.9313,
+      "step": 5376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3894910698535124e-05,
+      "loss": 0.9516,
+      "step": 5377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3892850876041829e-05,
+      "loss": 0.9466,
+      "step": 5378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3890790858860833e-05,
+      "loss": 0.6634,
+      "step": 5379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3888730647095159e-05,
+      "loss": 0.9795,
+      "step": 5380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3886670240847849e-05,
+      "loss": 0.9141,
+      "step": 5381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3884609640221937e-05,
+      "loss": 1.0652,
+      "step": 5382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3882548845320486e-05,
+      "loss": 0.9553,
+      "step": 5383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3880487856246556e-05,
+      "loss": 1.0059,
+      "step": 5384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3878426673103222e-05,
+      "loss": 0.8834,
+      "step": 5385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3876365295993565e-05,
+      "loss": 0.8141,
+      "step": 5386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3874303725020675e-05,
+      "loss": 1.0073,
+      "step": 5387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3872241960287662e-05,
+      "loss": 0.856,
+      "step": 5388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3870180001897633e-05,
+      "loss": 0.8634,
+      "step": 5389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.386811784995371e-05,
+      "loss": 0.9286,
+      "step": 5390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3866055504559029e-05,
+      "loss": 1.0403,
+      "step": 5391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3863992965816722e-05,
+      "loss": 0.9681,
+      "step": 5392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.386193023382995e-05,
+      "loss": 1.0226,
+      "step": 5393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3859867308701866e-05,
+      "loss": 0.8938,
+      "step": 5394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3857804190535646e-05,
+      "loss": 0.9074,
+      "step": 5395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3855740879434468e-05,
+      "loss": 0.9576,
+      "step": 5396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3853677375501518e-05,
+      "loss": 0.9738,
+      "step": 5397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3851613678839999e-05,
+      "loss": 0.6295,
+      "step": 5398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3849549789553116e-05,
+      "loss": 0.8785,
+      "step": 5399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3847485707744093e-05,
+      "loss": 0.8402,
+      "step": 5400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3845421433516154e-05,
+      "loss": 1.0168,
+      "step": 5401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3843356966972535e-05,
+      "loss": 0.8905,
+      "step": 5402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3841292308216488e-05,
+      "loss": 0.9735,
+      "step": 5403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3839227457351267e-05,
+      "loss": 0.9294,
+      "step": 5404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.383716241448014e-05,
+      "loss": 0.6671,
+      "step": 5405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.383509717970638e-05,
+      "loss": 1.0681,
+      "step": 5406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3833031753133276e-05,
+      "loss": 1.0246,
+      "step": 5407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3830966134864126e-05,
+      "loss": 0.94,
+      "step": 5408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3828900325002226e-05,
+      "loss": 0.9668,
+      "step": 5409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.915,
+      "step": 5410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3824768130913466e-05,
+      "loss": 0.6577,
+      "step": 5411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.382270174689326e-05,
+      "loss": 1.0053,
+      "step": 5412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3820635171693621e-05,
+      "loss": 0.9527,
+      "step": 5413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3818568405417906e-05,
+      "loss": 0.9296,
+      "step": 5414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.381650144816948e-05,
+      "loss": 0.9764,
+      "step": 5415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3814434300051709e-05,
+      "loss": 1.0081,
+      "step": 5416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3812366961167978e-05,
+      "loss": 1.0077,
+      "step": 5417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3810299431621674e-05,
+      "loss": 0.9647,
+      "step": 5418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.38082317115162e-05,
+      "loss": 0.9798,
+      "step": 5419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3806163800954969e-05,
+      "loss": 0.8582,
+      "step": 5420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3804095700041396e-05,
+      "loss": 1.0879,
+      "step": 5421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.380202740887891e-05,
+      "loss": 0.9714,
+      "step": 5422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3799958927570956e-05,
+      "loss": 0.8358,
+      "step": 5423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3797890256220971e-05,
+      "loss": 0.8783,
+      "step": 5424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3795821394932422e-05,
+      "loss": 1.0353,
+      "step": 5425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3793752343808775e-05,
+      "loss": 0.9084,
+      "step": 5426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3791683102953501e-05,
+      "loss": 1.0196,
+      "step": 5427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.378961367247009e-05,
+      "loss": 0.9868,
+      "step": 5428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3787544052462036e-05,
+      "loss": 0.9792,
+      "step": 5429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3785474243032847e-05,
+      "loss": 0.9374,
+      "step": 5430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3783404244286037e-05,
+      "loss": 0.9149,
+      "step": 5431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3781334056325126e-05,
+      "loss": 0.6584,
+      "step": 5432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.377926367925365e-05,
+      "loss": 0.9984,
+      "step": 5433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3777193113175149e-05,
+      "loss": 0.9864,
+      "step": 5434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3775122358193182e-05,
+      "loss": 0.6388,
+      "step": 5435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3773051414411308e-05,
+      "loss": 0.9436,
+      "step": 5436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3770980281933091e-05,
+      "loss": 0.8491,
+      "step": 5437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3768908960862122e-05,
+      "loss": 1.0381,
+      "step": 5438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3766837451301985e-05,
+      "loss": 0.8974,
+      "step": 5439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3764765753356282e-05,
+      "loss": 0.6655,
+      "step": 5440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.376269386712862e-05,
+      "loss": 1.0217,
+      "step": 5441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3760621792722622e-05,
+      "loss": 0.9199,
+      "step": 5442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.375854953024191e-05,
+      "loss": 1.0924,
+      "step": 5443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3756477079790122e-05,
+      "loss": 0.811,
+      "step": 5444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3754404441470908e-05,
+      "loss": 1.0267,
+      "step": 5445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3752331615387919e-05,
+      "loss": 1.0125,
+      "step": 5446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3750258601644829e-05,
+      "loss": 0.9518,
+      "step": 5447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3748185400345302e-05,
+      "loss": 0.9205,
+      "step": 5448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3746112011593026e-05,
+      "loss": 0.9778,
+      "step": 5449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3744038435491698e-05,
+      "loss": 0.9562,
+      "step": 5450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3741964672145021e-05,
+      "loss": 1.1067,
+      "step": 5451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3739890721656699e-05,
+      "loss": 1.0735,
+      "step": 5452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3737816584130466e-05,
+      "loss": 0.9587,
+      "step": 5453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.373574225967004e-05,
+      "loss": 0.621,
+      "step": 5454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.373366774837917e-05,
+      "loss": 0.88,
+      "step": 5455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3731593050361603e-05,
+      "loss": 1.0079,
+      "step": 5456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.37295181657211e-05,
+      "loss": 0.6682,
+      "step": 5457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3727443094561426e-05,
+      "loss": 1.0384,
+      "step": 5458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3725367836986358e-05,
+      "loss": 1.0443,
+      "step": 5459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.372329239309969e-05,
+      "loss": 1.0005,
+      "step": 5460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3721216763005208e-05,
+      "loss": 0.9341,
+      "step": 5461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3719140946806727e-05,
+      "loss": 1.0927,
+      "step": 5462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3717064944608054e-05,
+      "loss": 0.9426,
+      "step": 5463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3714988756513017e-05,
+      "loss": 0.9491,
+      "step": 5464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3712912382625452e-05,
+      "loss": 0.8981,
+      "step": 5465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3710835823049196e-05,
+      "loss": 1.0853,
+      "step": 5466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3708759077888108e-05,
+      "loss": 0.8984,
+      "step": 5467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3706682147246044e-05,
+      "loss": 0.9842,
+      "step": 5468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3704605031226874e-05,
+      "loss": 0.9783,
+      "step": 5469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3702527729934483e-05,
+      "loss": 1.0609,
+      "step": 5470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3700450243472757e-05,
+      "loss": 0.9691,
+      "step": 5471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3698372571945593e-05,
+      "loss": 0.9688,
+      "step": 5472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3696294715456904e-05,
+      "loss": 1.0432,
+      "step": 5473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3694216674110599e-05,
+      "loss": 0.9423,
+      "step": 5474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.369213844801061e-05,
+      "loss": 0.923,
+      "step": 5475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3690060037260871e-05,
+      "loss": 0.6842,
+      "step": 5476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3687981441965325e-05,
+      "loss": 0.9863,
+      "step": 5477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.368590266222793e-05,
+      "loss": 0.949,
+      "step": 5478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3683823698152645e-05,
+      "loss": 0.8868,
+      "step": 5479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3681744549843445e-05,
+      "loss": 1.0031,
+      "step": 5480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3679665217404306e-05,
+      "loss": 0.9306,
+      "step": 5481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3677585700939227e-05,
+      "loss": 0.8918,
+      "step": 5482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3675506000552206e-05,
+      "loss": 0.9877,
+      "step": 5483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3673426116347246e-05,
+      "loss": 0.9735,
+      "step": 5484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3671346048428373e-05,
+      "loss": 0.98,
+      "step": 5485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3669265796899607e-05,
+      "loss": 0.9019,
+      "step": 5486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3667185361864992e-05,
+      "loss": 0.8325,
+      "step": 5487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3665104743428568e-05,
+      "loss": 0.9894,
+      "step": 5488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3663023941694396e-05,
+      "loss": 1.032,
+      "step": 5489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3660942956766536e-05,
+      "loss": 0.9687,
+      "step": 5490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.365886178874906e-05,
+      "loss": 0.658,
+      "step": 5491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3656780437746056e-05,
+      "loss": 0.9209,
+      "step": 5492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.365469890386161e-05,
+      "loss": 0.9214,
+      "step": 5493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3652617187199826e-05,
+      "loss": 1.0359,
+      "step": 5494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3650535287864814e-05,
+      "loss": 0.6751,
+      "step": 5495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3648453205960692e-05,
+      "loss": 0.9737,
+      "step": 5496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3646370941591587e-05,
+      "loss": 1.0099,
+      "step": 5497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.364428849486164e-05,
+      "loss": 0.9457,
+      "step": 5498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3642205865874995e-05,
+      "loss": 0.944,
+      "step": 5499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3640123054735809e-05,
+      "loss": 0.8774,
+      "step": 5500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3638040061548243e-05,
+      "loss": 0.6425,
+      "step": 5501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3635956886416479e-05,
+      "loss": 0.823,
+      "step": 5502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.363387352944469e-05,
+      "loss": 0.9269,
+      "step": 5503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3631789990737074e-05,
+      "loss": 0.8045,
+      "step": 5504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3629706270397834e-05,
+      "loss": 0.9329,
+      "step": 5505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3627622368531172e-05,
+      "loss": 0.9778,
+      "step": 5506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3625538285241316e-05,
+      "loss": 1.0242,
+      "step": 5507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3623454020632488e-05,
+      "loss": 0.9614,
+      "step": 5508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.362136957480893e-05,
+      "loss": 0.963,
+      "step": 5509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3619284947874882e-05,
+      "loss": 0.6184,
+      "step": 5510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3617200139934607e-05,
+      "loss": 0.9651,
+      "step": 5511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3615115151092369e-05,
+      "loss": 0.9622,
+      "step": 5512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3613029981452432e-05,
+      "loss": 1.0274,
+      "step": 5513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3610944631119095e-05,
+      "loss": 1.0578,
+      "step": 5514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3608859100196634e-05,
+      "loss": 0.9888,
+      "step": 5515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3606773388789359e-05,
+      "loss": 0.9103,
+      "step": 5516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3604687497001576e-05,
+      "loss": 0.9576,
+      "step": 5517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3602601424937605e-05,
+      "loss": 0.9829,
+      "step": 5518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3600515172701774e-05,
+      "loss": 0.9905,
+      "step": 5519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3598428740398421e-05,
+      "loss": 0.9904,
+      "step": 5520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3596342128131887e-05,
+      "loss": 1.0016,
+      "step": 5521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3594255336006535e-05,
+      "loss": 0.8995,
+      "step": 5522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3592168364126717e-05,
+      "loss": 1.0205,
+      "step": 5523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3590081212596817e-05,
+      "loss": 0.9304,
+      "step": 5524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.358799388152121e-05,
+      "loss": 0.8676,
+      "step": 5525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3585906371004295e-05,
+      "loss": 0.9081,
+      "step": 5526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3583818681150462e-05,
+      "loss": 1.1845,
+      "step": 5527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.358173081206412e-05,
+      "loss": 0.9941,
+      "step": 5528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3579642763849695e-05,
+      "loss": 0.898,
+      "step": 5529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3577554536611607e-05,
+      "loss": 1.0075,
+      "step": 5530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3575466130454293e-05,
+      "loss": 0.8757,
+      "step": 5531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3573377545482203e-05,
+      "loss": 0.6161,
+      "step": 5532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3571288781799781e-05,
+      "loss": 0.6041,
+      "step": 5533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3569199839511496e-05,
+      "loss": 0.8593,
+      "step": 5534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3567110718721815e-05,
+      "loss": 1.022,
+      "step": 5535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3565021419535223e-05,
+      "loss": 0.9904,
+      "step": 5536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3562931942056205e-05,
+      "loss": 0.9864,
+      "step": 5537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3560842286389262e-05,
+      "loss": 0.9528,
+      "step": 5538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3558752452638901e-05,
+      "loss": 0.9362,
+      "step": 5539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3556662440909634e-05,
+      "loss": 0.9296,
+      "step": 5540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.355457225130599e-05,
+      "loss": 0.9412,
+      "step": 5541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3552481883932505e-05,
+      "loss": 1.0183,
+      "step": 5542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3550391338893713e-05,
+      "loss": 0.808,
+      "step": 5543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.354830061629417e-05,
+      "loss": 0.9138,
+      "step": 5544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3546209716238438e-05,
+      "loss": 0.9847,
+      "step": 5545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3544118638831089e-05,
+      "loss": 0.9779,
+      "step": 5546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3542027384176696e-05,
+      "loss": 0.9503,
+      "step": 5547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3539935952379844e-05,
+      "loss": 1.0157,
+      "step": 5548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3537844343545135e-05,
+      "loss": 0.9498,
+      "step": 5549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3535752557777171e-05,
+      "loss": 0.9514,
+      "step": 5550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3533660595180565e-05,
+      "loss": 0.9653,
+      "step": 5551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.353156845585994e-05,
+      "loss": 0.8993,
+      "step": 5552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3529476139919923e-05,
+      "loss": 0.9808,
+      "step": 5553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3527383647465165e-05,
+      "loss": 0.9275,
+      "step": 5554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3525290978600303e-05,
+      "loss": 0.6624,
+      "step": 5555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3523198133430002e-05,
+      "loss": 1.0032,
+      "step": 5556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3521105112058924e-05,
+      "loss": 0.934,
+      "step": 5557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3519011914591749e-05,
+      "loss": 0.9718,
+      "step": 5558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3516918541133155e-05,
+      "loss": 1.0547,
+      "step": 5559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3514824991787841e-05,
+      "loss": 0.8036,
+      "step": 5560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3512731266660509e-05,
+      "loss": 0.9748,
+      "step": 5561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3510637365855865e-05,
+      "loss": 1.0026,
+      "step": 5562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3508543289478628e-05,
+      "loss": 0.9623,
+      "step": 5563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3506449037633534e-05,
+      "loss": 0.9024,
+      "step": 5564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.350435461042531e-05,
+      "loss": 0.9792,
+      "step": 5565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3502260007958708e-05,
+      "loss": 0.9247,
+      "step": 5566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.350016523033848e-05,
+      "loss": 0.8896,
+      "step": 5567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3498070277669386e-05,
+      "loss": 1.0675,
+      "step": 5568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3495975150056206e-05,
+      "loss": 0.9873,
+      "step": 5569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3493879847603713e-05,
+      "loss": 0.9295,
+      "step": 5570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3491784370416703e-05,
+      "loss": 0.9538,
+      "step": 5571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3489688718599968e-05,
+      "loss": 0.8654,
+      "step": 5572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.348759289225832e-05,
+      "loss": 0.8412,
+      "step": 5573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3485496891496572e-05,
+      "loss": 1.0232,
+      "step": 5574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3483400716419546e-05,
+      "loss": 0.9624,
+      "step": 5575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3481304367132077e-05,
+      "loss": 0.9292,
+      "step": 5576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3479207843739013e-05,
+      "loss": 0.6729,
+      "step": 5577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3477111146345194e-05,
+      "loss": 0.9822,
+      "step": 5578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3475014275055487e-05,
+      "loss": 1.0402,
+      "step": 5579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3472917229974755e-05,
+      "loss": 0.9053,
+      "step": 5580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3470820011207879e-05,
+      "loss": 0.9394,
+      "step": 5581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3468722618859743e-05,
+      "loss": 0.8899,
+      "step": 5582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3466625053035236e-05,
+      "loss": 1.0908,
+      "step": 5583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3464527313839268e-05,
+      "loss": 0.9889,
+      "step": 5584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3462429401376744e-05,
+      "loss": 0.9815,
+      "step": 5585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3460331315752589e-05,
+      "loss": 0.9043,
+      "step": 5586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3458233057071728e-05,
+      "loss": 0.946,
+      "step": 5587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.34561346254391e-05,
+      "loss": 0.965,
+      "step": 5588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3454036020959651e-05,
+      "loss": 0.9721,
+      "step": 5589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3451937243738334e-05,
+      "loss": 0.8445,
+      "step": 5590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3449838293880112e-05,
+      "loss": 0.8833,
+      "step": 5591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.344773917148996e-05,
+      "loss": 1.0172,
+      "step": 5592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3445639876672858e-05,
+      "loss": 0.9783,
+      "step": 5593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3443540409533793e-05,
+      "loss": 0.843,
+      "step": 5594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3441440770177763e-05,
+      "loss": 0.9848,
+      "step": 5595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3439340958709774e-05,
+      "loss": 0.8745,
+      "step": 5596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3437240975234842e-05,
+      "loss": 0.9159,
+      "step": 5597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.343514081985799e-05,
+      "loss": 0.8448,
+      "step": 5598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3433040492684253e-05,
+      "loss": 0.9892,
+      "step": 5599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3430939993818665e-05,
+      "loss": 0.9295,
+      "step": 5600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3428839323366282e-05,
+      "loss": 0.9314,
+      "step": 5601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.342673848143216e-05,
+      "loss": 0.9604,
+      "step": 5602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3424637468121364e-05,
+      "loss": 0.9936,
+      "step": 5603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3422536283538969e-05,
+      "loss": 0.9632,
+      "step": 5604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.342043492779006e-05,
+      "loss": 1.0344,
+      "step": 5605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3418333400979726e-05,
+      "loss": 0.9637,
+      "step": 5606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3416231703213072e-05,
+      "loss": 0.9943,
+      "step": 5607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3414129834595207e-05,
+      "loss": 0.8582,
+      "step": 5608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3412027795231248e-05,
+      "loss": 0.9411,
+      "step": 5609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3409925585226317e-05,
+      "loss": 0.9224,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3407823204685558e-05,
+      "loss": 0.8474,
+      "step": 5611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3405720653714104e-05,
+      "loss": 0.9707,
+      "step": 5612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3403617932417114e-05,
+      "loss": 0.8817,
+      "step": 5613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3401515040899746e-05,
+      "loss": 0.9431,
+      "step": 5614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.339941197926717e-05,
+      "loss": 0.8976,
+      "step": 5615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3397308747624562e-05,
+      "loss": 0.936,
+      "step": 5616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3395205346077109e-05,
+      "loss": 0.9796,
+      "step": 5617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3393101774730006e-05,
+      "loss": 0.963,
+      "step": 5618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3390998033688454e-05,
+      "loss": 0.9769,
+      "step": 5619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3388894123057667e-05,
+      "loss": 1.038,
+      "step": 5620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3386790042942862e-05,
+      "loss": 0.9746,
+      "step": 5621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.338468579344927e-05,
+      "loss": 1.0521,
+      "step": 5622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3382581374682127e-05,
+      "loss": 0.9529,
+      "step": 5623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.338047678674668e-05,
+      "loss": 0.98,
+      "step": 5624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3378372029748179e-05,
+      "loss": 0.8508,
+      "step": 5625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.337626710379189e-05,
+      "loss": 1.0491,
+      "step": 5626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3374162008983078e-05,
+      "loss": 0.8323,
+      "step": 5627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.337205674542703e-05,
+      "loss": 1.0502,
+      "step": 5628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3369951313229028e-05,
+      "loss": 0.8671,
+      "step": 5629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3367845712494373e-05,
+      "loss": 0.9939,
+      "step": 5630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3365739943328367e-05,
+      "loss": 0.9151,
+      "step": 5631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3363634005836316e-05,
+      "loss": 1.0526,
+      "step": 5632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3361527900123552e-05,
+      "loss": 1.0011,
+      "step": 5633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.33594216262954e-05,
+      "loss": 0.9154,
+      "step": 5634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3357315184457199e-05,
+      "loss": 0.9366,
+      "step": 5635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3355208574714293e-05,
+      "loss": 0.9258,
+      "step": 5636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3353101797172036e-05,
+      "loss": 1.0033,
+      "step": 5637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.33509948519358e-05,
+      "loss": 0.9969,
+      "step": 5638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3348887739110944e-05,
+      "loss": 0.89,
+      "step": 5639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3346780458802861e-05,
+      "loss": 0.9654,
+      "step": 5640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3344673011116934e-05,
+      "loss": 0.9377,
+      "step": 5641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3342565396158555e-05,
+      "loss": 0.8624,
+      "step": 5642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3340457614033136e-05,
+      "loss": 0.8842,
+      "step": 5643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3338349664846087e-05,
+      "loss": 0.9422,
+      "step": 5644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.333624154870283e-05,
+      "loss": 1.0561,
+      "step": 5645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.33341332657088e-05,
+      "loss": 0.9498,
+      "step": 5646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.333202481596943e-05,
+      "loss": 0.997,
+      "step": 5647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3329916199590164e-05,
+      "loss": 0.8867,
+      "step": 5648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3327807416676466e-05,
+      "loss": 0.9792,
+      "step": 5649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3325698467333796e-05,
+      "loss": 0.9486,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3323589351667624e-05,
+      "loss": 0.832,
+      "step": 5651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3321480069783434e-05,
+      "loss": 0.9481,
+      "step": 5652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3319370621786711e-05,
+      "loss": 0.9129,
+      "step": 5653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3317261007782949e-05,
+      "loss": 1.0255,
+      "step": 5654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3315151227877664e-05,
+      "loss": 0.9375,
+      "step": 5655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.331304128217636e-05,
+      "loss": 0.8618,
+      "step": 5656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.331093117078456e-05,
+      "loss": 0.9686,
+      "step": 5657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3308820893807799e-05,
+      "loss": 1.0666,
+      "step": 5658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3306710451351608e-05,
+      "loss": 1.0156,
+      "step": 5659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3304599843521542e-05,
+      "loss": 0.9523,
+      "step": 5660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3302489070423148e-05,
+      "loss": 0.9055,
+      "step": 5661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3300378132161992e-05,
+      "loss": 0.9,
+      "step": 5662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.329826702884365e-05,
+      "loss": 0.9371,
+      "step": 5663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3296155760573694e-05,
+      "loss": 0.8751,
+      "step": 5664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3294044327457718e-05,
+      "loss": 1.0198,
+      "step": 5665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3291932729601314e-05,
+      "loss": 0.85,
+      "step": 5666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.328982096711009e-05,
+      "loss": 0.9701,
+      "step": 5667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3287709040089656e-05,
+      "loss": 0.8957,
+      "step": 5668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.328559694864563e-05,
+      "loss": 0.9582,
+      "step": 5669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3283484692883648e-05,
+      "loss": 0.9104,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3281372272909343e-05,
+      "loss": 0.8904,
+      "step": 5671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3279259688828361e-05,
+      "loss": 1.0222,
+      "step": 5672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3277146940746358e-05,
+      "loss": 0.8949,
+      "step": 5673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3275034028768992e-05,
+      "loss": 1.1009,
+      "step": 5674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3272920953001935e-05,
+      "loss": 0.9525,
+      "step": 5675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3270807713550866e-05,
+      "loss": 0.8353,
+      "step": 5676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3268694310521473e-05,
+      "loss": 1.0907,
+      "step": 5677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3266580744019446e-05,
+      "loss": 0.9218,
+      "step": 5678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3264467014150491e-05,
+      "loss": 0.9222,
+      "step": 5679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3262353121020319e-05,
+      "loss": 0.9401,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3260239064734646e-05,
+      "loss": 0.8423,
+      "step": 5681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3258124845399205e-05,
+      "loss": 0.9348,
+      "step": 5682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3256010463119728e-05,
+      "loss": 0.6484,
+      "step": 5683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3253895918001958e-05,
+      "loss": 0.9354,
+      "step": 5684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3251781210151648e-05,
+      "loss": 0.8757,
+      "step": 5685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3249666339674555e-05,
+      "loss": 0.9978,
+      "step": 5686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3247551306676457e-05,
+      "loss": 0.9419,
+      "step": 5687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.324543611126312e-05,
+      "loss": 0.9174,
+      "step": 5688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3243320753540328e-05,
+      "loss": 1.1178,
+      "step": 5689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3241205233613879e-05,
+      "loss": 0.9396,
+      "step": 5690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323908955158957e-05,
+      "loss": 1.0197,
+      "step": 5691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3236973707573214e-05,
+      "loss": 0.9022,
+      "step": 5692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3234857701670625e-05,
+      "loss": 0.994,
+      "step": 5693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3232741533987624e-05,
+      "loss": 0.9129,
+      "step": 5694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323062520463005e-05,
+      "loss": 0.9263,
+      "step": 5695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3228508713703738e-05,
+      "loss": 0.6894,
+      "step": 5696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3226392061314544e-05,
+      "loss": 0.9705,
+      "step": 5697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.322427524756832e-05,
+      "loss": 1.0041,
+      "step": 5698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3222158272570934e-05,
+      "loss": 0.9228,
+      "step": 5699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3220041136428259e-05,
+      "loss": 0.9677,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3217923839246169e-05,
+      "loss": 0.8048,
+      "step": 5701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3215806381130567e-05,
+      "loss": 0.869,
+      "step": 5702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3213688762187345e-05,
+      "loss": 1.0185,
+      "step": 5703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3211570982522401e-05,
+      "loss": 0.9763,
+      "step": 5704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.320945304224166e-05,
+      "loss": 0.9646,
+      "step": 5705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3207334941451033e-05,
+      "loss": 0.9643,
+      "step": 5706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3205216680256458e-05,
+      "loss": 0.8771,
+      "step": 5707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3203098258763871e-05,
+      "loss": 0.8872,
+      "step": 5708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3200979677079214e-05,
+      "loss": 0.6534,
+      "step": 5709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3198860935308445e-05,
+      "loss": 0.9051,
+      "step": 5710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.319674203355752e-05,
+      "loss": 0.9786,
+      "step": 5711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3194622971932412e-05,
+      "loss": 1.006,
+      "step": 5712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3192503750539099e-05,
+      "loss": 0.9867,
+      "step": 5713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3190384369483569e-05,
+      "loss": 0.8556,
+      "step": 5714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.318826482887181e-05,
+      "loss": 0.8535,
+      "step": 5715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3186145128809827e-05,
+      "loss": 1.0057,
+      "step": 5716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.318402526940363e-05,
+      "loss": 0.994,
+      "step": 5717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3181905250759237e-05,
+      "loss": 0.9164,
+      "step": 5718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3179785072982669e-05,
+      "loss": 0.9132,
+      "step": 5719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3177664736179967e-05,
+      "loss": 0.9233,
+      "step": 5720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3175544240457163e-05,
+      "loss": 0.8923,
+      "step": 5721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3173423585920318e-05,
+      "loss": 0.9107,
+      "step": 5722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3171302772675476e-05,
+      "loss": 0.9395,
+      "step": 5723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3169181800828713e-05,
+      "loss": 0.8931,
+      "step": 5724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.31670606704861e-05,
+      "loss": 0.8496,
+      "step": 5725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3164939381753714e-05,
+      "loss": 0.9635,
+      "step": 5726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3162817934737648e-05,
+      "loss": 0.9926,
+      "step": 5727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3160696329543996e-05,
+      "loss": 0.9043,
+      "step": 5728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3158574566278866e-05,
+      "loss": 0.9495,
+      "step": 5729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3156452645048368e-05,
+      "loss": 0.9751,
+      "step": 5730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3154330565958624e-05,
+      "loss": 0.9498,
+      "step": 5731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3152208329115759e-05,
+      "loss": 0.9197,
+      "step": 5732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3150085934625916e-05,
+      "loss": 0.9315,
+      "step": 5733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3147963382595237e-05,
+      "loss": 0.9329,
+      "step": 5734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3145840673129873e-05,
+      "loss": 0.9724,
+      "step": 5735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3143717806335983e-05,
+      "loss": 1.0605,
+      "step": 5736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.314159478231974e-05,
+      "loss": 0.9773,
+      "step": 5737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3139471601187312e-05,
+      "loss": 0.6178,
+      "step": 5738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3137348263044891e-05,
+      "loss": 0.8667,
+      "step": 5739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3135224767998663e-05,
+      "loss": 1.0009,
+      "step": 5740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3133101116154827e-05,
+      "loss": 0.978,
+      "step": 5741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3130977307619595e-05,
+      "loss": 0.9457,
+      "step": 5742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3128853342499175e-05,
+      "loss": 1.0308,
+      "step": 5743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.31267292208998e-05,
+      "loss": 0.9095,
+      "step": 5744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3124604942927687e-05,
+      "loss": 0.9046,
+      "step": 5745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3122480508689091e-05,
+      "loss": 1.0067,
+      "step": 5746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3120355918290244e-05,
+      "loss": 0.8591,
+      "step": 5747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3118231171837405e-05,
+      "loss": 0.9556,
+      "step": 5748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3116106269436841e-05,
+      "loss": 1.0256,
+      "step": 5749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3113981211194817e-05,
+      "loss": 0.9942,
+      "step": 5750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3111855997217609e-05,
+      "loss": 0.9557,
+      "step": 5751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3109730627611505e-05,
+      "loss": 1.0047,
+      "step": 5752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.31076051024828e-05,
+      "loss": 1.0517,
+      "step": 5753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3105479421937788e-05,
+      "loss": 0.6413,
+      "step": 5754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3103353586082788e-05,
+      "loss": 0.8608,
+      "step": 5755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3101227595024106e-05,
+      "loss": 0.93,
+      "step": 5756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3099101448868074e-05,
+      "loss": 0.9446,
+      "step": 5757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3096975147721018e-05,
+      "loss": 0.6355,
+      "step": 5758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3094848691689283e-05,
+      "loss": 0.9176,
+      "step": 5759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3092722080879211e-05,
+      "loss": 0.8964,
+      "step": 5760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3090595315397162e-05,
+      "loss": 0.9715,
+      "step": 5761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.30884683953495e-05,
+      "loss": 0.8963,
+      "step": 5762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3086341320842588e-05,
+      "loss": 0.9409,
+      "step": 5763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3084214091982812e-05,
+      "loss": 0.8784,
+      "step": 5764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3082086708876556e-05,
+      "loss": 0.9369,
+      "step": 5765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3079959171630212e-05,
+      "loss": 0.9777,
+      "step": 5766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3077831480350182e-05,
+      "loss": 0.676,
+      "step": 5767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3075703635142877e-05,
+      "loss": 0.9057,
+      "step": 5768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3073575636114714e-05,
+      "loss": 0.8443,
+      "step": 5769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3071447483372115e-05,
+      "loss": 0.8734,
+      "step": 5770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3069319177021518e-05,
+      "loss": 0.9689,
+      "step": 5771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3067190717169356e-05,
+      "loss": 0.9046,
+      "step": 5772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3065062103922077e-05,
+      "loss": 0.9417,
+      "step": 5773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3062933337386144e-05,
+      "loss": 0.9653,
+      "step": 5774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3060804417668009e-05,
+      "loss": 0.9426,
+      "step": 5775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3058675344874154e-05,
+      "loss": 0.7865,
+      "step": 5776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.305654611911105e-05,
+      "loss": 0.6823,
+      "step": 5777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3054416740485185e-05,
+      "loss": 0.9916,
+      "step": 5778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3052287209103051e-05,
+      "loss": 0.9321,
+      "step": 5779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3050157525071152e-05,
+      "loss": 1.0628,
+      "step": 5780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3048027688495996e-05,
+      "loss": 0.9086,
+      "step": 5781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.30458976994841e-05,
+      "loss": 1.006,
+      "step": 5782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3043767558141988e-05,
+      "loss": 0.9188,
+      "step": 5783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.304163726457619e-05,
+      "loss": 0.9254,
+      "step": 5784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3039506818893246e-05,
+      "loss": 0.9845,
+      "step": 5785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3037376221199707e-05,
+      "loss": 0.9297,
+      "step": 5786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3035245471602122e-05,
+      "loss": 0.8652,
+      "step": 5787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3033114570207055e-05,
+      "loss": 0.9345,
+      "step": 5788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3030983517121076e-05,
+      "loss": 0.8973,
+      "step": 5789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3028852312450764e-05,
+      "loss": 0.7177,
+      "step": 5790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3026720956302703e-05,
+      "loss": 0.9675,
+      "step": 5791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3024589448783482e-05,
+      "loss": 0.8672,
+      "step": 5792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3022457789999707e-05,
+      "loss": 1.029,
+      "step": 5793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3020325980057984e-05,
+      "loss": 0.8514,
+      "step": 5794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.301819401906492e-05,
+      "loss": 1.033,
+      "step": 5795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.301606190712715e-05,
+      "loss": 0.8705,
+      "step": 5796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3013929644351302e-05,
+      "loss": 1.0217,
+      "step": 5797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3011797230844008e-05,
+      "loss": 0.973,
+      "step": 5798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3009664666711916e-05,
+      "loss": 0.9498,
+      "step": 5799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.300753195206168e-05,
+      "loss": 0.951,
+      "step": 5800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3005399086999962e-05,
+      "loss": 0.8802,
+      "step": 5801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3003266071633424e-05,
+      "loss": 1.073,
+      "step": 5802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3001132906068747e-05,
+      "loss": 0.9173,
+      "step": 5803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2998999590412616e-05,
+      "loss": 0.9069,
+      "step": 5804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2996866124771712e-05,
+      "loss": 0.948,
+      "step": 5805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2994732509252744e-05,
+      "loss": 0.9582,
+      "step": 5806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2992598743962408e-05,
+      "loss": 0.9306,
+      "step": 5807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2990464829007422e-05,
+      "loss": 0.9291,
+      "step": 5808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.298833076449451e-05,
+      "loss": 0.9812,
+      "step": 5809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2986196550530392e-05,
+      "loss": 1.0089,
+      "step": 5810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.298406218722181e-05,
+      "loss": 0.8416,
+      "step": 5811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2981927674675502e-05,
+      "loss": 0.8697,
+      "step": 5812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2979793012998224e-05,
+      "loss": 0.895,
+      "step": 5813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2977658202296731e-05,
+      "loss": 0.9828,
+      "step": 5814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2975523242677783e-05,
+      "loss": 1.0136,
+      "step": 5815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2973388134248162e-05,
+      "loss": 0.8887,
+      "step": 5816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2971252877114639e-05,
+      "loss": 0.9113,
+      "step": 5817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2969117471384012e-05,
+      "loss": 1.015,
+      "step": 5818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.296698191716307e-05,
+      "loss": 0.9421,
+      "step": 5819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2964846214558613e-05,
+      "loss": 0.9433,
+      "step": 5820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2962710363677458e-05,
+      "loss": 0.8279,
+      "step": 5821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2960574364626413e-05,
+      "loss": 0.9163,
+      "step": 5822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2958438217512313e-05,
+      "loss": 0.9359,
+      "step": 5823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2956301922441983e-05,
+      "loss": 0.9376,
+      "step": 5824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2954165479522263e-05,
+      "loss": 0.9438,
+      "step": 5825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2952028888860002e-05,
+      "loss": 0.9266,
+      "step": 5826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.294989215056205e-05,
+      "loss": 0.8563,
+      "step": 5827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.294775526473528e-05,
+      "loss": 1.0201,
+      "step": 5828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2945618231486552e-05,
+      "loss": 0.8783,
+      "step": 5829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2943481050922742e-05,
+      "loss": 0.9241,
+      "step": 5830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2941343723150738e-05,
+      "loss": 0.9189,
+      "step": 5831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2939206248277427e-05,
+      "loss": 0.9995,
+      "step": 5832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2937068626409712e-05,
+      "loss": 0.9684,
+      "step": 5833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2934930857654497e-05,
+      "loss": 0.8887,
+      "step": 5834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.293279294211869e-05,
+      "loss": 0.9158,
+      "step": 5835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2930654879909222e-05,
+      "loss": 0.9274,
+      "step": 5836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.292851667113301e-05,
+      "loss": 0.9935,
+      "step": 5837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2926378315897e-05,
+      "loss": 0.9645,
+      "step": 5838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2924239814308123e-05,
+      "loss": 0.6573,
+      "step": 5839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.292210116647334e-05,
+      "loss": 0.9597,
+      "step": 5840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2919962372499602e-05,
+      "loss": 0.8832,
+      "step": 5841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.291782343249387e-05,
+      "loss": 0.6635,
+      "step": 5842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2915684346563126e-05,
+      "loss": 0.9758,
+      "step": 5843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2913545114814343e-05,
+      "loss": 0.9367,
+      "step": 5844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2911405737354505e-05,
+      "loss": 0.9129,
+      "step": 5845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2909266214290612e-05,
+      "loss": 1.0768,
+      "step": 5846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.290712654572966e-05,
+      "loss": 1.0381,
+      "step": 5847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2904986731778659e-05,
+      "loss": 0.9313,
+      "step": 5848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2902846772544625e-05,
+      "loss": 0.994,
+      "step": 5849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.290070666813458e-05,
+      "loss": 0.8902,
+      "step": 5850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2898566418655554e-05,
+      "loss": 0.8477,
+      "step": 5851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2896426024214583e-05,
+      "loss": 0.9802,
+      "step": 5852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2894285484918715e-05,
+      "loss": 0.9531,
+      "step": 5853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2892144800875e-05,
+      "loss": 1.0274,
+      "step": 5854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2890003972190497e-05,
+      "loss": 0.9786,
+      "step": 5855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2887862998972274e-05,
+      "loss": 1.0203,
+      "step": 5856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2885721881327395e-05,
+      "loss": 0.9821,
+      "step": 5857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2883580619362956e-05,
+      "loss": 0.96,
+      "step": 5858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2881439213186036e-05,
+      "loss": 0.9332,
+      "step": 5859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287929766290373e-05,
+      "loss": 0.9605,
+      "step": 5860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2877155968623142e-05,
+      "loss": 0.8121,
+      "step": 5861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287501413045138e-05,
+      "loss": 0.9211,
+      "step": 5862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2872872148495565e-05,
+      "loss": 1.0527,
+      "step": 5863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2870730022862816e-05,
+      "loss": 0.9299,
+      "step": 5864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2868587753660269e-05,
+      "loss": 0.9164,
+      "step": 5865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2866445340995057e-05,
+      "loss": 0.923,
+      "step": 5866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.286430278497433e-05,
+      "loss": 0.8551,
+      "step": 5867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2862160085705237e-05,
+      "loss": 1.0437,
+      "step": 5868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.286001724329494e-05,
+      "loss": 0.9718,
+      "step": 5869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2857874257850605e-05,
+      "loss": 0.9091,
+      "step": 5870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2855731129479407e-05,
+      "loss": 0.8847,
+      "step": 5871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2853587858288527e-05,
+      "loss": 0.97,
+      "step": 5872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2851444444385152e-05,
+      "loss": 0.9479,
+      "step": 5873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2849300887876479e-05,
+      "loss": 0.9505,
+      "step": 5874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2847157188869713e-05,
+      "loss": 1.0344,
+      "step": 5875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.284501334747206e-05,
+      "loss": 0.6418,
+      "step": 5876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2842869363790738e-05,
+      "loss": 0.9305,
+      "step": 5877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2840725237932972e-05,
+      "loss": 0.8945,
+      "step": 5878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.283858097000599e-05,
+      "loss": 0.6174,
+      "step": 5879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2836436560117036e-05,
+      "loss": 0.6459,
+      "step": 5880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2834292008373353e-05,
+      "loss": 1.0566,
+      "step": 5881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2832147314882188e-05,
+      "loss": 0.9619,
+      "step": 5882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.283000247975081e-05,
+      "loss": 0.612,
+      "step": 5883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2827857503086476e-05,
+      "loss": 0.9071,
+      "step": 5884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2825712384996467e-05,
+      "loss": 0.9016,
+      "step": 5885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.282356712558806e-05,
+      "loss": 0.9545,
+      "step": 5886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2821421724968549e-05,
+      "loss": 0.9617,
+      "step": 5887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.281927618324522e-05,
+      "loss": 0.9129,
+      "step": 5888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2817130500525377e-05,
+      "loss": 0.8698,
+      "step": 5889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2814984676916334e-05,
+      "loss": 0.9807,
+      "step": 5890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2812838712525405e-05,
+      "loss": 0.9871,
+      "step": 5891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.281069260745991e-05,
+      "loss": 1.0176,
+      "step": 5892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2808546361827183e-05,
+      "loss": 0.9022,
+      "step": 5893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2806399975734557e-05,
+      "loss": 0.9205,
+      "step": 5894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2804253449289384e-05,
+      "loss": 0.9584,
+      "step": 5895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2802106782599004e-05,
+      "loss": 0.8637,
+      "step": 5896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2799959975770785e-05,
+      "loss": 0.6061,
+      "step": 5897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.279781302891209e-05,
+      "loss": 1.0271,
+      "step": 5898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2795665942130286e-05,
+      "loss": 0.9234,
+      "step": 5899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2793518715532757e-05,
+      "loss": 0.8464,
+      "step": 5900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2791371349226888e-05,
+      "loss": 1.0244,
+      "step": 5901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2789223843320074e-05,
+      "loss": 0.8483,
+      "step": 5902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2787076197919715e-05,
+      "loss": 1.0097,
+      "step": 5903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.278492841313321e-05,
+      "loss": 0.9019,
+      "step": 5904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2782780489067985e-05,
+      "loss": 0.9047,
+      "step": 5905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2780632425831455e-05,
+      "loss": 0.9823,
+      "step": 5906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2778484223531052e-05,
+      "loss": 1.0273,
+      "step": 5907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2776335882274204e-05,
+      "loss": 0.8651,
+      "step": 5908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2774187402168358e-05,
+      "loss": 0.8416,
+      "step": 5909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2772038783320965e-05,
+      "loss": 0.9514,
+      "step": 5910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2769890025839476e-05,
+      "loss": 0.9406,
+      "step": 5911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2767741129831357e-05,
+      "loss": 0.9845,
+      "step": 5912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2765592095404075e-05,
+      "loss": 0.9361,
+      "step": 5913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.276344292266511e-05,
+      "loss": 0.9716,
+      "step": 5914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2761293611721941e-05,
+      "loss": 0.9483,
+      "step": 5915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2759144162682063e-05,
+      "loss": 0.9005,
+      "step": 5916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2756994575652973e-05,
+      "loss": 0.8952,
+      "step": 5917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2754844850742171e-05,
+      "loss": 0.9698,
+      "step": 5918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2752694988057173e-05,
+      "loss": 0.9729,
+      "step": 5919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2750544987705494e-05,
+      "loss": 0.9748,
+      "step": 5920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2748394849794656e-05,
+      "loss": 0.9109,
+      "step": 5921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.27462445744322e-05,
+      "loss": 0.9273,
+      "step": 5922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2744094161725662e-05,
+      "loss": 0.9343,
+      "step": 5923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2741943611782582e-05,
+      "loss": 1.0126,
+      "step": 5924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2739792924710516e-05,
+      "loss": 0.9858,
+      "step": 5925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.273764210061702e-05,
+      "loss": 0.6714,
+      "step": 5926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2735491139609668e-05,
+      "loss": 0.9718,
+      "step": 5927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2733340041796028e-05,
+      "loss": 0.9057,
+      "step": 5928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2731188807283676e-05,
+      "loss": 0.9901,
+      "step": 5929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2729037436180205e-05,
+      "loss": 0.9627,
+      "step": 5930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2726885928593205e-05,
+      "loss": 0.8651,
+      "step": 5931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.272473428463028e-05,
+      "loss": 0.951,
+      "step": 5932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.272258250439903e-05,
+      "loss": 0.8824,
+      "step": 5933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2720430588007078e-05,
+      "loss": 0.931,
+      "step": 5934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.271827853556204e-05,
+      "loss": 1.1151,
+      "step": 5935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.271612634717154e-05,
+      "loss": 0.9126,
+      "step": 5936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.271397402294322e-05,
+      "loss": 0.9282,
+      "step": 5937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2711821562984719e-05,
+      "loss": 0.9075,
+      "step": 5938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2709668967403678e-05,
+      "loss": 0.8636,
+      "step": 5939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2707516236307763e-05,
+      "loss": 0.8844,
+      "step": 5940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2705363369804626e-05,
+      "loss": 1.0323,
+      "step": 5941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.270321036800194e-05,
+      "loss": 0.9389,
+      "step": 5942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2701057231007379e-05,
+      "loss": 0.854,
+      "step": 5943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2698903958928628e-05,
+      "loss": 1.0276,
+      "step": 5944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.269675055187337e-05,
+      "loss": 0.9514,
+      "step": 5945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.26945970099493e-05,
+      "loss": 0.9167,
+      "step": 5946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2692443333264129e-05,
+      "loss": 0.8996,
+      "step": 5947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2690289521925557e-05,
+      "loss": 0.9658,
+      "step": 5948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.26881355760413e-05,
+      "loss": 0.9389,
+      "step": 5949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2685981495719087e-05,
+      "loss": 0.9518,
+      "step": 5950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2683827281066636e-05,
+      "loss": 0.6543,
+      "step": 5951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2681672932191696e-05,
+      "loss": 0.896,
+      "step": 5952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2679518449202e-05,
+      "loss": 0.6768,
+      "step": 5953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2677363832205299e-05,
+      "loss": 0.8924,
+      "step": 5954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2675209081309352e-05,
+      "loss": 0.9951,
+      "step": 5955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2673054196621917e-05,
+      "loss": 0.9816,
+      "step": 5956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2670899178250769e-05,
+      "loss": 0.9833,
+      "step": 5957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2668744026303676e-05,
+      "loss": 0.962,
+      "step": 5958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2666588740888429e-05,
+      "loss": 0.9357,
+      "step": 5959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.266443332211281e-05,
+      "loss": 1.0846,
+      "step": 5960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2662277770084618e-05,
+      "loss": 1.0264,
+      "step": 5961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2660122084911661e-05,
+      "loss": 0.9377,
+      "step": 5962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2657966266701737e-05,
+      "loss": 0.8686,
+      "step": 5963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2655810315562674e-05,
+      "loss": 0.8825,
+      "step": 5964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2653654231602284e-05,
+      "loss": 0.8746,
+      "step": 5965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2651498014928402e-05,
+      "loss": 0.9633,
+      "step": 5966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2649341665648863e-05,
+      "loss": 0.9392,
+      "step": 5967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2647185183871507e-05,
+      "loss": 1.0141,
+      "step": 5968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2645028569704187e-05,
+      "loss": 0.9129,
+      "step": 5969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2642871823254758e-05,
+      "loss": 0.9749,
+      "step": 5970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2640714944631078e-05,
+      "loss": 0.8293,
+      "step": 5971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2638557933941025e-05,
+      "loss": 0.9372,
+      "step": 5972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2636400791292462e-05,
+      "loss": 0.9942,
+      "step": 5973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2634243516793283e-05,
+      "loss": 0.9617,
+      "step": 5974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2632086110551371e-05,
+      "loss": 0.9166,
+      "step": 5975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2629928572674621e-05,
+      "loss": 0.8287,
+      "step": 5976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2627770903270937e-05,
+      "loss": 0.8685,
+      "step": 5977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2625613102448224e-05,
+      "loss": 0.938,
+      "step": 5978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2623455170314403e-05,
+      "loss": 0.9244,
+      "step": 5979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2621297106977389e-05,
+      "loss": 0.8774,
+      "step": 5980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2619138912545115e-05,
+      "loss": 0.9938,
+      "step": 5981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2616980587125514e-05,
+      "loss": 0.9217,
+      "step": 5982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2614822130826522e-05,
+      "loss": 0.8695,
+      "step": 5983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2612663543756098e-05,
+      "loss": 0.9802,
+      "step": 5984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2610504826022192e-05,
+      "loss": 0.9539,
+      "step": 5985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2608345977732761e-05,
+      "loss": 1.0401,
+      "step": 5986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2606186998995777e-05,
+      "loss": 0.6215,
+      "step": 5987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2604027889919208e-05,
+      "loss": 1.0366,
+      "step": 5988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2601868650611044e-05,
+      "loss": 0.925,
+      "step": 5989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2599709281179262e-05,
+      "loss": 0.9987,
+      "step": 5990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2597549781731864e-05,
+      "loss": 0.9873,
+      "step": 5991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2595390152376844e-05,
+      "loss": 0.9716,
+      "step": 5992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2593230393222209e-05,
+      "loss": 1.0213,
+      "step": 5993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2591070504375977e-05,
+      "loss": 0.6598,
+      "step": 5994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2588910485946163e-05,
+      "loss": 0.9453,
+      "step": 5995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2586750338040793e-05,
+      "loss": 0.7097,
+      "step": 5996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2584590060767901e-05,
+      "loss": 1.005,
+      "step": 5997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2582429654235522e-05,
+      "loss": 1.0836,
+      "step": 5998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.258026911855171e-05,
+      "loss": 0.9601,
+      "step": 5999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.257810845382451e-05,
+      "loss": 0.6847,
+      "step": 6000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2575947660161983e-05,
+      "loss": 1.121,
+      "step": 6001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2573786737672191e-05,
+      "loss": 0.9714,
+      "step": 6002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2571625686463205e-05,
+      "loss": 1.0248,
+      "step": 6003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.256946450664311e-05,
+      "loss": 0.9894,
+      "step": 6004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.256730319831998e-05,
+      "loss": 0.992,
+      "step": 6005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2565141761601914e-05,
+      "loss": 0.9196,
+      "step": 6006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2562980196597003e-05,
+      "loss": 0.9244,
+      "step": 6007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.256081850341335e-05,
+      "loss": 0.8091,
+      "step": 6008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2558656682159072e-05,
+      "loss": 0.9178,
+      "step": 6009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2556494732942274e-05,
+      "loss": 0.8958,
+      "step": 6010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2554332655871091e-05,
+      "loss": 0.9395,
+      "step": 6011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2552170451053645e-05,
+      "loss": 0.8658,
+      "step": 6012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2550008118598067e-05,
+      "loss": 0.9699,
+      "step": 6013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2547845658612508e-05,
+      "loss": 0.9601,
+      "step": 6014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2545683071205107e-05,
+      "loss": 0.9838,
+      "step": 6015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2543520356484029e-05,
+      "loss": 0.9988,
+      "step": 6016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2541357514557429e-05,
+      "loss": 0.9225,
+      "step": 6017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2539194545533469e-05,
+      "loss": 0.8961,
+      "step": 6018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2537031449520331e-05,
+      "loss": 1.0097,
+      "step": 6019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2534868226626191e-05,
+      "loss": 0.6482,
+      "step": 6020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2532704876959238e-05,
+      "loss": 1.0469,
+      "step": 6021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2530541400627663e-05,
+      "loss": 0.9971,
+      "step": 6022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2528377797739659e-05,
+      "loss": 0.8493,
+      "step": 6023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2526214068403442e-05,
+      "loss": 0.9285,
+      "step": 6024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2524050212727215e-05,
+      "loss": 1.0273,
+      "step": 6025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2521886230819202e-05,
+      "loss": 0.9459,
+      "step": 6026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.251972212278762e-05,
+      "loss": 0.9237,
+      "step": 6027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2517557888740708e-05,
+      "loss": 0.9875,
+      "step": 6028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2515393528786699e-05,
+      "loss": 0.9569,
+      "step": 6029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2513229043033829e-05,
+      "loss": 0.9539,
+      "step": 6030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2511064431590362e-05,
+      "loss": 0.8812,
+      "step": 6031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2508899694564545e-05,
+      "loss": 0.8524,
+      "step": 6032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2506734832064638e-05,
+      "loss": 0.9849,
+      "step": 6033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2504569844198913e-05,
+      "loss": 0.9804,
+      "step": 6034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2502404731075641e-05,
+      "loss": 0.8446,
+      "step": 6035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2500239492803107e-05,
+      "loss": 0.9098,
+      "step": 6036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2498074129489596e-05,
+      "loss": 0.8763,
+      "step": 6037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2495908641243405e-05,
+      "loss": 0.9264,
+      "step": 6038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2493743028172829e-05,
+      "loss": 0.8988,
+      "step": 6039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.249157729038617e-05,
+      "loss": 0.8651,
+      "step": 6040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2489411427991751e-05,
+      "loss": 0.9898,
+      "step": 6041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.248724544109788e-05,
+      "loss": 0.894,
+      "step": 6042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2485079329812887e-05,
+      "loss": 0.6483,
+      "step": 6043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2482913094245101e-05,
+      "loss": 1.0517,
+      "step": 6044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2480746734502858e-05,
+      "loss": 0.9718,
+      "step": 6045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2478580250694505e-05,
+      "loss": 0.6619,
+      "step": 6046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2476413642928389e-05,
+      "loss": 0.9888,
+      "step": 6047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2474246911312862e-05,
+      "loss": 0.8975,
+      "step": 6048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.247208005595629e-05,
+      "loss": 0.9281,
+      "step": 6049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2469913076967039e-05,
+      "loss": 1.0162,
+      "step": 6050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2467745974453486e-05,
+      "loss": 0.9596,
+      "step": 6051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.246557874852401e-05,
+      "loss": 0.8682,
+      "step": 6052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2463411399286995e-05,
+      "loss": 0.8506,
+      "step": 6053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2461243926850837e-05,
+      "loss": 0.9826,
+      "step": 6054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2459076331323929e-05,
+      "loss": 0.9721,
+      "step": 6055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2456908612814683e-05,
+      "loss": 1.0027,
+      "step": 6056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2454740771431503e-05,
+      "loss": 1.0385,
+      "step": 6057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2452572807282814e-05,
+      "loss": 0.9071,
+      "step": 6058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2450404720477036e-05,
+      "loss": 0.6213,
+      "step": 6059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2448236511122595e-05,
+      "loss": 0.9814,
+      "step": 6060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.244606817932793e-05,
+      "loss": 1.0165,
+      "step": 6061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2443899725201483e-05,
+      "loss": 0.84,
+      "step": 6062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2441731148851703e-05,
+      "loss": 0.9306,
+      "step": 6063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2439562450387042e-05,
+      "loss": 0.9811,
+      "step": 6064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2437393629915956e-05,
+      "loss": 0.9888,
+      "step": 6065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2435224687546921e-05,
+      "loss": 0.8311,
+      "step": 6066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2433055623388399e-05,
+      "loss": 0.8574,
+      "step": 6067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2430886437548877e-05,
+      "loss": 0.7651,
+      "step": 6068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2428717130136836e-05,
+      "loss": 1.0,
+      "step": 6069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2426547701260763e-05,
+      "loss": 0.8643,
+      "step": 6070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.242437815102916e-05,
+      "loss": 0.9785,
+      "step": 6071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2422208479550523e-05,
+      "loss": 1.1495,
+      "step": 6072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2420038686933368e-05,
+      "loss": 0.9738,
+      "step": 6073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2417868773286208e-05,
+      "loss": 0.9767,
+      "step": 6074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2415698738717564e-05,
+      "loss": 0.8727,
+      "step": 6075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2413528583335959e-05,
+      "loss": 0.898,
+      "step": 6076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2411358307249925e-05,
+      "loss": 1.0067,
+      "step": 6077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.240918791056801e-05,
+      "loss": 0.9889,
+      "step": 6078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2407017393398753e-05,
+      "loss": 0.8679,
+      "step": 6079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2404846755850704e-05,
+      "loss": 1.0362,
+      "step": 6080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2402675998032422e-05,
+      "loss": 0.9392,
+      "step": 6081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2400505120052471e-05,
+      "loss": 0.9618,
+      "step": 6082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.239833412201942e-05,
+      "loss": 0.8832,
+      "step": 6083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2396163004041838e-05,
+      "loss": 1.0044,
+      "step": 6084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2393991766228317e-05,
+      "loss": 0.8261,
+      "step": 6085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2391820408687436e-05,
+      "loss": 1.0517,
+      "step": 6086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2389648931527788e-05,
+      "loss": 0.9166,
+      "step": 6087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2387477334857975e-05,
+      "loss": 0.9554,
+      "step": 6088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2385305618786601e-05,
+      "loss": 1.0021,
+      "step": 6089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.238313378342228e-05,
+      "loss": 0.9141,
+      "step": 6090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2380961828873624e-05,
+      "loss": 0.9362,
+      "step": 6091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2378789755249256e-05,
+      "loss": 0.9768,
+      "step": 6092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.237661756265781e-05,
+      "loss": 0.6871,
+      "step": 6093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2374445251207917e-05,
+      "loss": 0.9247,
+      "step": 6094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2372272821008218e-05,
+      "loss": 1.0168,
+      "step": 6095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.237010027216736e-05,
+      "loss": 0.9147,
+      "step": 6096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2367927604793997e-05,
+      "loss": 0.955,
+      "step": 6097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2365754818996787e-05,
+      "loss": 1.0844,
+      "step": 6098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.236358191488439e-05,
+      "loss": 1.024,
+      "step": 6099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2361408892565485e-05,
+      "loss": 0.8487,
+      "step": 6100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2359235752148744e-05,
+      "loss": 0.9497,
+      "step": 6101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2357062493742847e-05,
+      "loss": 0.9718,
+      "step": 6102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2354889117456484e-05,
+      "loss": 0.8892,
+      "step": 6103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2352715623398349e-05,
+      "loss": 0.9469,
+      "step": 6104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2350542011677142e-05,
+      "loss": 1.0403,
+      "step": 6105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.234836828240157e-05,
+      "loss": 0.8968,
+      "step": 6106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2346194435680344e-05,
+      "loss": 0.8782,
+      "step": 6107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2344020471622175e-05,
+      "loss": 0.9792,
+      "step": 6108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.23418463903358e-05,
+      "loss": 1.1146,
+      "step": 6109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2339672191929937e-05,
+      "loss": 0.6301,
+      "step": 6110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2337497876513326e-05,
+      "loss": 0.9918,
+      "step": 6111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2335323444194706e-05,
+      "loss": 0.8736,
+      "step": 6112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2333148895082828e-05,
+      "loss": 0.9615,
+      "step": 6113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2330974229286437e-05,
+      "loss": 0.9215,
+      "step": 6114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2328799446914297e-05,
+      "loss": 0.9751,
+      "step": 6115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2326624548075173e-05,
+      "loss": 0.9174,
+      "step": 6116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2324449532877832e-05,
+      "loss": 0.9608,
+      "step": 6117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2322274401431051e-05,
+      "loss": 1.0479,
+      "step": 6118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2320099153843612e-05,
+      "loss": 0.9219,
+      "step": 6119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2317923790224307e-05,
+      "loss": 0.9343,
+      "step": 6120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2315748310681918e-05,
+      "loss": 0.9187,
+      "step": 6121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2313572715325257e-05,
+      "loss": 0.9104,
+      "step": 6122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2311397004263124e-05,
+      "loss": 0.8319,
+      "step": 6123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2309221177604325e-05,
+      "loss": 0.9504,
+      "step": 6124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2307045235457682e-05,
+      "loss": 0.9075,
+      "step": 6125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.230486917793202e-05,
+      "loss": 0.9369,
+      "step": 6126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2302693005136156e-05,
+      "loss": 0.9628,
+      "step": 6127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2300516717178938e-05,
+      "loss": 0.9413,
+      "step": 6128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2298340314169197e-05,
+      "loss": 0.9037,
+      "step": 6129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2296163796215778e-05,
+      "loss": 0.9926,
+      "step": 6130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2293987163427537e-05,
+      "loss": 0.8734,
+      "step": 6131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2291810415913328e-05,
+      "loss": 0.8784,
+      "step": 6132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2289633553782015e-05,
+      "loss": 0.9467,
+      "step": 6133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2287456577142463e-05,
+      "loss": 0.6277,
+      "step": 6134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2285279486103547e-05,
+      "loss": 0.9665,
+      "step": 6135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2283102280774149e-05,
+      "loss": 0.914,
+      "step": 6136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2280924961263157e-05,
+      "loss": 0.9191,
+      "step": 6137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2278747527679456e-05,
+      "loss": 0.9693,
+      "step": 6138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2276569980131944e-05,
+      "loss": 1.0893,
+      "step": 6139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2274392318729527e-05,
+      "loss": 0.8549,
+      "step": 6140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2272214543581113e-05,
+      "loss": 0.9724,
+      "step": 6141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2270036654795614e-05,
+      "loss": 0.956,
+      "step": 6142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2267858652481953e-05,
+      "loss": 0.9192,
+      "step": 6143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2265680536749048e-05,
+      "loss": 0.9352,
+      "step": 6144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2263502307705839e-05,
+      "loss": 0.866,
+      "step": 6145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2261323965461255e-05,
+      "loss": 1.0011,
+      "step": 6146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2259145510124246e-05,
+      "loss": 0.8716,
+      "step": 6147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2256966941803754e-05,
+      "loss": 0.6503,
+      "step": 6148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2254788260608735e-05,
+      "loss": 0.9612,
+      "step": 6149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2252609466648147e-05,
+      "loss": 0.916,
+      "step": 6150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2250430560030957e-05,
+      "loss": 0.8974,
+      "step": 6151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2248251540866134e-05,
+      "loss": 0.6824,
+      "step": 6152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2246072409262657e-05,
+      "loss": 0.9296,
+      "step": 6153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2243893165329504e-05,
+      "loss": 0.947,
+      "step": 6154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2241713809175664e-05,
+      "loss": 0.8897,
+      "step": 6155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.223953434091013e-05,
+      "loss": 0.934,
+      "step": 6156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2237354760641902e-05,
+      "loss": 0.9815,
+      "step": 6157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2235175068479984e-05,
+      "loss": 0.9972,
+      "step": 6158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2232995264533383e-05,
+      "loss": 0.8514,
+      "step": 6159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2230815348911119e-05,
+      "loss": 1.0026,
+      "step": 6160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2228635321722208e-05,
+      "loss": 0.6619,
+      "step": 6161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2226455183075683e-05,
+      "loss": 1.0331,
+      "step": 6162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2224274933080571e-05,
+      "loss": 0.9526,
+      "step": 6163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.222209457184591e-05,
+      "loss": 0.8237,
+      "step": 6164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2219914099480748e-05,
+      "loss": 0.9184,
+      "step": 6165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2217733516094126e-05,
+      "loss": 0.974,
+      "step": 6166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2215552821795107e-05,
+      "loss": 0.8912,
+      "step": 6167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2213372016692743e-05,
+      "loss": 0.9377,
+      "step": 6168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.221119110089611e-05,
+      "loss": 1.0187,
+      "step": 6169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.220901007451427e-05,
+      "loss": 0.94,
+      "step": 6170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2206828937656299e-05,
+      "loss": 0.8696,
+      "step": 6171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2204647690431285e-05,
+      "loss": 0.8943,
+      "step": 6172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2202466332948315e-05,
+      "loss": 0.9962,
+      "step": 6173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2200284865316476e-05,
+      "loss": 0.9963,
+      "step": 6174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2198103287644875e-05,
+      "loss": 0.8787,
+      "step": 6175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.219592160004261e-05,
+      "loss": 0.9276,
+      "step": 6176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2193739802618797e-05,
+      "loss": 0.9045,
+      "step": 6177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2191557895482542e-05,
+      "loss": 0.888,
+      "step": 6178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2189375878742975e-05,
+      "loss": 1.0042,
+      "step": 6179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2187193752509217e-05,
+      "loss": 0.9424,
+      "step": 6180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2185011516890397e-05,
+      "loss": 0.8377,
+      "step": 6181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.218282917199566e-05,
+      "loss": 1.0694,
+      "step": 6182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.218064671793414e-05,
+      "loss": 0.9029,
+      "step": 6183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2178464154814994e-05,
+      "loss": 0.9529,
+      "step": 6184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.217628148274737e-05,
+      "loss": 0.9193,
+      "step": 6185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2174098701840425e-05,
+      "loss": 1.0309,
+      "step": 6186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2171915812203326e-05,
+      "loss": 1.0252,
+      "step": 6187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2169732813945244e-05,
+      "loss": 0.8188,
+      "step": 6188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.216754970717535e-05,
+      "loss": 0.9713,
+      "step": 6189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2165366492002832e-05,
+      "loss": 0.9703,
+      "step": 6190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2163183168536869e-05,
+      "loss": 0.9123,
+      "step": 6191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2160999736886658e-05,
+      "loss": 1.0082,
+      "step": 6192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.215881619716139e-05,
+      "loss": 1.0098,
+      "step": 6193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2156632549470274e-05,
+      "loss": 0.9795,
+      "step": 6194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2154448793922512e-05,
+      "loss": 0.8992,
+      "step": 6195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.215226493062732e-05,
+      "loss": 0.9373,
+      "step": 6196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2150080959693916e-05,
+      "loss": 0.897,
+      "step": 6197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2147896881231523e-05,
+      "loss": 0.8656,
+      "step": 6198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2145712695349373e-05,
+      "loss": 0.996,
+      "step": 6199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2143528402156695e-05,
+      "loss": 0.9106,
+      "step": 6200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2141344001762734e-05,
+      "loss": 0.851,
+      "step": 6201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2139159494276734e-05,
+      "loss": 1.0412,
+      "step": 6202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2136974879807943e-05,
+      "loss": 0.9401,
+      "step": 6203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2134790158465623e-05,
+      "loss": 0.8355,
+      "step": 6204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2132605330359035e-05,
+      "loss": 0.9814,
+      "step": 6205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2130420395597437e-05,
+      "loss": 0.9064,
+      "step": 6206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2128235354290112e-05,
+      "loss": 0.9143,
+      "step": 6207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2126050206546328e-05,
+      "loss": 0.9899,
+      "step": 6208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2123864952475375e-05,
+      "loss": 0.9349,
+      "step": 6209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.212167959218654e-05,
+      "loss": 1.0059,
+      "step": 6210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2119494125789109e-05,
+      "loss": 1.0342,
+      "step": 6211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2117308553392391e-05,
+      "loss": 0.9523,
+      "step": 6212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2115122875105683e-05,
+      "loss": 0.895,
+      "step": 6213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2112937091038295e-05,
+      "loss": 1.0067,
+      "step": 6214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2110751201299544e-05,
+      "loss": 0.6339,
+      "step": 6215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.210856520599875e-05,
+      "loss": 0.9202,
+      "step": 6216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2106379105245239e-05,
+      "loss": 1.1066,
+      "step": 6217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2104192899148333e-05,
+      "loss": 0.9579,
+      "step": 6218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2102006587817379e-05,
+      "loss": 0.9601,
+      "step": 6219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2099820171361711e-05,
+      "loss": 0.9453,
+      "step": 6220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2097633649890677e-05,
+      "loss": 0.8962,
+      "step": 6221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2095447023513631e-05,
+      "loss": 0.6297,
+      "step": 6222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2093260292339926e-05,
+      "loss": 0.9937,
+      "step": 6223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2091073456478923e-05,
+      "loss": 0.9505,
+      "step": 6224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2088886516039992e-05,
+      "loss": 0.9308,
+      "step": 6225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2086699471132508e-05,
+      "loss": 0.9079,
+      "step": 6226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2084512321865844e-05,
+      "loss": 0.9131,
+      "step": 6227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2082325068349384e-05,
+      "loss": 0.9557,
+      "step": 6228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2080137710692514e-05,
+      "loss": 0.9527,
+      "step": 6229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.207795024900463e-05,
+      "loss": 0.9502,
+      "step": 6230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2075762683395133e-05,
+      "loss": 0.9341,
+      "step": 6231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2073575013973422e-05,
+      "loss": 0.9158,
+      "step": 6232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2071387240848906e-05,
+      "loss": 0.8568,
+      "step": 6233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2069199364131e-05,
+      "loss": 1.0384,
+      "step": 6234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2067011383929127e-05,
+      "loss": 0.968,
+      "step": 6235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2064823300352704e-05,
+      "loss": 0.949,
+      "step": 6236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2062635113511168e-05,
+      "loss": 1.0425,
+      "step": 6237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2060446823513949e-05,
+      "loss": 1.0198,
+      "step": 6238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2058258430470488e-05,
+      "loss": 0.9067,
+      "step": 6239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.205606993449023e-05,
+      "loss": 1.0365,
+      "step": 6240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2053881335682626e-05,
+      "loss": 0.9642,
+      "step": 6241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2051692634157133e-05,
+      "loss": 0.6088,
+      "step": 6242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2049503830023206e-05,
+      "loss": 0.9397,
+      "step": 6243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2047314923390317e-05,
+      "loss": 0.9646,
+      "step": 6244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.204512591436793e-05,
+      "loss": 0.942,
+      "step": 6245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2042936803065527e-05,
+      "loss": 0.9632,
+      "step": 6246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2040747589592584e-05,
+      "loss": 0.685,
+      "step": 6247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2038558274058591e-05,
+      "loss": 1.0018,
+      "step": 6248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2036368856573037e-05,
+      "loss": 0.8668,
+      "step": 6249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2034179337245417e-05,
+      "loss": 0.9508,
+      "step": 6250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.203198971618524e-05,
+      "loss": 0.9017,
+      "step": 6251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2029799993502003e-05,
+      "loss": 0.8892,
+      "step": 6252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2027610169305219e-05,
+      "loss": 0.9765,
+      "step": 6253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2025420243704411e-05,
+      "loss": 0.9532,
+      "step": 6254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2023230216809095e-05,
+      "loss": 0.8777,
+      "step": 6255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2021040088728798e-05,
+      "loss": 1.0861,
+      "step": 6256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2018849859573056e-05,
+      "loss": 0.93,
+      "step": 6257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.20166595294514e-05,
+      "loss": 0.9029,
+      "step": 6258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2014469098473377e-05,
+      "loss": 0.9512,
+      "step": 6259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.201227856674853e-05,
+      "loss": 0.9861,
+      "step": 6260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2010087934386415e-05,
+      "loss": 0.9954,
+      "step": 6261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2007897201496583e-05,
+      "loss": 0.8946,
+      "step": 6262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2005706368188606e-05,
+      "loss": 0.7835,
+      "step": 6263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2003515434572041e-05,
+      "loss": 0.9732,
+      "step": 6264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2001324400756462e-05,
+      "loss": 1.0807,
+      "step": 6265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.199913326685145e-05,
+      "loss": 0.9135,
+      "step": 6266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1996942032966587e-05,
+      "loss": 0.9812,
+      "step": 6267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1994750699211457e-05,
+      "loss": 1.0001,
+      "step": 6268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1992559265695656e-05,
+      "loss": 0.8724,
+      "step": 6269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1990367732528775e-05,
+      "loss": 0.8934,
+      "step": 6270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1988176099820421e-05,
+      "loss": 0.9274,
+      "step": 6271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1985984367680199e-05,
+      "loss": 0.9796,
+      "step": 6272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1983792536217724e-05,
+      "loss": 0.9228,
+      "step": 6273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.198160060554261e-05,
+      "loss": 0.9614,
+      "step": 6274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1979408575764477e-05,
+      "loss": 1.0137,
+      "step": 6275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1977216446992959e-05,
+      "loss": 1.0836,
+      "step": 6276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1975024219337681e-05,
+      "loss": 0.9649,
+      "step": 6277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1972831892908284e-05,
+      "loss": 1.0033,
+      "step": 6278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1970639467814406e-05,
+      "loss": 0.9287,
+      "step": 6279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1968446944165695e-05,
+      "loss": 0.9269,
+      "step": 6280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1966254322071805e-05,
+      "loss": 0.9912,
+      "step": 6281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1964061601642391e-05,
+      "loss": 0.8361,
+      "step": 6282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1961868782987111e-05,
+      "loss": 0.9578,
+      "step": 6283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1959675866215641e-05,
+      "loss": 1.107,
+      "step": 6284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.195748285143764e-05,
+      "loss": 0.9555,
+      "step": 6285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1955289738762796e-05,
+      "loss": 0.8456,
+      "step": 6286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.195309652830078e-05,
+      "loss": 0.8738,
+      "step": 6287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1950903220161286e-05,
+      "loss": 1.0042,
+      "step": 6288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1948709814453998e-05,
+      "loss": 0.8646,
+      "step": 6289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1946516311288613e-05,
+      "loss": 0.8931,
+      "step": 6290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1944322710774838e-05,
+      "loss": 0.9078,
+      "step": 6291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.194212901302237e-05,
+      "loss": 0.9203,
+      "step": 6292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1939935218140928e-05,
+      "loss": 0.9899,
+      "step": 6293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1937741326240218e-05,
+      "loss": 0.9263,
+      "step": 6294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1935547337429965e-05,
+      "loss": 1.0617,
+      "step": 6295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1933353251819893e-05,
+      "loss": 0.9377,
+      "step": 6296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1931159069519733e-05,
+      "loss": 0.9293,
+      "step": 6297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1928964790639219e-05,
+      "loss": 1.0449,
+      "step": 6298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1926770415288091e-05,
+      "loss": 0.9958,
+      "step": 6299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1924575943576091e-05,
+      "loss": 0.9413,
+      "step": 6300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.192238137561297e-05,
+      "loss": 0.9407,
+      "step": 6301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1920186711508479e-05,
+      "loss": 0.8583,
+      "step": 6302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.191799195137238e-05,
+      "loss": 0.9716,
+      "step": 6303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.191579709531444e-05,
+      "loss": 1.0424,
+      "step": 6304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1913602143444416e-05,
+      "loss": 1.0477,
+      "step": 6305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.191140709587209e-05,
+      "loss": 0.9452,
+      "step": 6306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1909211952707236e-05,
+      "loss": 1.0035,
+      "step": 6307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.190701671405964e-05,
+      "loss": 0.9223,
+      "step": 6308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1904821380039086e-05,
+      "loss": 0.9645,
+      "step": 6309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1902625950755371e-05,
+      "loss": 0.9913,
+      "step": 6310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1900430426318285e-05,
+      "loss": 1.0117,
+      "step": 6311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1898234806837633e-05,
+      "loss": 0.969,
+      "step": 6312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1896039092423223e-05,
+      "loss": 0.9707,
+      "step": 6313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1893843283184865e-05,
+      "loss": 0.8828,
+      "step": 6314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1891647379232373e-05,
+      "loss": 0.9206,
+      "step": 6315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1889451380675574e-05,
+      "loss": 0.9844,
+      "step": 6316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1887255287624283e-05,
+      "loss": 1.0031,
+      "step": 6317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1885059100188341e-05,
+      "loss": 0.9996,
+      "step": 6318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1882862818477575e-05,
+      "loss": 0.9219,
+      "step": 6319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1880666442601828e-05,
+      "loss": 0.8538,
+      "step": 6320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1878469972670946e-05,
+      "loss": 0.9483,
+      "step": 6321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1876273408794771e-05,
+      "loss": 0.9805,
+      "step": 6322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1874076751083166e-05,
+      "loss": 1.0329,
+      "step": 6323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1871879999645982e-05,
+      "loss": 0.857,
+      "step": 6324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1869683154593085e-05,
+      "loss": 0.9199,
+      "step": 6325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1867486216034345e-05,
+      "loss": 0.9681,
+      "step": 6326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1865289184079627e-05,
+      "loss": 0.6415,
+      "step": 6327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1863092058838816e-05,
+      "loss": 0.865,
+      "step": 6328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1860894840421791e-05,
+      "loss": 1.0186,
+      "step": 6329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1858697528938437e-05,
+      "loss": 0.907,
+      "step": 6330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1856500124498647e-05,
+      "loss": 0.9612,
+      "step": 6331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1854302627212315e-05,
+      "loss": 0.8938,
+      "step": 6332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1852105037189342e-05,
+      "loss": 0.9248,
+      "step": 6333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1849907354539634e-05,
+      "loss": 1.0723,
+      "step": 6334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.18477095793731e-05,
+      "loss": 0.9524,
+      "step": 6335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1845511711799657e-05,
+      "loss": 0.9264,
+      "step": 6336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1843313751929217e-05,
+      "loss": 0.9128,
+      "step": 6337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.184111569987171e-05,
+      "loss": 0.9675,
+      "step": 6338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1838917555737059e-05,
+      "loss": 0.8836,
+      "step": 6339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1836719319635201e-05,
+      "loss": 0.9634,
+      "step": 6340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.183452099167607e-05,
+      "loss": 0.9991,
+      "step": 6341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1832322571969613e-05,
+      "loss": 0.9005,
+      "step": 6342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.183012406062577e-05,
+      "loss": 0.9955,
+      "step": 6343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1827925457754495e-05,
+      "loss": 0.9439,
+      "step": 6344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1825726763465744e-05,
+      "loss": 0.9563,
+      "step": 6345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1823527977869478e-05,
+      "loss": 0.9148,
+      "step": 6346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.182132910107566e-05,
+      "loss": 0.9049,
+      "step": 6347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1819130133194261e-05,
+      "loss": 0.953,
+      "step": 6348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1816931074335252e-05,
+      "loss": 0.7033,
+      "step": 6349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1814731924608616e-05,
+      "loss": 0.9455,
+      "step": 6350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1812532684124332e-05,
+      "loss": 0.9586,
+      "step": 6351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1810333352992386e-05,
+      "loss": 0.9994,
+      "step": 6352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1808133931322778e-05,
+      "loss": 0.9135,
+      "step": 6353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1805934419225497e-05,
+      "loss": 0.9756,
+      "step": 6354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1803734816810549e-05,
+      "loss": 0.9281,
+      "step": 6355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1801535124187935e-05,
+      "loss": 0.919,
+      "step": 6356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1799335341467668e-05,
+      "loss": 0.9671,
+      "step": 6357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1797135468759764e-05,
+      "loss": 0.9271,
+      "step": 6358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1794935506174235e-05,
+      "loss": 0.8755,
+      "step": 6359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1792735453821114e-05,
+      "loss": 0.9703,
+      "step": 6360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1790535311810425e-05,
+      "loss": 0.9738,
+      "step": 6361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.17883350802522e-05,
+      "loss": 0.9406,
+      "step": 6362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1786134759256478e-05,
+      "loss": 0.9279,
+      "step": 6363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1783934348933297e-05,
+      "loss": 1.0597,
+      "step": 6364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.178173384939271e-05,
+      "loss": 0.6115,
+      "step": 6365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1779533260744758e-05,
+      "loss": 0.9935,
+      "step": 6366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1777332583099503e-05,
+      "loss": 0.9102,
+      "step": 6367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1775131816567003e-05,
+      "loss": 0.6406,
+      "step": 6368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1772930961257318e-05,
+      "loss": 0.9242,
+      "step": 6369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1770730017280525e-05,
+      "loss": 0.8755,
+      "step": 6370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1768528984746685e-05,
+      "loss": 0.9511,
+      "step": 6371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1766327863765885e-05,
+      "loss": 0.8911,
+      "step": 6372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1764126654448205e-05,
+      "loss": 0.8636,
+      "step": 6373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1761925356903726e-05,
+      "loss": 0.9098,
+      "step": 6374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1759723971242538e-05,
+      "loss": 0.9715,
+      "step": 6375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1757522497574745e-05,
+      "loss": 0.9781,
+      "step": 6376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1755320936010435e-05,
+      "loss": 0.8785,
+      "step": 6377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1753119286659722e-05,
+      "loss": 0.9701,
+      "step": 6378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1750917549632706e-05,
+      "loss": 1.0397,
+      "step": 6379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1748715725039504e-05,
+      "loss": 0.9553,
+      "step": 6380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1746513812990228e-05,
+      "loss": 1.0015,
+      "step": 6381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1744311813595006e-05,
+      "loss": 0.9838,
+      "step": 6382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1742109726963959e-05,
+      "loss": 0.9368,
+      "step": 6383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1739907553207216e-05,
+      "loss": 0.9197,
+      "step": 6384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1737705292434916e-05,
+      "loss": 0.8218,
+      "step": 6385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1735502944757192e-05,
+      "loss": 1.021,
+      "step": 6386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1733300510284191e-05,
+      "loss": 0.8858,
+      "step": 6387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1731097989126057e-05,
+      "loss": 0.8846,
+      "step": 6388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1728895381392944e-05,
+      "loss": 0.8663,
+      "step": 6389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1726692687195007e-05,
+      "loss": 0.9452,
+      "step": 6390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1724489906642404e-05,
+      "loss": 0.9589,
+      "step": 6391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1722287039845308e-05,
+      "loss": 0.8575,
+      "step": 6392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1720084086913879e-05,
+      "loss": 0.95,
+      "step": 6393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1717881047958292e-05,
+      "loss": 0.9332,
+      "step": 6394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.171567792308873e-05,
+      "loss": 0.9189,
+      "step": 6395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1713474712415364e-05,
+      "loss": 1.1035,
+      "step": 6396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.171127141604839e-05,
+      "loss": 0.9266,
+      "step": 6397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1709068034097996e-05,
+      "loss": 0.9938,
+      "step": 6398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1706864566674374e-05,
+      "loss": 0.9801,
+      "step": 6399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1704661013887725e-05,
+      "loss": 0.8603,
+      "step": 6400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.170245737584825e-05,
+      "loss": 0.9076,
+      "step": 6401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1700253652666162e-05,
+      "loss": 0.9191,
+      "step": 6402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1698049844451665e-05,
+      "loss": 0.6293,
+      "step": 6403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1695845951314982e-05,
+      "loss": 0.9402,
+      "step": 6404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.169364197336633e-05,
+      "loss": 0.9176,
+      "step": 6405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1691437910715928e-05,
+      "loss": 0.9306,
+      "step": 6406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1689233763474018e-05,
+      "loss": 0.8948,
+      "step": 6407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1687029531750822e-05,
+      "loss": 0.8267,
+      "step": 6408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.168482521565658e-05,
+      "loss": 1.0687,
+      "step": 6409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1682620815301533e-05,
+      "loss": 0.9793,
+      "step": 6410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1680416330795927e-05,
+      "loss": 1.0452,
+      "step": 6411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1678211762250014e-05,
+      "loss": 1.0266,
+      "step": 6412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1676007109774044e-05,
+      "loss": 0.8483,
+      "step": 6413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.167380237347828e-05,
+      "loss": 0.8853,
+      "step": 6414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1671597553472982e-05,
+      "loss": 0.988,
+      "step": 6415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1669392649868414e-05,
+      "loss": 0.9402,
+      "step": 6416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.166718766277485e-05,
+      "loss": 0.6579,
+      "step": 6417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1664982592302562e-05,
+      "loss": 0.5978,
+      "step": 6418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1662777438561834e-05,
+      "loss": 0.9074,
+      "step": 6419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1660572201662947e-05,
+      "loss": 0.9325,
+      "step": 6420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1658366881716185e-05,
+      "loss": 1.0232,
+      "step": 6421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1656161478831842e-05,
+      "loss": 1.1003,
+      "step": 6422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1653955993120214e-05,
+      "loss": 0.964,
+      "step": 6423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1651750424691604e-05,
+      "loss": 0.8989,
+      "step": 6424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1649544773656312e-05,
+      "loss": 0.6466,
+      "step": 6425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1647339040124643e-05,
+      "loss": 1.0134,
+      "step": 6426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1645133224206919e-05,
+      "loss": 1.0323,
+      "step": 6427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1642927326013446e-05,
+      "loss": 0.8002,
+      "step": 6428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1640721345654553e-05,
+      "loss": 0.857,
+      "step": 6429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1638515283240562e-05,
+      "loss": 1.0316,
+      "step": 6430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1636309138881796e-05,
+      "loss": 1.0345,
+      "step": 6431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1634102912688593e-05,
+      "loss": 1.0291,
+      "step": 6432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.163189660477129e-05,
+      "loss": 0.8897,
+      "step": 6433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.162969021524023e-05,
+      "loss": 0.6342,
+      "step": 6434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1627483744205751e-05,
+      "loss": 1.0158,
+      "step": 6435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1625277191778211e-05,
+      "loss": 0.9185,
+      "step": 6436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1623070558067956e-05,
+      "loss": 0.9311,
+      "step": 6437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1620863843185344e-05,
+      "loss": 1.0046,
+      "step": 6438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.161865704724074e-05,
+      "loss": 0.8002,
+      "step": 6439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1616450170344509e-05,
+      "loss": 0.9686,
+      "step": 6440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1614243212607017e-05,
+      "loss": 0.8659,
+      "step": 6441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.161203617413864e-05,
+      "loss": 0.9347,
+      "step": 6442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1609829055049757e-05,
+      "loss": 0.9274,
+      "step": 6443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1607621855450748e-05,
+      "loss": 0.9221,
+      "step": 6444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1605414575451996e-05,
+      "loss": 0.8908,
+      "step": 6445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1603207215163894e-05,
+      "loss": 0.9013,
+      "step": 6446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1600999774696835e-05,
+      "loss": 0.9776,
+      "step": 6447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1598792254161216e-05,
+      "loss": 0.8826,
+      "step": 6448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1596584653667438e-05,
+      "loss": 0.9584,
+      "step": 6449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1594376973325907e-05,
+      "loss": 0.9971,
+      "step": 6450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1592169213247034e-05,
+      "loss": 0.8889,
+      "step": 6451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1589961373541232e-05,
+      "loss": 0.9194,
+      "step": 6452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1587753454318916e-05,
+      "loss": 1.0017,
+      "step": 6453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.158554545569051e-05,
+      "loss": 0.8997,
+      "step": 6454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1583337377766441e-05,
+      "loss": 0.9189,
+      "step": 6455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1581129220657137e-05,
+      "loss": 1.0573,
+      "step": 6456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1578920984473032e-05,
+      "loss": 1.0454,
+      "step": 6457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1576712669324559e-05,
+      "loss": 1.0221,
+      "step": 6458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1574504275322168e-05,
+      "loss": 0.9609,
+      "step": 6459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1572295802576295e-05,
+      "loss": 0.655,
+      "step": 6460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1570087251197398e-05,
+      "loss": 1.0032,
+      "step": 6461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1567878621295922e-05,
+      "loss": 0.9865,
+      "step": 6462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1565669912982331e-05,
+      "loss": 0.9082,
+      "step": 6463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1563461126367082e-05,
+      "loss": 0.9867,
+      "step": 6464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1561252261560641e-05,
+      "loss": 0.8578,
+      "step": 6465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1559043318673475e-05,
+      "loss": 1.0159,
+      "step": 6466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1556834297816061e-05,
+      "loss": 0.8264,
+      "step": 6467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1554625199098872e-05,
+      "loss": 0.8682,
+      "step": 6468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1552416022632392e-05,
+      "loss": 0.9651,
+      "step": 6469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.15502067685271e-05,
+      "loss": 0.8688,
+      "step": 6470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1547997436893487e-05,
+      "loss": 0.9632,
+      "step": 6471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.154578802784205e-05,
+      "loss": 0.9072,
+      "step": 6472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1543578541483278e-05,
+      "loss": 1.0028,
+      "step": 6473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1541368977927676e-05,
+      "loss": 1.0208,
+      "step": 6474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1539159337285744e-05,
+      "loss": 0.7972,
+      "step": 6475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1536949619667991e-05,
+      "loss": 0.9867,
+      "step": 6476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1534739825184933e-05,
+      "loss": 0.6844,
+      "step": 6477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1532529953947077e-05,
+      "loss": 0.9224,
+      "step": 6478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1530320006064947e-05,
+      "loss": 0.775,
+      "step": 6479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1528109981649065e-05,
+      "loss": 1.0981,
+      "step": 6480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1525899880809961e-05,
+      "loss": 0.9491,
+      "step": 6481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1523689703658158e-05,
+      "loss": 0.9523,
+      "step": 6482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1521479450304201e-05,
+      "loss": 1.0312,
+      "step": 6483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1519269120858618e-05,
+      "loss": 0.8344,
+      "step": 6484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1517058715431958e-05,
+      "loss": 0.9321,
+      "step": 6485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1514848234134765e-05,
+      "loss": 0.9938,
+      "step": 6486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1512637677077589e-05,
+      "loss": 1.043,
+      "step": 6487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1510427044370982e-05,
+      "loss": 0.9431,
+      "step": 6488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1508216336125502e-05,
+      "loss": 0.9698,
+      "step": 6489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1506005552451712e-05,
+      "loss": 0.9015,
+      "step": 6490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1503794693460174e-05,
+      "loss": 0.9844,
+      "step": 6491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1501583759261459e-05,
+      "loss": 1.0129,
+      "step": 6492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1499372749966135e-05,
+      "loss": 0.9118,
+      "step": 6493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1497161665684785e-05,
+      "loss": 0.8468,
+      "step": 6494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1494950506527982e-05,
+      "loss": 0.903,
+      "step": 6495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1492739272606315e-05,
+      "loss": 1.0212,
+      "step": 6496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1490527964030368e-05,
+      "loss": 0.9952,
+      "step": 6497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1488316580910732e-05,
+      "loss": 1.0132,
+      "step": 6498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1486105123358004e-05,
+      "loss": 0.8749,
+      "step": 6499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.148389359148278e-05,
+      "loss": 0.9195,
+      "step": 6500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1481681985395666e-05,
+      "loss": 0.9881,
+      "step": 6501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1479470305207266e-05,
+      "loss": 0.8712,
+      "step": 6502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1477258551028184e-05,
+      "loss": 0.9507,
+      "step": 6503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1475046722969045e-05,
+      "loss": 0.9329,
+      "step": 6504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1472834821140457e-05,
+      "loss": 1.0115,
+      "step": 6505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1470622845653046e-05,
+      "loss": 0.9001,
+      "step": 6506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1468410796617429e-05,
+      "loss": 0.8866,
+      "step": 6507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1466198674144243e-05,
+      "loss": 0.6739,
+      "step": 6508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1463986478344116e-05,
+      "loss": 0.8251,
+      "step": 6509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.146177420932768e-05,
+      "loss": 0.9276,
+      "step": 6510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1459561867205582e-05,
+      "loss": 0.975,
+      "step": 6511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1457349452088455e-05,
+      "loss": 0.9586,
+      "step": 6512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1455136964086954e-05,
+      "loss": 0.9414,
+      "step": 6513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1452924403311725e-05,
+      "loss": 0.955,
+      "step": 6514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1450711769873422e-05,
+      "loss": 0.9975,
+      "step": 6515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1448499063882701e-05,
+      "loss": 1.063,
+      "step": 6516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1446286285450226e-05,
+      "loss": 0.99,
+      "step": 6517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.144407343468666e-05,
+      "loss": 0.968,
+      "step": 6518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1441860511702673e-05,
+      "loss": 1.0128,
+      "step": 6519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1439647516608933e-05,
+      "loss": 0.9362,
+      "step": 6520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.143743444951612e-05,
+      "loss": 0.8951,
+      "step": 6521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.143522131053491e-05,
+      "loss": 0.9148,
+      "step": 6522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1433008099775986e-05,
+      "loss": 0.8624,
+      "step": 6523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1430794817350036e-05,
+      "loss": 0.669,
+      "step": 6524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1428581463367745e-05,
+      "loss": 1.0067,
+      "step": 6525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1426368037939812e-05,
+      "loss": 0.9248,
+      "step": 6526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.142415454117693e-05,
+      "loss": 0.8537,
+      "step": 6527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1421940973189805e-05,
+      "loss": 0.9639,
+      "step": 6528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1419727334089133e-05,
+      "loss": 0.9948,
+      "step": 6529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1417513623985627e-05,
+      "loss": 0.9578,
+      "step": 6530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.141529984299e-05,
+      "loss": 0.9795,
+      "step": 6531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1413085991212957e-05,
+      "loss": 1.0092,
+      "step": 6532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.141087206876523e-05,
+      "loss": 0.9921,
+      "step": 6533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.140865807575753e-05,
+      "loss": 0.8702,
+      "step": 6534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1406444012300585e-05,
+      "loss": 0.8702,
+      "step": 6535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1404229878505132e-05,
+      "loss": 0.98,
+      "step": 6536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1402015674481888e-05,
+      "loss": 0.8503,
+      "step": 6537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1399801400341603e-05,
+      "loss": 0.8984,
+      "step": 6538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.139758705619501e-05,
+      "loss": 0.9592,
+      "step": 6539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.139537264215285e-05,
+      "loss": 1.0252,
+      "step": 6540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1393158158325874e-05,
+      "loss": 0.8527,
+      "step": 6541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1390943604824827e-05,
+      "loss": 0.9536,
+      "step": 6542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1388728981760468e-05,
+      "loss": 0.8689,
+      "step": 6543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.138651428924355e-05,
+      "loss": 0.6477,
+      "step": 6544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1384299527384837e-05,
+      "loss": 0.8339,
+      "step": 6545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1382084696295088e-05,
+      "loss": 0.9847,
+      "step": 6546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.137986979608507e-05,
+      "loss": 0.938,
+      "step": 6547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1377654826865559e-05,
+      "loss": 0.9943,
+      "step": 6548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1375439788747327e-05,
+      "loss": 0.9599,
+      "step": 6549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1373224681841148e-05,
+      "loss": 0.6083,
+      "step": 6550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.137100950625781e-05,
+      "loss": 0.9561,
+      "step": 6551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1368794262108087e-05,
+      "loss": 0.629,
+      "step": 6552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.136657894950278e-05,
+      "loss": 0.8876,
+      "step": 6553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.136436356855267e-05,
+      "loss": 0.9586,
+      "step": 6554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1362148119368557e-05,
+      "loss": 0.9279,
+      "step": 6555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1359932602061236e-05,
+      "loss": 0.9209,
+      "step": 6556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.135771701674151e-05,
+      "loss": 0.956,
+      "step": 6557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1355501363520186e-05,
+      "loss": 0.9241,
+      "step": 6558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1353285642508068e-05,
+      "loss": 0.8864,
+      "step": 6559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1351069853815973e-05,
+      "loss": 0.9725,
+      "step": 6560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1348853997554712e-05,
+      "loss": 0.9004,
+      "step": 6561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1346638073835103e-05,
+      "loss": 0.9155,
+      "step": 6562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.134442208276797e-05,
+      "loss": 0.9088,
+      "step": 6563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1342206024464137e-05,
+      "loss": 0.9901,
+      "step": 6564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1339989899034438e-05,
+      "loss": 0.888,
+      "step": 6565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1337773706589699e-05,
+      "loss": 0.8817,
+      "step": 6566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1335557447240755e-05,
+      "loss": 1.0612,
+      "step": 6567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1333341121098446e-05,
+      "loss": 1.056,
+      "step": 6568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1331124728273617e-05,
+      "loss": 0.9221,
+      "step": 6569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.132890826887711e-05,
+      "loss": 0.9242,
+      "step": 6570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1326691743019778e-05,
+      "loss": 0.9292,
+      "step": 6571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1324475150812464e-05,
+      "loss": 1.0093,
+      "step": 6572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1322258492366032e-05,
+      "loss": 0.9121,
+      "step": 6573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1320041767791337e-05,
+      "loss": 0.6796,
+      "step": 6574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.131782497719924e-05,
+      "loss": 0.8358,
+      "step": 6575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.131560812070061e-05,
+      "loss": 0.8201,
+      "step": 6576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1313391198406312e-05,
+      "loss": 0.8925,
+      "step": 6577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1311174210427221e-05,
+      "loss": 0.9177,
+      "step": 6578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1308957156874205e-05,
+      "loss": 0.6537,
+      "step": 6579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1306740037858152e-05,
+      "loss": 0.9633,
+      "step": 6580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130452285348994e-05,
+      "loss": 0.915,
+      "step": 6581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130230560388045e-05,
+      "loss": 0.8461,
+      "step": 6582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1300088289140576e-05,
+      "loss": 1.0603,
+      "step": 6583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1297870909381205e-05,
+      "loss": 1.0404,
+      "step": 6584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1295653464713236e-05,
+      "loss": 1.0386,
+      "step": 6585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1293435955247564e-05,
+      "loss": 0.9457,
+      "step": 6586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.129121838109509e-05,
+      "loss": 0.952,
+      "step": 6587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128900074236672e-05,
+      "loss": 0.9376,
+      "step": 6588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128678303917336e-05,
+      "loss": 0.9514,
+      "step": 6589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1284565271625922e-05,
+      "loss": 1.028,
+      "step": 6590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128234743983532e-05,
+      "loss": 0.9669,
+      "step": 6591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128012954391247e-05,
+      "loss": 0.9116,
+      "step": 6592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1277911583968298e-05,
+      "loss": 0.7057,
+      "step": 6593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1275693560113717e-05,
+      "loss": 0.8869,
+      "step": 6594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1273475472459663e-05,
+      "loss": 0.9872,
+      "step": 6595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1271257321117064e-05,
+      "loss": 0.8161,
+      "step": 6596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.126903910619685e-05,
+      "loss": 1.0127,
+      "step": 6597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1266820827809965e-05,
+      "loss": 0.9472,
+      "step": 6598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1264602486067342e-05,
+      "loss": 0.9338,
+      "step": 6599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1262384081079925e-05,
+      "loss": 1.0047,
+      "step": 6600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.126016561295866e-05,
+      "loss": 0.9876,
+      "step": 6601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1257947081814497e-05,
+      "loss": 0.9931,
+      "step": 6602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1255728487758391e-05,
+      "loss": 0.8221,
+      "step": 6603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.125350983090129e-05,
+      "loss": 0.8931,
+      "step": 6604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.125129111135416e-05,
+      "loss": 0.9953,
+      "step": 6605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.124907232922796e-05,
+      "loss": 0.9801,
+      "step": 6606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1246853484633655e-05,
+      "loss": 0.9078,
+      "step": 6607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1244634577682212e-05,
+      "loss": 0.9769,
+      "step": 6608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.12424156084846e-05,
+      "loss": 0.9324,
+      "step": 6609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1240196577151798e-05,
+      "loss": 0.9182,
+      "step": 6610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1237977483794779e-05,
+      "loss": 0.9868,
+      "step": 6611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1235758328524529e-05,
+      "loss": 0.8297,
+      "step": 6612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1233539111452028e-05,
+      "loss": 0.9476,
+      "step": 6613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.123131983268826e-05,
+      "loss": 0.9329,
+      "step": 6614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1229100492344222e-05,
+      "loss": 1.0364,
+      "step": 6615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1226881090530899e-05,
+      "loss": 0.9573,
+      "step": 6616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.122466162735929e-05,
+      "loss": 0.9739,
+      "step": 6617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1222442102940396e-05,
+      "loss": 0.9496,
+      "step": 6618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1220222517385214e-05,
+      "loss": 1.0057,
+      "step": 6619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1218002870804755e-05,
+      "loss": 1.0448,
+      "step": 6620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.121578316331002e-05,
+      "loss": 0.6459,
+      "step": 6621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1213563395012025e-05,
+      "loss": 0.8566,
+      "step": 6622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1211343566021784e-05,
+      "loss": 0.6525,
+      "step": 6623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1209123676450316e-05,
+      "loss": 0.9178,
+      "step": 6624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1206903726408638e-05,
+      "loss": 0.9496,
+      "step": 6625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.120468371600777e-05,
+      "loss": 1.0015,
+      "step": 6626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1202463645358745e-05,
+      "loss": 0.6309,
+      "step": 6627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.120024351457259e-05,
+      "loss": 0.9998,
+      "step": 6628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1198023323760334e-05,
+      "loss": 0.9564,
+      "step": 6629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1195803073033017e-05,
+      "loss": 0.9255,
+      "step": 6630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1193582762501676e-05,
+      "loss": 0.98,
+      "step": 6631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1191362392277352e-05,
+      "loss": 0.852,
+      "step": 6632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1189141962471088e-05,
+      "loss": 0.8092,
+      "step": 6633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1186921473193933e-05,
+      "loss": 1.0087,
+      "step": 6634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1184700924556938e-05,
+      "loss": 0.9036,
+      "step": 6635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1182480316671149e-05,
+      "loss": 0.9471,
+      "step": 6636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1180259649647631e-05,
+      "loss": 0.9305,
+      "step": 6637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.117803892359744e-05,
+      "loss": 0.9347,
+      "step": 6638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1175818138631638e-05,
+      "loss": 0.9111,
+      "step": 6639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.117359729486129e-05,
+      "loss": 0.9914,
+      "step": 6640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1171376392397461e-05,
+      "loss": 0.9123,
+      "step": 6641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1169155431351229e-05,
+      "loss": 0.9162,
+      "step": 6642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1166934411833663e-05,
+      "loss": 0.9546,
+      "step": 6643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1164713333955839e-05,
+      "loss": 0.9681,
+      "step": 6644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1162492197828838e-05,
+      "loss": 0.9998,
+      "step": 6645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.116027100356374e-05,
+      "loss": 0.9812,
+      "step": 6646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1158049751271639e-05,
+      "loss": 0.9757,
+      "step": 6647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1155828441063612e-05,
+      "loss": 0.8974,
+      "step": 6648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1153607073050758e-05,
+      "loss": 0.8747,
+      "step": 6649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1151385647344171e-05,
+      "loss": 0.9581,
+      "step": 6650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1149164164054942e-05,
+      "loss": 0.883,
+      "step": 6651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1146942623294176e-05,
+      "loss": 0.8444,
+      "step": 6652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1144721025172975e-05,
+      "loss": 0.8387,
+      "step": 6653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1142499369802444e-05,
+      "loss": 0.9474,
+      "step": 6654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1140277657293694e-05,
+      "loss": 1.0052,
+      "step": 6655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.113805588775783e-05,
+      "loss": 0.9715,
+      "step": 6656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1135834061305972e-05,
+      "loss": 1.0272,
+      "step": 6657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1133612178049235e-05,
+      "loss": 0.9334,
+      "step": 6658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1131390238098743e-05,
+      "loss": 0.9567,
+      "step": 6659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1129168241565615e-05,
+      "loss": 0.9266,
+      "step": 6660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1126946188560973e-05,
+      "loss": 0.8661,
+      "step": 6661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1124724079195952e-05,
+      "loss": 0.8861,
+      "step": 6662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.112250191358168e-05,
+      "loss": 0.9442,
+      "step": 6663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1120279691829293e-05,
+      "loss": 1.0084,
+      "step": 6664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1118057414049925e-05,
+      "loss": 0.8639,
+      "step": 6665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1115835080354718e-05,
+      "loss": 0.9164,
+      "step": 6666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1113612690854813e-05,
+      "loss": 0.9466,
+      "step": 6667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1111390245661355e-05,
+      "loss": 0.91,
+      "step": 6668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1109167744885494e-05,
+      "loss": 1.0175,
+      "step": 6669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1106945188638378e-05,
+      "loss": 0.8915,
+      "step": 6670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1104722577031164e-05,
+      "loss": 0.9694,
+      "step": 6671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1102499910175008e-05,
+      "loss": 0.8957,
+      "step": 6672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1100277188181063e-05,
+      "loss": 1.0114,
+      "step": 6673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1098054411160501e-05,
+      "loss": 0.9329,
+      "step": 6674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1095831579224478e-05,
+      "loss": 0.9516,
+      "step": 6675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1093608692484166e-05,
+      "loss": 0.9186,
+      "step": 6676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1091385751050735e-05,
+      "loss": 0.9482,
+      "step": 6677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1089162755035354e-05,
+      "loss": 0.8964,
+      "step": 6678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1086939704549204e-05,
+      "loss": 0.9397,
+      "step": 6679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.108471659970346e-05,
+      "loss": 0.8906,
+      "step": 6680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1082493440609304e-05,
+      "loss": 0.8584,
+      "step": 6681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1080270227377918e-05,
+      "loss": 0.9946,
+      "step": 6682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.107804696012049e-05,
+      "loss": 0.9867,
+      "step": 6683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1075823638948209e-05,
+      "loss": 0.9043,
+      "step": 6684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1073600263972267e-05,
+      "loss": 0.909,
+      "step": 6685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1071376835303858e-05,
+      "loss": 0.8313,
+      "step": 6686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.106915335305418e-05,
+      "loss": 0.672,
+      "step": 6687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.106692981733443e-05,
+      "loss": 0.9922,
+      "step": 6688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1064706228255817e-05,
+      "loss": 0.9048,
+      "step": 6689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1062482585929541e-05,
+      "loss": 0.9203,
+      "step": 6690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1060258890466811e-05,
+      "loss": 0.9393,
+      "step": 6691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1058035141978838e-05,
+      "loss": 0.9556,
+      "step": 6692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1055811340576834e-05,
+      "loss": 0.6539,
+      "step": 6693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1053587486372017e-05,
+      "loss": 1.0477,
+      "step": 6694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1051363579475603e-05,
+      "loss": 0.8991,
+      "step": 6695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1049139619998816e-05,
+      "loss": 0.9545,
+      "step": 6696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.104691560805288e-05,
+      "loss": 0.8988,
+      "step": 6697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1044691543749016e-05,
+      "loss": 0.9283,
+      "step": 6698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1042467427198458e-05,
+      "loss": 0.9825,
+      "step": 6699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1040243258512435e-05,
+      "loss": 0.787,
+      "step": 6700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1038019037802186e-05,
+      "loss": 0.879,
+      "step": 6701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1035794765178941e-05,
+      "loss": 0.8607,
+      "step": 6702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1033570440753945e-05,
+      "loss": 1.0596,
+      "step": 6703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1031346064638436e-05,
+      "loss": 0.9018,
+      "step": 6704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.102912163694366e-05,
+      "loss": 0.6559,
+      "step": 6705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1026897157780867e-05,
+      "loss": 0.956,
+      "step": 6706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1024672627261304e-05,
+      "loss": 0.9067,
+      "step": 6707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1022448045496224e-05,
+      "loss": 1.0692,
+      "step": 6708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.102022341259688e-05,
+      "loss": 0.92,
+      "step": 6709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1017998728674531e-05,
+      "loss": 1.0309,
+      "step": 6710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.101577399384044e-05,
+      "loss": 0.879,
+      "step": 6711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1013549208205865e-05,
+      "loss": 0.9285,
+      "step": 6712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.101132437188207e-05,
+      "loss": 1.0087,
+      "step": 6713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.100909948498033e-05,
+      "loss": 0.8928,
+      "step": 6714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1006874547611908e-05,
+      "loss": 0.8868,
+      "step": 6715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1004649559888079e-05,
+      "loss": 0.6171,
+      "step": 6716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1002424521920119e-05,
+      "loss": 0.8759,
+      "step": 6717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1000199433819305e-05,
+      "loss": 0.9302,
+      "step": 6718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0997974295696918e-05,
+      "loss": 1.0534,
+      "step": 6719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0995749107664235e-05,
+      "loss": 0.6181,
+      "step": 6720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0993523869832552e-05,
+      "loss": 0.8514,
+      "step": 6721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0991298582313151e-05,
+      "loss": 0.6693,
+      "step": 6722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0989073245217318e-05,
+      "loss": 0.93,
+      "step": 6723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0986847858656353e-05,
+      "loss": 0.9364,
+      "step": 6724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0984622422741547e-05,
+      "loss": 0.8846,
+      "step": 6725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0982396937584199e-05,
+      "loss": 1.1415,
+      "step": 6726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 0.8684,
+      "step": 6727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0977945819987077e-05,
+      "loss": 0.9429,
+      "step": 6728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0975720187769912e-05,
+      "loss": 0.8858,
+      "step": 6729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0973494506755417e-05,
+      "loss": 0.9394,
+      "step": 6730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0971268777054909e-05,
+      "loss": 0.9222,
+      "step": 6731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.096904299877969e-05,
+      "loss": 0.9633,
+      "step": 6732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0966817172041087e-05,
+      "loss": 0.9949,
+      "step": 6733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0964591296950407e-05,
+      "loss": 0.8681,
+      "step": 6734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.096236537361897e-05,
+      "loss": 0.9085,
+      "step": 6735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0960139402158105e-05,
+      "loss": 0.8462,
+      "step": 6736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0957913382679133e-05,
+      "loss": 0.8974,
+      "step": 6737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0955687315293378e-05,
+      "loss": 1.0515,
+      "step": 6738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0953461200112174e-05,
+      "loss": 0.8861,
+      "step": 6739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0951235037246848e-05,
+      "loss": 0.8935,
+      "step": 6740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0949008826808738e-05,
+      "loss": 0.9365,
+      "step": 6741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0946782568909175e-05,
+      "loss": 1.0544,
+      "step": 6742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0944556263659505e-05,
+      "loss": 0.6571,
+      "step": 6743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0942329911171066e-05,
+      "loss": 0.8757,
+      "step": 6744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0940103511555198e-05,
+      "loss": 0.6219,
+      "step": 6745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.093787706492325e-05,
+      "loss": 0.9367,
+      "step": 6746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0935650571386569e-05,
+      "loss": 0.8667,
+      "step": 6747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0933424031056507e-05,
+      "loss": 0.993,
+      "step": 6748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0931197444044417e-05,
+      "loss": 0.8537,
+      "step": 6749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0928970810461652e-05,
+      "loss": 0.6447,
+      "step": 6750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0926744130419569e-05,
+      "loss": 0.8924,
+      "step": 6751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.092451740402953e-05,
+      "loss": 0.9745,
+      "step": 6752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0922290631402899e-05,
+      "loss": 1.0125,
+      "step": 6753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0920063812651039e-05,
+      "loss": 0.6482,
+      "step": 6754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0917836947885312e-05,
+      "loss": 0.9646,
+      "step": 6755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0915610037217093e-05,
+      "loss": 0.8575,
+      "step": 6756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.091338308075775e-05,
+      "loss": 0.9408,
+      "step": 6757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0911156078618662e-05,
+      "loss": 0.9637,
+      "step": 6758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0908929030911198e-05,
+      "loss": 0.6255,
+      "step": 6759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0906701937746738e-05,
+      "loss": 0.9305,
+      "step": 6760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0904474799236666e-05,
+      "loss": 0.964,
+      "step": 6761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.090224761549236e-05,
+      "loss": 0.9528,
+      "step": 6762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.090002038662521e-05,
+      "loss": 1.0434,
+      "step": 6763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0897793112746597e-05,
+      "loss": 0.9058,
+      "step": 6764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.089556579396792e-05,
+      "loss": 0.9134,
+      "step": 6765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0893338430400561e-05,
+      "loss": 1.0135,
+      "step": 6766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0891111022155915e-05,
+      "loss": 0.9289,
+      "step": 6767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0888883569345388e-05,
+      "loss": 0.9341,
+      "step": 6768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0886656072080368e-05,
+      "loss": 0.967,
+      "step": 6769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.088442853047226e-05,
+      "loss": 0.9166,
+      "step": 6770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0882200944632467e-05,
+      "loss": 0.925,
+      "step": 6771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0879973314672391e-05,
+      "loss": 0.8428,
+      "step": 6772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0877745640703445e-05,
+      "loss": 1.0373,
+      "step": 6773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0875517922837034e-05,
+      "loss": 0.8509,
+      "step": 6774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.087329016118457e-05,
+      "loss": 0.8936,
+      "step": 6775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0871062355857472e-05,
+      "loss": 0.9431,
+      "step": 6776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0868834506967147e-05,
+      "loss": 0.8963,
+      "step": 6777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0866606614625023e-05,
+      "loss": 0.8296,
+      "step": 6778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0864378678942514e-05,
+      "loss": 0.9705,
+      "step": 6779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0862150700031044e-05,
+      "loss": 0.625,
+      "step": 6780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.085992267800204e-05,
+      "loss": 0.9908,
+      "step": 6781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.085769461296692e-05,
+      "loss": 1.0364,
+      "step": 6782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0855466505037128e-05,
+      "loss": 0.8864,
+      "step": 6783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0853238354324087e-05,
+      "loss": 0.8666,
+      "step": 6784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0851010160939229e-05,
+      "loss": 0.9615,
+      "step": 6785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0848781924993993e-05,
+      "loss": 1.0217,
+      "step": 6786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0846553646599814e-05,
+      "loss": 1.0322,
+      "step": 6787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0844325325868132e-05,
+      "loss": 0.8263,
+      "step": 6788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0842096962910392e-05,
+      "loss": 0.6395,
+      "step": 6789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0839868557838034e-05,
+      "loss": 0.9606,
+      "step": 6790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0837640110762508e-05,
+      "loss": 0.9842,
+      "step": 6791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0835411621795258e-05,
+      "loss": 1.0445,
+      "step": 6792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.083318309104774e-05,
+      "loss": 0.9094,
+      "step": 6793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0830954518631395e-05,
+      "loss": 0.9448,
+      "step": 6794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0828725904657693e-05,
+      "loss": 0.9168,
+      "step": 6795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0826497249238082e-05,
+      "loss": 0.9629,
+      "step": 6796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0824268552484016e-05,
+      "loss": 0.8749,
+      "step": 6797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0822039814506964e-05,
+      "loss": 0.9304,
+      "step": 6798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0819811035418384e-05,
+      "loss": 1.0043,
+      "step": 6799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0817582215329746e-05,
+      "loss": 0.6707,
+      "step": 6800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0815353354352513e-05,
+      "loss": 0.9221,
+      "step": 6801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0813124452598154e-05,
+      "loss": 0.9094,
+      "step": 6802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.081089551017814e-05,
+      "loss": 1.0462,
+      "step": 6803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0808666527203944e-05,
+      "loss": 0.9344,
+      "step": 6804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0806437503787043e-05,
+      "loss": 0.9007,
+      "step": 6805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0804208440038913e-05,
+      "loss": 0.8737,
+      "step": 6806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0801979336071028e-05,
+      "loss": 0.9655,
+      "step": 6807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0799750191994877e-05,
+      "loss": 0.9663,
+      "step": 6808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0797521007921939e-05,
+      "loss": 0.7841,
+      "step": 6809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.07952917839637e-05,
+      "loss": 0.9368,
+      "step": 6810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0793062520231644e-05,
+      "loss": 0.9566,
+      "step": 6811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0790833216837267e-05,
+      "loss": 0.9582,
+      "step": 6812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0788603873892052e-05,
+      "loss": 0.9935,
+      "step": 6813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0786374491507494e-05,
+      "loss": 1.0007,
+      "step": 6814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0784145069795093e-05,
+      "loss": 1.0024,
+      "step": 6815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0781915608866344e-05,
+      "loss": 0.8808,
+      "step": 6816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0779686108832744e-05,
+      "loss": 0.8694,
+      "step": 6817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0777456569805794e-05,
+      "loss": 0.9389,
+      "step": 6818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0775226991896993e-05,
+      "loss": 0.9288,
+      "step": 6819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0772997375217856e-05,
+      "loss": 1.0332,
+      "step": 6820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.077076771987988e-05,
+      "loss": 0.9132,
+      "step": 6821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076853802599458e-05,
+      "loss": 0.9818,
+      "step": 6822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0766308293673462e-05,
+      "loss": 0.6412,
+      "step": 6823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0764078523028041e-05,
+      "loss": 0.9406,
+      "step": 6824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0761848714169831e-05,
+      "loss": 0.7741,
+      "step": 6825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0759618867210346e-05,
+      "loss": 0.9374,
+      "step": 6826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.075738898226111e-05,
+      "loss": 0.8932,
+      "step": 6827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0755159059433638e-05,
+      "loss": 0.9941,
+      "step": 6828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0752929098839449e-05,
+      "loss": 0.9335,
+      "step": 6829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0750699100590076e-05,
+      "loss": 0.9445,
+      "step": 6830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.074846906479704e-05,
+      "loss": 0.8909,
+      "step": 6831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0746238991571867e-05,
+      "loss": 0.6456,
+      "step": 6832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0744008881026091e-05,
+      "loss": 0.8495,
+      "step": 6833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0741778733271238e-05,
+      "loss": 0.9321,
+      "step": 6834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0739548548418845e-05,
+      "loss": 0.8924,
+      "step": 6835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0737318326580443e-05,
+      "loss": 0.9412,
+      "step": 6836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0735088067867577e-05,
+      "loss": 0.6681,
+      "step": 6837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.073285777239178e-05,
+      "loss": 0.925,
+      "step": 6838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.073062744026459e-05,
+      "loss": 1.0098,
+      "step": 6839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0728397071597556e-05,
+      "loss": 1.0103,
+      "step": 6840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0726166666502218e-05,
+      "loss": 0.9976,
+      "step": 6841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0723936225090123e-05,
+      "loss": 0.8853,
+      "step": 6842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0721705747472821e-05,
+      "loss": 0.9288,
+      "step": 6843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0719475233761859e-05,
+      "loss": 0.9382,
+      "step": 6844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0717244684068789e-05,
+      "loss": 0.9726,
+      "step": 6845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0715014098505163e-05,
+      "loss": 0.9911,
+      "step": 6846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.071278347718254e-05,
+      "loss": 0.9064,
+      "step": 6847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0710552820212476e-05,
+      "loss": 0.9704,
+      "step": 6848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.070832212770653e-05,
+      "loss": 0.9699,
+      "step": 6849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0706091399776259e-05,
+      "loss": 1.0019,
+      "step": 6850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0703860636533227e-05,
+      "loss": 1.0427,
+      "step": 6851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0701629838089e-05,
+      "loss": 0.9227,
+      "step": 6852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0699399004555145e-05,
+      "loss": 0.8783,
+      "step": 6853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0697168136043221e-05,
+      "loss": 0.853,
+      "step": 6854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0694937232664808e-05,
+      "loss": 0.9423,
+      "step": 6855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.069270629453147e-05,
+      "loss": 1.0265,
+      "step": 6856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0690475321754783e-05,
+      "loss": 0.8003,
+      "step": 6857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0688244314446319e-05,
+      "loss": 0.9271,
+      "step": 6858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.068601327271766e-05,
+      "loss": 1.0039,
+      "step": 6859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0683782196680376e-05,
+      "loss": 0.8205,
+      "step": 6860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0681551086446049e-05,
+      "loss": 0.97,
+      "step": 6861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0679319942126265e-05,
+      "loss": 0.9254,
+      "step": 6862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0677088763832606e-05,
+      "loss": 0.9849,
+      "step": 6863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0674857551676649e-05,
+      "loss": 1.0419,
+      "step": 6864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0672626305769991e-05,
+      "loss": 0.9211,
+      "step": 6865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0670395026224215e-05,
+      "loss": 0.8823,
+      "step": 6866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.066816371315091e-05,
+      "loss": 0.6635,
+      "step": 6867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0665932366661667e-05,
+      "loss": 0.9522,
+      "step": 6868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0663700986868084e-05,
+      "loss": 0.992,
+      "step": 6869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0661469573881754e-05,
+      "loss": 0.6142,
+      "step": 6870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0659238127814268e-05,
+      "loss": 1.0248,
+      "step": 6871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0657006648777233e-05,
+      "loss": 1.0136,
+      "step": 6872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.065477513688224e-05,
+      "loss": 0.6746,
+      "step": 6873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0652543592240898e-05,
+      "loss": 0.8513,
+      "step": 6874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0650312014964807e-05,
+      "loss": 0.8768,
+      "step": 6875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0648080405165568e-05,
+      "loss": 0.8445,
+      "step": 6876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0645848762954795e-05,
+      "loss": 0.9327,
+      "step": 6877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0643617088444094e-05,
+      "loss": 0.9284,
+      "step": 6878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0641385381745071e-05,
+      "loss": 0.9823,
+      "step": 6879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0639153642969337e-05,
+      "loss": 0.9398,
+      "step": 6880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0636921872228509e-05,
+      "loss": 0.8945,
+      "step": 6881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0634690069634198e-05,
+      "loss": 0.9415,
+      "step": 6882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0632458235298021e-05,
+      "loss": 0.9387,
+      "step": 6883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06302263693316e-05,
+      "loss": 0.8821,
+      "step": 6884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0627994471846548e-05,
+      "loss": 1.0157,
+      "step": 6885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0625762542954486e-05,
+      "loss": 0.9268,
+      "step": 6886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.062353058276704e-05,
+      "loss": 0.8327,
+      "step": 6887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0621298591395828e-05,
+      "loss": 0.9734,
+      "step": 6888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0619066568952485e-05,
+      "loss": 0.9918,
+      "step": 6889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.061683451554863e-05,
+      "loss": 0.9385,
+      "step": 6890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0614602431295892e-05,
+      "loss": 0.8462,
+      "step": 6891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0612370316305907e-05,
+      "loss": 1.0042,
+      "step": 6892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0610138170690298e-05,
+      "loss": 0.8828,
+      "step": 6893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.060790599456071e-05,
+      "loss": 0.7852,
+      "step": 6894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0605673788028768e-05,
+      "loss": 0.8997,
+      "step": 6895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0603441551206109e-05,
+      "loss": 0.892,
+      "step": 6896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0601209284204374e-05,
+      "loss": 0.8324,
+      "step": 6897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.05989769871352e-05,
+      "loss": 0.9886,
+      "step": 6898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.059674466011023e-05,
+      "loss": 0.9677,
+      "step": 6899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0594512303241105e-05,
+      "loss": 0.9263,
+      "step": 6900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0592279916639468e-05,
+      "loss": 0.983,
+      "step": 6901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0590047500416964e-05,
+      "loss": 0.9117,
+      "step": 6902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0587815054685241e-05,
+      "loss": 1.0252,
+      "step": 6903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0585582579555948e-05,
+      "loss": 0.9507,
+      "step": 6904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0583350075140732e-05,
+      "loss": 0.9517,
+      "step": 6905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.058111754155125e-05,
+      "loss": 0.7682,
+      "step": 6906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0578884978899146e-05,
+      "loss": 0.8672,
+      "step": 6907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0576652387296073e-05,
+      "loss": 0.9232,
+      "step": 6908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0574419766853701e-05,
+      "loss": 0.9256,
+      "step": 6909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0572187117683674e-05,
+      "loss": 0.6403,
+      "step": 6910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0569954439897652e-05,
+      "loss": 0.8195,
+      "step": 6911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0567721733607298e-05,
+      "loss": 0.7733,
+      "step": 6912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0565488998924274e-05,
+      "loss": 0.8878,
+      "step": 6913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.056325623596024e-05,
+      "loss": 1.0193,
+      "step": 6914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0561023444826856e-05,
+      "loss": 0.6511,
+      "step": 6915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0558790625635798e-05,
+      "loss": 0.9931,
+      "step": 6916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0556557778498726e-05,
+      "loss": 0.6288,
+      "step": 6917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0554324903527306e-05,
+      "loss": 1.0147,
+      "step": 6918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0552092000833211e-05,
+      "loss": 1.0095,
+      "step": 6919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0549859070528115e-05,
+      "loss": 0.8733,
+      "step": 6920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0547626112723685e-05,
+      "loss": 0.9372,
+      "step": 6921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.05453931275316e-05,
+      "loss": 0.9611,
+      "step": 6922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0543160115063524e-05,
+      "loss": 0.9546,
+      "step": 6923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0540927075431152e-05,
+      "loss": 0.8561,
+      "step": 6924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0538694008746147e-05,
+      "loss": 0.8729,
+      "step": 6925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.053646091512019e-05,
+      "loss": 1.0156,
+      "step": 6926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0534227794664971e-05,
+      "loss": 1.002,
+      "step": 6927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0531994647492158e-05,
+      "loss": 0.9527,
+      "step": 6928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0529761473713447e-05,
+      "loss": 0.89,
+      "step": 6929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0527528273440515e-05,
+      "loss": 0.8664,
+      "step": 6930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0525295046785054e-05,
+      "loss": 0.9073,
+      "step": 6931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0523061793858744e-05,
+      "loss": 0.9009,
+      "step": 6932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0520828514773275e-05,
+      "loss": 0.9133,
+      "step": 6933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0518595209640344e-05,
+      "loss": 0.8801,
+      "step": 6934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0516361878571633e-05,
+      "loss": 0.9221,
+      "step": 6935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0514128521678839e-05,
+      "loss": 0.9059,
+      "step": 6936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0511895139073655e-05,
+      "loss": 0.9498,
+      "step": 6937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0509661730867778e-05,
+      "loss": 0.9725,
+      "step": 6938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0507428297172899e-05,
+      "loss": 0.936,
+      "step": 6939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0505194838100721e-05,
+      "loss": 0.8235,
+      "step": 6940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0502961353762942e-05,
+      "loss": 0.9815,
+      "step": 6941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.050072784427126e-05,
+      "loss": 0.8643,
+      "step": 6942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0498494309737379e-05,
+      "loss": 0.9054,
+      "step": 6943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0496260750273e-05,
+      "loss": 0.9722,
+      "step": 6944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0494027165989825e-05,
+      "loss": 0.8801,
+      "step": 6945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0491793556999563e-05,
+      "loss": 0.9213,
+      "step": 6946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.048955992341392e-05,
+      "loss": 0.9827,
+      "step": 6947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0487326265344597e-05,
+      "loss": 0.9372,
+      "step": 6948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0485092582903314e-05,
+      "loss": 0.8285,
+      "step": 6949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.048285887620177e-05,
+      "loss": 0.8633,
+      "step": 6950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0480625145351681e-05,
+      "loss": 0.9009,
+      "step": 6951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0478391390464762e-05,
+      "loss": 0.9417,
+      "step": 6952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0476157611652724e-05,
+      "loss": 0.8124,
+      "step": 6953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0473923809027285e-05,
+      "loss": 0.8872,
+      "step": 6954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0471689982700151e-05,
+      "loss": 0.91,
+      "step": 6955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0469456132783051e-05,
+      "loss": 0.9393,
+      "step": 6956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.04672222593877e-05,
+      "loss": 0.8653,
+      "step": 6957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0464988362625812e-05,
+      "loss": 0.9721,
+      "step": 6958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0462754442609115e-05,
+      "loss": 0.8262,
+      "step": 6959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0460520499449327e-05,
+      "loss": 0.9366,
+      "step": 6960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0458286533258175e-05,
+      "loss": 0.9931,
+      "step": 6961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0456052544147377e-05,
+      "loss": 0.9396,
+      "step": 6962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0453818532228664e-05,
+      "loss": 0.9677,
+      "step": 6963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.045158449761376e-05,
+      "loss": 0.9723,
+      "step": 6964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0449350440414391e-05,
+      "loss": 0.9854,
+      "step": 6965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0447116360742288e-05,
+      "loss": 0.8725,
+      "step": 6966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0444882258709181e-05,
+      "loss": 0.9992,
+      "step": 6967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0442648134426802e-05,
+      "loss": 1.0177,
+      "step": 6968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0440413988006882e-05,
+      "loss": 1.0129,
+      "step": 6969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0438179819561152e-05,
+      "loss": 0.9536,
+      "step": 6970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0435945629201348e-05,
+      "loss": 0.9029,
+      "step": 6971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0433711417039207e-05,
+      "loss": 0.8388,
+      "step": 6972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0431477183186462e-05,
+      "loss": 1.0011,
+      "step": 6973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0429242927754855e-05,
+      "loss": 0.8924,
+      "step": 6974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0427008650856121e-05,
+      "loss": 0.8169,
+      "step": 6975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0424774352602002e-05,
+      "loss": 0.9453,
+      "step": 6976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0422540033104238e-05,
+      "loss": 0.5945,
+      "step": 6977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0420305692474572e-05,
+      "loss": 1.0059,
+      "step": 6978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0418071330824749e-05,
+      "loss": 0.932,
+      "step": 6979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0415836948266505e-05,
+      "loss": 0.8797,
+      "step": 6980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0413602544911593e-05,
+      "loss": 0.8988,
+      "step": 6981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0411368120871751e-05,
+      "loss": 0.8622,
+      "step": 6982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0409133676258735e-05,
+      "loss": 0.8576,
+      "step": 6983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0406899211184292e-05,
+      "loss": 0.8701,
+      "step": 6984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0404664725760162e-05,
+      "loss": 0.9834,
+      "step": 6985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0402430220098106e-05,
+      "loss": 0.9539,
+      "step": 6986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0400195694309866e-05,
+      "loss": 1.0046,
+      "step": 6987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0397961148507204e-05,
+      "loss": 1.0077,
+      "step": 6988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.039572658280187e-05,
+      "loss": 0.9918,
+      "step": 6989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0393491997305613e-05,
+      "loss": 0.5879,
+      "step": 6990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0391257392130194e-05,
+      "loss": 1.0234,
+      "step": 6991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0389022767387364e-05,
+      "loss": 0.8972,
+      "step": 6992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0386788123188886e-05,
+      "loss": 0.8662,
+      "step": 6993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0384553459646515e-05,
+      "loss": 1.006,
+      "step": 6994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0382318776872009e-05,
+      "loss": 0.9172,
+      "step": 6995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.038008407497713e-05,
+      "loss": 0.926,
+      "step": 6996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0377849354073638e-05,
+      "loss": 0.9574,
+      "step": 6997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0375614614273297e-05,
+      "loss": 0.9872,
+      "step": 6998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0373379855687864e-05,
+      "loss": 0.9728,
+      "step": 6999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0371145078429113e-05,
+      "loss": 0.9833,
+      "step": 7000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.03689102826088e-05,
+      "loss": 0.8814,
+      "step": 7001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.036667546833869e-05,
+      "loss": 0.9749,
+      "step": 7002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0364440635730558e-05,
+      "loss": 0.9568,
+      "step": 7003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0362205784896165e-05,
+      "loss": 0.9308,
+      "step": 7004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0359970915947282e-05,
+      "loss": 0.8326,
+      "step": 7005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0357736028995677e-05,
+      "loss": 1.0028,
+      "step": 7006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.035550112415312e-05,
+      "loss": 0.9722,
+      "step": 7007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0353266201531386e-05,
+      "loss": 0.6094,
+      "step": 7008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0351031261242242e-05,
+      "loss": 1.0075,
+      "step": 7009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0348796303397463e-05,
+      "loss": 0.9361,
+      "step": 7010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0346561328108826e-05,
+      "loss": 1.1007,
+      "step": 7011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0344326335488098e-05,
+      "loss": 1.0057,
+      "step": 7012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.034209132564706e-05,
+      "loss": 0.9433,
+      "step": 7013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.033985629869749e-05,
+      "loss": 0.626,
+      "step": 7014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0337621254751163e-05,
+      "loss": 1.005,
+      "step": 7015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0335386193919856e-05,
+      "loss": 0.6609,
+      "step": 7016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0333151116315348e-05,
+      "loss": 0.8379,
+      "step": 7017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0330916022049418e-05,
+      "loss": 1.0099,
+      "step": 7018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0328680911233852e-05,
+      "loss": 0.6304,
+      "step": 7019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0326445783980426e-05,
+      "loss": 0.9868,
+      "step": 7020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0324210640400924e-05,
+      "loss": 0.9545,
+      "step": 7021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.032197548060713e-05,
+      "loss": 0.8343,
+      "step": 7022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0319740304710828e-05,
+      "loss": 0.6772,
+      "step": 7023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0317505112823799e-05,
+      "loss": 0.9573,
+      "step": 7024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0315269905057834e-05,
+      "loss": 0.9425,
+      "step": 7025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0313034681524716e-05,
+      "loss": 0.8857,
+      "step": 7026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0310799442336233e-05,
+      "loss": 0.9254,
+      "step": 7027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0308564187604174e-05,
+      "loss": 1.0234,
+      "step": 7028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0306328917440323e-05,
+      "loss": 0.9748,
+      "step": 7029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0304093631956475e-05,
+      "loss": 0.8345,
+      "step": 7030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0301858331264418e-05,
+      "loss": 0.7964,
+      "step": 7031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.029962301547594e-05,
+      "loss": 0.9164,
+      "step": 7032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0297387684702837e-05,
+      "loss": 0.6485,
+      "step": 7033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0295152339056902e-05,
+      "loss": 0.8525,
+      "step": 7034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0292916978649927e-05,
+      "loss": 1.0128,
+      "step": 7035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0290681603593704e-05,
+      "loss": 0.9971,
+      "step": 7036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0288446214000027e-05,
+      "loss": 0.9555,
+      "step": 7037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0286210809980696e-05,
+      "loss": 0.9831,
+      "step": 7038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0283975391647505e-05,
+      "loss": 0.6115,
+      "step": 7039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0281739959112249e-05,
+      "loss": 0.9254,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0279504512486729e-05,
+      "loss": 0.9812,
+      "step": 7041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0277269051882739e-05,
+      "loss": 0.9659,
+      "step": 7042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0275033577412082e-05,
+      "loss": 0.9334,
+      "step": 7043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0272798089186553e-05,
+      "loss": 0.9062,
+      "step": 7044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0270562587317955e-05,
+      "loss": 0.9066,
+      "step": 7045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0268327071918091e-05,
+      "loss": 0.9058,
+      "step": 7046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0266091543098763e-05,
+      "loss": 0.9974,
+      "step": 7047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.026385600097177e-05,
+      "loss": 0.9111,
+      "step": 7048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0261620445648912e-05,
+      "loss": 0.9608,
+      "step": 7049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0259384877242e-05,
+      "loss": 0.9659,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0257149295862838e-05,
+      "loss": 0.9728,
+      "step": 7051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0254913701623224e-05,
+      "loss": 0.8855,
+      "step": 7052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0252678094634972e-05,
+      "loss": 0.8814,
+      "step": 7053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.025044247500988e-05,
+      "loss": 0.9297,
+      "step": 7054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0248206842859763e-05,
+      "loss": 0.9693,
+      "step": 7055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.024597119829642e-05,
+      "loss": 0.9638,
+      "step": 7056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.024373554143167e-05,
+      "loss": 0.6282,
+      "step": 7057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0241499872377314e-05,
+      "loss": 0.9121,
+      "step": 7058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0239264191245159e-05,
+      "loss": 1.0809,
+      "step": 7059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0237028498147024e-05,
+      "loss": 0.8767,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.023479279319471e-05,
+      "loss": 0.953,
+      "step": 7061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0232557076500038e-05,
+      "loss": 0.9682,
+      "step": 7062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0230321348174812e-05,
+      "loss": 0.8605,
+      "step": 7063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0228085608330847e-05,
+      "loss": 0.8514,
+      "step": 7064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0225849857079953e-05,
+      "loss": 0.8443,
+      "step": 7065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0223614094533949e-05,
+      "loss": 0.9932,
+      "step": 7066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0221378320804647e-05,
+      "loss": 0.8683,
+      "step": 7067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.021914253600386e-05,
+      "loss": 0.945,
+      "step": 7068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0216906740243404e-05,
+      "loss": 0.9505,
+      "step": 7069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0214670933635095e-05,
+      "loss": 0.9661,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.021243511629075e-05,
+      "loss": 0.9337,
+      "step": 7071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0210199288322187e-05,
+      "loss": 0.9793,
+      "step": 7072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0207963449841219e-05,
+      "loss": 0.8508,
+      "step": 7073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0205727600959664e-05,
+      "loss": 0.9379,
+      "step": 7074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0203491741789346e-05,
+      "loss": 0.8979,
+      "step": 7075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.020125587244208e-05,
+      "loss": 1.0182,
+      "step": 7076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0199019993029686e-05,
+      "loss": 0.9293,
+      "step": 7077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0196784103663981e-05,
+      "loss": 0.9059,
+      "step": 7078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.019454820445679e-05,
+      "loss": 0.9438,
+      "step": 7079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0192312295519933e-05,
+      "loss": 0.851,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0190076376965227e-05,
+      "loss": 0.8626,
+      "step": 7081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0187840448904503e-05,
+      "loss": 0.9689,
+      "step": 7082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0185604511449573e-05,
+      "loss": 0.6341,
+      "step": 7083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0183368564712265e-05,
+      "loss": 0.883,
+      "step": 7084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0181132608804405e-05,
+      "loss": 0.9991,
+      "step": 7085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.017889664383781e-05,
+      "loss": 0.9478,
+      "step": 7086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.017666066992431e-05,
+      "loss": 0.6392,
+      "step": 7087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0174424687175729e-05,
+      "loss": 0.9705,
+      "step": 7088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0172188695703887e-05,
+      "loss": 1.0339,
+      "step": 7089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0169952695620613e-05,
+      "loss": 0.9572,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0167716687037732e-05,
+      "loss": 0.8786,
+      "step": 7091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0165480670067073e-05,
+      "loss": 0.9345,
+      "step": 7092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.016324464482046e-05,
+      "loss": 0.9299,
+      "step": 7093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0161008611409723e-05,
+      "loss": 0.82,
+      "step": 7094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0158772569946686e-05,
+      "loss": 0.9523,
+      "step": 7095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0156536520543176e-05,
+      "loss": 0.9082,
+      "step": 7096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0154300463311029e-05,
+      "loss": 0.9266,
+      "step": 7097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0152064398362069e-05,
+      "loss": 0.8661,
+      "step": 7098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0149828325808122e-05,
+      "loss": 0.9608,
+      "step": 7099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0147592245761025e-05,
+      "loss": 0.8665,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0145356158332598e-05,
+      "loss": 0.6656,
+      "step": 7101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0143120063634681e-05,
+      "loss": 0.8643,
+      "step": 7102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.01408839617791e-05,
+      "loss": 0.9513,
+      "step": 7103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0138647852877688e-05,
+      "loss": 0.9468,
+      "step": 7104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0136411737042273e-05,
+      "loss": 1.045,
+      "step": 7105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.013417561438469e-05,
+      "loss": 0.9714,
+      "step": 7106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0131939485016768e-05,
+      "loss": 0.9095,
+      "step": 7107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0129703349050342e-05,
+      "loss": 0.9268,
+      "step": 7108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0127467206597245e-05,
+      "loss": 0.8505,
+      "step": 7109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.012523105776931e-05,
+      "loss": 0.9684,
+      "step": 7110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0122994902678366e-05,
+      "loss": 0.8856,
+      "step": 7111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0120758741436251e-05,
+      "loss": 0.944,
+      "step": 7112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0118522574154802e-05,
+      "loss": 0.87,
+      "step": 7113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0116286400945844e-05,
+      "loss": 0.9057,
+      "step": 7114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0114050221921221e-05,
+      "loss": 0.9253,
+      "step": 7115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0111814037192761e-05,
+      "loss": 0.9876,
+      "step": 7116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0109577846872307e-05,
+      "loss": 0.9104,
+      "step": 7117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0107341651071685e-05,
+      "loss": 0.9141,
+      "step": 7118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0105105449902737e-05,
+      "loss": 0.9339,
+      "step": 7119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0102869243477298e-05,
+      "loss": 0.9253,
+      "step": 7120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100633031907204e-05,
+      "loss": 0.8934,
+      "step": 7121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009839681530429e-05,
+      "loss": 0.8469,
+      "step": 7122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0096160593780391e-05,
+      "loss": 0.8433,
+      "step": 7123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0093924367447352e-05,
+      "loss": 0.8629,
+      "step": 7124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0091688136417002e-05,
+      "loss": 0.8687,
+      "step": 7125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0089451900801182e-05,
+      "loss": 0.8776,
+      "step": 7126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0087215660711728e-05,
+      "loss": 0.9768,
+      "step": 7127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084979416260481e-05,
+      "loss": 0.7769,
+      "step": 7128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0082743167559279e-05,
+      "loss": 0.8252,
+      "step": 7129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0080506914719956e-05,
+      "loss": 0.8863,
+      "step": 7130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0078270657854355e-05,
+      "loss": 1.0507,
+      "step": 7131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076034397074315e-05,
+      "loss": 0.9467,
+      "step": 7132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073798132491668e-05,
+      "loss": 1.0093,
+      "step": 7133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0071561864218262e-05,
+      "loss": 0.5939,
+      "step": 7134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069325592365933e-05,
+      "loss": 0.8547,
+      "step": 7135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006708931704652e-05,
+      "loss": 0.9365,
+      "step": 7136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0064853038371862e-05,
+      "loss": 1.0692,
+      "step": 7137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0062616756453798e-05,
+      "loss": 0.9189,
+      "step": 7138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0060380471404174e-05,
+      "loss": 1.0063,
+      "step": 7139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0058144183334821e-05,
+      "loss": 1.038,
+      "step": 7140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0055907892357591e-05,
+      "loss": 1.0354,
+      "step": 7141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0053671598584313e-05,
+      "loss": 0.7778,
+      "step": 7142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0051435302126831e-05,
+      "loss": 1.0971,
+      "step": 7143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0049199003096992e-05,
+      "loss": 0.645,
+      "step": 7144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0046962701606631e-05,
+      "loss": 0.9138,
+      "step": 7145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0044726397767589e-05,
+      "loss": 0.879,
+      "step": 7146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0042490091691708e-05,
+      "loss": 0.9472,
+      "step": 7147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0040253783490827e-05,
+      "loss": 0.8725,
+      "step": 7148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0038017473276796e-05,
+      "loss": 0.9233,
+      "step": 7149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0035781161161446e-05,
+      "loss": 0.9897,
+      "step": 7150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0033544847256625e-05,
+      "loss": 0.9511,
+      "step": 7151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0031308531674174e-05,
+      "loss": 0.8991,
+      "step": 7152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0029072214525927e-05,
+      "loss": 0.9229,
+      "step": 7153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0026835895923739e-05,
+      "loss": 0.9347,
+      "step": 7154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0024599575979439e-05,
+      "loss": 0.9503,
+      "step": 7155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.002236325480488e-05,
+      "loss": 0.9772,
+      "step": 7156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0020126932511897e-05,
+      "loss": 0.9481,
+      "step": 7157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0017890609212334e-05,
+      "loss": 0.6606,
+      "step": 7158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.001565428501803e-05,
+      "loss": 0.8905,
+      "step": 7159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0013417960040835e-05,
+      "loss": 0.8768,
+      "step": 7160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0011181634392581e-05,
+      "loss": 0.9571,
+      "step": 7161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0008945308185121e-05,
+      "loss": 0.9509,
+      "step": 7162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0006708981530287e-05,
+      "loss": 0.8982,
+      "step": 7163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.000447265453993e-05,
+      "loss": 0.8393,
+      "step": 7164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0002236327325885e-05,
+      "loss": 0.6386,
+      "step": 7165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1e-05,
+      "loss": 0.955,
+      "step": 7166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.997763672674117e-06,
+      "loss": 1.0545,
+      "step": 7167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.995527345460073e-06,
+      "loss": 0.8873,
+      "step": 7168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.993291018469714e-06,
+      "loss": 0.8997,
+      "step": 7169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.991054691814884e-06,
+      "loss": 0.9621,
+      "step": 7170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.98881836560742e-06,
+      "loss": 0.8811,
+      "step": 7171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.98658203995917e-06,
+      "loss": 0.9489,
+      "step": 7172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.984345714981973e-06,
+      "loss": 0.9572,
+      "step": 7173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.982109390787671e-06,
+      "loss": 0.9495,
+      "step": 7174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.979873067488108e-06,
+      "loss": 0.9467,
+      "step": 7175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.977636745195126e-06,
+      "loss": 0.9741,
+      "step": 7176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.975400424020563e-06,
+      "loss": 0.609,
+      "step": 7177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.973164104076268e-06,
+      "loss": 0.8843,
+      "step": 7178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.970927785474078e-06,
+      "loss": 0.8,
+      "step": 7179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.968691468325833e-06,
+      "loss": 0.6606,
+      "step": 7180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.96645515274338e-06,
+      "loss": 0.9769,
+      "step": 7181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.964218838838555e-06,
+      "loss": 0.7906,
+      "step": 7182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.961982526723204e-06,
+      "loss": 0.8644,
+      "step": 7183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.959746216509171e-06,
+      "loss": 0.8597,
+      "step": 7184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.957509908308292e-06,
+      "loss": 0.8514,
+      "step": 7185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.955273602232413e-06,
+      "loss": 0.9772,
+      "step": 7186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.953037298393374e-06,
+      "loss": 0.9732,
+      "step": 7187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.95080099690301e-06,
+      "loss": 0.9696,
+      "step": 7188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.948564697873169e-06,
+      "loss": 0.9828,
+      "step": 7189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.94632840141569e-06,
+      "loss": 0.8386,
+      "step": 7190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.944092107642414e-06,
+      "loss": 0.8481,
+      "step": 7191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.94185581666518e-06,
+      "loss": 0.901,
+      "step": 7192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.939619528595828e-06,
+      "loss": 0.9088,
+      "step": 7193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.937383243546203e-06,
+      "loss": 0.8668,
+      "step": 7194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.93514696162814e-06,
+      "loss": 0.8162,
+      "step": 7195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.932910682953484e-06,
+      "loss": 0.8742,
+      "step": 7196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.930674407634069e-06,
+      "loss": 0.9527,
+      "step": 7197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.92843813578174e-06,
+      "loss": 0.8717,
+      "step": 7198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.926201867508335e-06,
+      "loss": 0.9133,
+      "step": 7199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.92396560292569e-06,
+      "loss": 0.97,
+      "step": 7200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.921729342145648e-06,
+      "loss": 1.033,
+      "step": 7201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.919493085280046e-06,
+      "loss": 0.8211,
+      "step": 7202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.917256832440725e-06,
+      "loss": 1.0255,
+      "step": 7203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.915020583739522e-06,
+      "loss": 0.9977,
+      "step": 7204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.912784339288274e-06,
+      "loss": 0.8815,
+      "step": 7205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.910548099198823e-06,
+      "loss": 0.8829,
+      "step": 7206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.908311863583002e-06,
+      "loss": 0.9098,
+      "step": 7207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.906075632552655e-06,
+      "loss": 0.9594,
+      "step": 7208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.90383940621961e-06,
+      "loss": 0.8466,
+      "step": 7209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.901603184695715e-06,
+      "loss": 0.967,
+      "step": 7210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.899366968092803e-06,
+      "loss": 0.9909,
+      "step": 7211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897130756522706e-06,
+      "loss": 0.8367,
+      "step": 7212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.894894550097266e-06,
+      "loss": 0.896,
+      "step": 7213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.892658348928315e-06,
+      "loss": 0.9906,
+      "step": 7214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.890422153127695e-06,
+      "loss": 0.9483,
+      "step": 7215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.888185962807239e-06,
+      "loss": 0.9278,
+      "step": 7216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.885949778078782e-06,
+      "loss": 1.0596,
+      "step": 7217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.883713599054156e-06,
+      "loss": 0.9482,
+      "step": 7218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.881477425845202e-06,
+      "loss": 0.9769,
+      "step": 7219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.879241258563748e-06,
+      "loss": 0.8975,
+      "step": 7220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.877005097321635e-06,
+      "loss": 0.8323,
+      "step": 7221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.874768942230692e-06,
+      "loss": 0.9387,
+      "step": 7222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.872532793402756e-06,
+      "loss": 0.9094,
+      "step": 7223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.870296650949662e-06,
+      "loss": 0.9317,
+      "step": 7224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.868060514983233e-06,
+      "loss": 0.91,
+      "step": 7225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.865824385615314e-06,
+      "loss": 0.8364,
+      "step": 7226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.86358826295773e-06,
+      "loss": 0.9075,
+      "step": 7227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.861352147122317e-06,
+      "loss": 0.937,
+      "step": 7228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.859116038220904e-06,
+      "loss": 0.907,
+      "step": 7229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.85687993636532e-06,
+      "loss": 0.6267,
+      "step": 7230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.854643841667403e-06,
+      "loss": 0.9644,
+      "step": 7231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.852407754238978e-06,
+      "loss": 0.9374,
+      "step": 7232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.850171674191881e-06,
+      "loss": 0.8681,
+      "step": 7233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.847935601637933e-06,
+      "loss": 0.938,
+      "step": 7234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.845699536688973e-06,
+      "loss": 0.9685,
+      "step": 7235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.843463479456825e-06,
+      "loss": 0.8454,
+      "step": 7236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.841227430053317e-06,
+      "loss": 0.9425,
+      "step": 7237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.838991388590282e-06,
+      "loss": 0.9888,
+      "step": 7238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.836755355179542e-06,
+      "loss": 0.8953,
+      "step": 7239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.83451932993293e-06,
+      "loss": 0.9306,
+      "step": 7240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.832283312962273e-06,
+      "loss": 0.9202,
+      "step": 7241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.830047304379392e-06,
+      "loss": 0.9187,
+      "step": 7242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.82781130429612e-06,
+      "loss": 1.0969,
+      "step": 7243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.825575312824276e-06,
+      "loss": 0.9106,
+      "step": 7244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.82333933007569e-06,
+      "loss": 1.0562,
+      "step": 7245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.821103356162189e-06,
+      "loss": 0.9842,
+      "step": 7246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.818867391195595e-06,
+      "loss": 0.9501,
+      "step": 7247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.816631435287733e-06,
+      "loss": 0.6133,
+      "step": 7248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.814395488550428e-06,
+      "loss": 0.9591,
+      "step": 7249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.812159551095499e-06,
+      "loss": 0.9257,
+      "step": 7250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.809923623034773e-06,
+      "loss": 0.9868,
+      "step": 7251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.807687704480067e-06,
+      "loss": 0.9099,
+      "step": 7252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.80545179554321e-06,
+      "loss": 0.9383,
+      "step": 7253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.803215896336022e-06,
+      "loss": 0.9914,
+      "step": 7254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.800980006970317e-06,
+      "loss": 0.9731,
+      "step": 7255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.798744127557922e-06,
+      "loss": 0.944,
+      "step": 7256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.796508258210656e-06,
+      "loss": 0.8902,
+      "step": 7257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.794272399040337e-06,
+      "loss": 0.9673,
+      "step": 7258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.792036550158783e-06,
+      "loss": 0.9762,
+      "step": 7259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.789800711677817e-06,
+      "loss": 0.6146,
+      "step": 7260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.787564883709253e-06,
+      "loss": 0.9453,
+      "step": 7261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.785329066364906e-06,
+      "loss": 0.8636,
+      "step": 7262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7830932597566e-06,
+      "loss": 0.934,
+      "step": 7263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.780857463996141e-06,
+      "loss": 0.8196,
+      "step": 7264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.778621679195357e-06,
+      "loss": 0.841,
+      "step": 7265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.776385905466053e-06,
+      "loss": 1.0204,
+      "step": 7266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.774150142920049e-06,
+      "loss": 0.9101,
+      "step": 7267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77191439166916e-06,
+      "loss": 0.9974,
+      "step": 7268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.769678651825193e-06,
+      "loss": 0.9436,
+      "step": 7269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.767442923499969e-06,
+      "loss": 1.0565,
+      "step": 7270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.765207206805291e-06,
+      "loss": 0.8708,
+      "step": 7271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.762971501852981e-06,
+      "loss": 0.9203,
+      "step": 7272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.760735808754846e-06,
+      "loss": 1.0453,
+      "step": 7273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.758500127622693e-06,
+      "loss": 0.9755,
+      "step": 7274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.756264458568335e-06,
+      "loss": 0.6376,
+      "step": 7275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.75402880170358e-06,
+      "loss": 0.8298,
+      "step": 7276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.751793157140237e-06,
+      "loss": 0.9442,
+      "step": 7277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.74955752499012e-06,
+      "loss": 0.9274,
+      "step": 7278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.74732190536503e-06,
+      "loss": 0.9291,
+      "step": 7279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.745086298376777e-06,
+      "loss": 0.9037,
+      "step": 7280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.742850704137165e-06,
+      "loss": 0.8483,
+      "step": 7281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.740615122758e-06,
+      "loss": 0.9182,
+      "step": 7282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.73837955435109e-06,
+      "loss": 0.8929,
+      "step": 7283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.736143999028233e-06,
+      "loss": 0.9545,
+      "step": 7284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.733908456901239e-06,
+      "loss": 0.9494,
+      "step": 7285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.731672928081912e-06,
+      "loss": 0.9453,
+      "step": 7286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.729437412682046e-06,
+      "loss": 0.9036,
+      "step": 7287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72720191081345e-06,
+      "loss": 0.935,
+      "step": 7288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.724966422587922e-06,
+      "loss": 0.8793,
+      "step": 7289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.722730948117264e-06,
+      "loss": 0.9499,
+      "step": 7290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.720495487513275e-06,
+      "loss": 0.9164,
+      "step": 7291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.718260040887753e-06,
+      "loss": 0.8857,
+      "step": 7292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.716024608352499e-06,
+      "loss": 0.9812,
+      "step": 7293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.713789190019306e-06,
+      "loss": 0.8335,
+      "step": 7294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.711553785999975e-06,
+      "loss": 0.8486,
+      "step": 7295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7093183964063e-06,
+      "loss": 0.8931,
+      "step": 7296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.707083021350076e-06,
+      "loss": 0.8583,
+      "step": 7297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.704847660943101e-06,
+      "loss": 0.9648,
+      "step": 7298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.702612315297165e-06,
+      "loss": 0.905,
+      "step": 7299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.700376984524064e-06,
+      "loss": 0.9499,
+      "step": 7300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.698141668735587e-06,
+      "loss": 1.0303,
+      "step": 7301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.69590636804353e-06,
+      "loss": 0.9136,
+      "step": 7302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.693671082559682e-06,
+      "loss": 0.9752,
+      "step": 7303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.691435812395833e-06,
+      "loss": 0.926,
+      "step": 7304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.689200557663772e-06,
+      "loss": 0.8338,
+      "step": 7305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.68696531847529e-06,
+      "loss": 0.9123,
+      "step": 7306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.684730094942172e-06,
+      "loss": 0.9407,
+      "step": 7307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.682494887176203e-06,
+      "loss": 0.9099,
+      "step": 7308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.680259695289174e-06,
+      "loss": 0.9302,
+      "step": 7309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678024519392872e-06,
+      "loss": 0.8965,
+      "step": 7310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.675789359599077e-06,
+      "loss": 0.909,
+      "step": 7311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.673554216019576e-06,
+      "loss": 0.8815,
+      "step": 7312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.671319088766151e-06,
+      "loss": 0.6182,
+      "step": 7313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.669083977950582e-06,
+      "loss": 0.8617,
+      "step": 7314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.666848883684656e-06,
+      "loss": 1.0287,
+      "step": 7315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.664613806080147e-06,
+      "loss": 1.0277,
+      "step": 7316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.66237874524884e-06,
+      "loss": 0.9959,
+      "step": 7317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.660143701302513e-06,
+      "loss": 0.858,
+      "step": 7318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.657908674352941e-06,
+      "loss": 1.0185,
+      "step": 7319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.655673664511903e-06,
+      "loss": 0.8987,
+      "step": 7320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.653438671891178e-06,
+      "loss": 0.8908,
+      "step": 7321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.651203696602538e-06,
+      "loss": 0.9246,
+      "step": 7322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.648968738757762e-06,
+      "loss": 0.9076,
+      "step": 7323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.646733798468617e-06,
+      "loss": 0.6437,
+      "step": 7324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.644498875846882e-06,
+      "loss": 0.9018,
+      "step": 7325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.642263971004325e-06,
+      "loss": 0.9599,
+      "step": 7326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.640029084052723e-06,
+      "loss": 0.8848,
+      "step": 7327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.637794215103837e-06,
+      "loss": 1.0018,
+      "step": 7328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.635559364269446e-06,
+      "loss": 0.9555,
+      "step": 7329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.633324531661314e-06,
+      "loss": 0.9066,
+      "step": 7330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.631089717391205e-06,
+      "loss": 0.9525,
+      "step": 7331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.628854921570894e-06,
+      "loss": 0.8999,
+      "step": 7332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.626620144312137e-06,
+      "loss": 1.0795,
+      "step": 7333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.624385385726708e-06,
+      "loss": 0.8615,
+      "step": 7334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.622150645926367e-06,
+      "loss": 0.8289,
+      "step": 7335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.619915925022874e-06,
+      "loss": 0.6362,
+      "step": 7336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.617681223127996e-06,
+      "loss": 0.6204,
+      "step": 7337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.61544654035349e-06,
+      "loss": 0.8975,
+      "step": 7338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.613211876811114e-06,
+      "loss": 0.9128,
+      "step": 7339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.610977232612636e-06,
+      "loss": 0.8603,
+      "step": 7340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.608742607869806e-06,
+      "loss": 0.9531,
+      "step": 7341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.606508002694387e-06,
+      "loss": 0.9661,
+      "step": 7342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.604273417198133e-06,
+      "loss": 0.9475,
+      "step": 7343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.602038851492794e-06,
+      "loss": 1.0837,
+      "step": 7344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.599804305690132e-06,
+      "loss": 0.8829,
+      "step": 7345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.597569779901896e-06,
+      "loss": 0.8888,
+      "step": 7346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.59533527423984e-06,
+      "loss": 0.8878,
+      "step": 7347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.593100788815712e-06,
+      "loss": 0.881,
+      "step": 7348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.590866323741266e-06,
+      "loss": 0.8577,
+      "step": 7349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.588631879128252e-06,
+      "loss": 0.9289,
+      "step": 7350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.586397455088412e-06,
+      "loss": 0.9101,
+      "step": 7351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.584163051733499e-06,
+      "loss": 0.9566,
+      "step": 7352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.581928669175255e-06,
+      "loss": 0.9147,
+      "step": 7353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.57969430752543e-06,
+      "loss": 0.8542,
+      "step": 7354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.577459966895765e-06,
+      "loss": 1.0224,
+      "step": 7355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.575225647398e-06,
+      "loss": 0.9066,
+      "step": 7356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.572991349143882e-06,
+      "loss": 1.0052,
+      "step": 7357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.570757072245147e-06,
+      "loss": 0.9314,
+      "step": 7358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.568522816813541e-06,
+      "loss": 0.875,
+      "step": 7359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.566288582960796e-06,
+      "loss": 1.0129,
+      "step": 7360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.564054370798655e-06,
+      "loss": 0.593,
+      "step": 7361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.561820180438853e-06,
+      "loss": 0.913,
+      "step": 7362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.559586011993123e-06,
+      "loss": 0.9427,
+      "step": 7363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.557351865573203e-06,
+      "loss": 0.8798,
+      "step": 7364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.555117741290822e-06,
+      "loss": 0.9545,
+      "step": 7365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.552883639257715e-06,
+      "loss": 0.9131,
+      "step": 7366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.550649559585614e-06,
+      "loss": 0.9373,
+      "step": 7367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.548415502386245e-06,
+      "loss": 1.0375,
+      "step": 7368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.546181467771341e-06,
+      "loss": 0.8903,
+      "step": 7369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.543947455852624e-06,
+      "loss": 0.9861,
+      "step": 7370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.541713466741825e-06,
+      "loss": 0.8928,
+      "step": 7371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.539479500550673e-06,
+      "loss": 0.9805,
+      "step": 7372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.537245557390884e-06,
+      "loss": 0.8836,
+      "step": 7373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.535011637374188e-06,
+      "loss": 0.9898,
+      "step": 7374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.532777740612304e-06,
+      "loss": 0.9613,
+      "step": 7375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.53054386721695e-06,
+      "loss": 0.9159,
+      "step": 7376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.52831001729985e-06,
+      "loss": 0.9993,
+      "step": 7377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.52607619097272e-06,
+      "loss": 0.9763,
+      "step": 7378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.523842388347278e-06,
+      "loss": 0.9464,
+      "step": 7379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.521608609535242e-06,
+      "loss": 0.9305,
+      "step": 7380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.51937485464832e-06,
+      "loss": 0.9375,
+      "step": 7381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.517141123798234e-06,
+      "loss": 0.9843,
+      "step": 7382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.51490741709669e-06,
+      "loss": 0.8995,
+      "step": 7383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.512673734655405e-06,
+      "loss": 0.9576,
+      "step": 7384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.510440076586084e-06,
+      "loss": 0.856,
+      "step": 7385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.50820644300044e-06,
+      "loss": 0.9035,
+      "step": 7386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.505972834010177e-06,
+      "loss": 1.0059,
+      "step": 7387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.503739249727004e-06,
+      "loss": 1.0807,
+      "step": 7388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.501505690262625e-06,
+      "loss": 0.8887,
+      "step": 7389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.499272155728741e-06,
+      "loss": 0.8435,
+      "step": 7390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49703864623706e-06,
+      "loss": 0.8911,
+      "step": 7391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.494805161899282e-06,
+      "loss": 0.8243,
+      "step": 7392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.492571702827104e-06,
+      "loss": 0.9702,
+      "step": 7393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.490338269132227e-06,
+      "loss": 0.8417,
+      "step": 7394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.488104860926348e-06,
+      "loss": 0.8831,
+      "step": 7395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.485871478321166e-06,
+      "loss": 0.9638,
+      "step": 7396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48363812142837e-06,
+      "loss": 1.0375,
+      "step": 7397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.481404790359661e-06,
+      "loss": 0.9816,
+      "step": 7398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.479171485226728e-06,
+      "loss": 0.9904,
+      "step": 7399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.476938206141261e-06,
+      "loss": 0.8384,
+      "step": 7400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.474704953214953e-06,
+      "loss": 0.6445,
+      "step": 7401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.472471726559485e-06,
+      "loss": 0.9265,
+      "step": 7402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.470238526286553e-06,
+      "loss": 1.0093,
+      "step": 7403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46800535250784e-06,
+      "loss": 0.8266,
+      "step": 7404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.465772205335034e-06,
+      "loss": 0.9489,
+      "step": 7405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46353908487981e-06,
+      "loss": 0.9642,
+      "step": 7406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.461305991253856e-06,
+      "loss": 0.8915,
+      "step": 7407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45907292456885e-06,
+      "loss": 0.9011,
+      "step": 7408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.456839884936475e-06,
+      "loss": 0.8973,
+      "step": 7409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.454606872468404e-06,
+      "loss": 0.9503,
+      "step": 7410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.452373887276317e-06,
+      "loss": 0.9163,
+      "step": 7411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45014092947189e-06,
+      "loss": 0.6373,
+      "step": 7412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.44790799916679e-06,
+      "loss": 1.0036,
+      "step": 7413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.445675096472697e-06,
+      "loss": 0.8997,
+      "step": 7414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.443442221501277e-06,
+      "loss": 0.8427,
+      "step": 7415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.441209374364205e-06,
+      "loss": 0.8179,
+      "step": 7416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.438976555173146e-06,
+      "loss": 0.911,
+      "step": 7417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.436743764039765e-06,
+      "loss": 0.9515,
+      "step": 7418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.43451100107573e-06,
+      "loss": 1.01,
+      "step": 7419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.432278266392703e-06,
+      "loss": 0.8794,
+      "step": 7420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.430045560102352e-06,
+      "loss": 0.9461,
+      "step": 7421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.42781288231633e-06,
+      "loss": 1.0105,
+      "step": 7422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.425580233146304e-06,
+      "loss": 0.7906,
+      "step": 7423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.423347612703929e-06,
+      "loss": 0.9544,
+      "step": 7424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.42111502110086e-06,
+      "loss": 0.914,
+      "step": 7425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.418882458448757e-06,
+      "loss": 0.9215,
+      "step": 7426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.416649924859271e-06,
+      "loss": 0.9066,
+      "step": 7427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.414417420444057e-06,
+      "loss": 0.9159,
+      "step": 7428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.41218494531476e-06,
+      "loss": 0.8803,
+      "step": 7429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.40995249958304e-06,
+      "loss": 1.0016,
+      "step": 7430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.407720083360537e-06,
+      "loss": 1.0177,
+      "step": 7431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.405487696758899e-06,
+      "loss": 1.0008,
+      "step": 7432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.40325533988977e-06,
+      "loss": 0.9379,
+      "step": 7433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.4010230128648e-06,
+      "loss": 0.9018,
+      "step": 7434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.398790715795627e-06,
+      "loss": 0.8354,
+      "step": 7435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.396558448793893e-06,
+      "loss": 0.8765,
+      "step": 7436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.394326211971235e-06,
+      "loss": 0.9459,
+      "step": 7437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.392094005439291e-06,
+      "loss": 1.0974,
+      "step": 7438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.3898618293097e-06,
+      "loss": 0.8906,
+      "step": 7439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.387629683694094e-06,
+      "loss": 0.8117,
+      "step": 7440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.38539756870411e-06,
+      "loss": 0.8347,
+      "step": 7441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.383165484451373e-06,
+      "loss": 0.6081,
+      "step": 7442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.380933431047519e-06,
+      "loss": 0.9392,
+      "step": 7443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.378701408604174e-06,
+      "loss": 0.8654,
+      "step": 7444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.376469417232964e-06,
+      "loss": 0.9308,
+      "step": 7445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.37423745704552e-06,
+      "loss": 0.9535,
+      "step": 7446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.372005528153457e-06,
+      "loss": 0.9106,
+      "step": 7447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.369773630668404e-06,
+      "loss": 0.6652,
+      "step": 7448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.36754176470198e-06,
+      "loss": 0.9253,
+      "step": 7449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.365309930365803e-06,
+      "loss": 0.8609,
+      "step": 7450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363078127771495e-06,
+      "loss": 0.8318,
+      "step": 7451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360846357030665e-06,
+      "loss": 0.8886,
+      "step": 7452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358614618254932e-06,
+      "loss": 0.9109,
+      "step": 7453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356382911555907e-06,
+      "loss": 0.8719,
+      "step": 7454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354151237045206e-06,
+      "loss": 1.0258,
+      "step": 7455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.351919594834433e-06,
+      "loss": 1.0206,
+      "step": 7456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.349687985035196e-06,
+      "loss": 0.9366,
+      "step": 7457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.347456407759107e-06,
+      "loss": 0.6391,
+      "step": 7458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.345224863117763e-06,
+      "loss": 0.9536,
+      "step": 7459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.342993351222773e-06,
+      "loss": 0.8788,
+      "step": 7460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.340761872185737e-06,
+      "loss": 0.9991,
+      "step": 7461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.338530426118251e-06,
+      "loss": 1.0693,
+      "step": 7462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.33629901313192e-06,
+      "loss": 1.0243,
+      "step": 7463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.334067633338335e-06,
+      "loss": 0.6078,
+      "step": 7464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.331836286849091e-06,
+      "loss": 0.8326,
+      "step": 7465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.329604973775787e-06,
+      "loss": 0.9738,
+      "step": 7466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.32737369423001e-06,
+      "loss": 0.9272,
+      "step": 7467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.325142448323351e-06,
+      "loss": 0.8761,
+      "step": 7468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.322911236167397e-06,
+      "loss": 0.9258,
+      "step": 7469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.320680057873735e-06,
+      "loss": 0.6368,
+      "step": 7470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.318448913553951e-06,
+      "loss": 0.7798,
+      "step": 7471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.316217803319625e-06,
+      "loss": 1.0108,
+      "step": 7472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.313986727282344e-06,
+      "loss": 0.931,
+      "step": 7473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.311755685553683e-06,
+      "loss": 0.8872,
+      "step": 7474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.309524678245219e-06,
+      "loss": 0.9336,
+      "step": 7475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.307293705468533e-06,
+      "loss": 0.8803,
+      "step": 7476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.305062767335195e-06,
+      "loss": 0.9053,
+      "step": 7477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.30283186395678e-06,
+      "loss": 0.9261,
+      "step": 7478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.300600995444858e-06,
+      "loss": 0.8785,
+      "step": 7479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.298370161911002e-06,
+      "loss": 0.6505,
+      "step": 7480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.296139363466775e-06,
+      "loss": 0.9734,
+      "step": 7481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.293908600223743e-06,
+      "loss": 0.8878,
+      "step": 7482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.291677872293474e-06,
+      "loss": 0.9568,
+      "step": 7483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.289447179787525e-06,
+      "loss": 0.9506,
+      "step": 7484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.287216522817462e-06,
+      "loss": 0.8512,
+      "step": 7485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.28498590149484e-06,
+      "loss": 0.6369,
+      "step": 7486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.282755315931216e-06,
+      "loss": 0.9514,
+      "step": 7487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.280524766238146e-06,
+      "loss": 0.9734,
+      "step": 7488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.278294252527184e-06,
+      "loss": 1.0117,
+      "step": 7489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.276063774909882e-06,
+      "loss": 0.9926,
+      "step": 7490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.273833333497786e-06,
+      "loss": 0.641,
+      "step": 7491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.271602928402449e-06,
+      "loss": 0.6397,
+      "step": 7492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.269372559735414e-06,
+      "loss": 0.6431,
+      "step": 7493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.267142227608225e-06,
+      "loss": 0.9718,
+      "step": 7494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.264911932132428e-06,
+      "loss": 0.8858,
+      "step": 7495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.262681673419557e-06,
+      "loss": 0.8672,
+      "step": 7496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.260451451581157e-06,
+      "loss": 0.6374,
+      "step": 7497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.258221266728764e-06,
+      "loss": 0.9365,
+      "step": 7498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.255991118973912e-06,
+      "loss": 0.9232,
+      "step": 7499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.253761008428133e-06,
+      "loss": 0.6354,
+      "step": 7500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.251530935202963e-06,
+      "loss": 0.9335,
+      "step": 7501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.249300899409924e-06,
+      "loss": 1.0377,
+      "step": 7502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.247070901160551e-06,
+      "loss": 0.9849,
+      "step": 7503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.244840940566365e-06,
+      "loss": 0.9328,
+      "step": 7504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.242611017738895e-06,
+      "loss": 0.9166,
+      "step": 7505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.240381132789657e-06,
+      "loss": 0.901,
+      "step": 7506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.238151285830172e-06,
+      "loss": 0.9391,
+      "step": 7507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.235921476971962e-06,
+      "loss": 0.9459,
+      "step": 7508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.23369170632654e-06,
+      "loss": 0.9558,
+      "step": 7509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.231461974005424e-06,
+      "loss": 0.9248,
+      "step": 7510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.229232280120124e-06,
+      "loss": 0.9346,
+      "step": 7511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.227002624782148e-06,
+      "loss": 0.8555,
+      "step": 7512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.224773008103008e-06,
+      "loss": 0.916,
+      "step": 7513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.222543430194212e-06,
+      "loss": 0.9051,
+      "step": 7514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.220313891167261e-06,
+      "loss": 0.9807,
+      "step": 7515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.218084391133659e-06,
+      "loss": 0.8722,
+      "step": 7516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.215854930204908e-06,
+      "loss": 0.9225,
+      "step": 7517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.213625508492509e-06,
+      "loss": 0.9211,
+      "step": 7518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.211396126107953e-06,
+      "loss": 0.937,
+      "step": 7519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.20916678316274e-06,
+      "loss": 0.9296,
+      "step": 7520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.206937479768358e-06,
+      "loss": 0.8013,
+      "step": 7521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.204708216036306e-06,
+      "loss": 0.9845,
+      "step": 7522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.202478992078065e-06,
+      "loss": 0.997,
+      "step": 7523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.200249808005128e-06,
+      "loss": 0.9435,
+      "step": 7524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.198020663928977e-06,
+      "loss": 0.9286,
+      "step": 7525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.195791559961094e-06,
+      "loss": 0.8832,
+      "step": 7526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.193562496212958e-06,
+      "loss": 0.7874,
+      "step": 7527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.191333472796058e-06,
+      "loss": 0.8695,
+      "step": 7528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.18910448982186e-06,
+      "loss": 0.9919,
+      "step": 7529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.186875547401847e-06,
+      "loss": 0.7887,
+      "step": 7530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.184646645647489e-06,
+      "loss": 0.8925,
+      "step": 7531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.182417784670255e-06,
+      "loss": 0.6388,
+      "step": 7532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.180188964581616e-06,
+      "loss": 0.9963,
+      "step": 7533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.177960185493037e-06,
+      "loss": 0.6011,
+      "step": 7534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.175731447515986e-06,
+      "loss": 0.9013,
+      "step": 7535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.173502750761922e-06,
+      "loss": 0.8474,
+      "step": 7536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.17127409534231e-06,
+      "loss": 0.9862,
+      "step": 7537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.169045481368607e-06,
+      "loss": 0.8512,
+      "step": 7538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.166816908952266e-06,
+      "loss": 0.8883,
+      "step": 7539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.164588378204745e-06,
+      "loss": 0.9374,
+      "step": 7540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.162359889237494e-06,
+      "loss": 0.9482,
+      "step": 7541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.160131442161968e-06,
+      "loss": 0.907,
+      "step": 7542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.157903037089612e-06,
+      "loss": 0.9833,
+      "step": 7543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15567467413187e-06,
+      "loss": 0.646,
+      "step": 7544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15344635340019e-06,
+      "loss": 0.926,
+      "step": 7545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15121807500601e-06,
+      "loss": 0.8609,
+      "step": 7546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.148989839060775e-06,
+      "loss": 0.9799,
+      "step": 7547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.146761645675916e-06,
+      "loss": 0.8862,
+      "step": 7548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.144533494962873e-06,
+      "loss": 0.9271,
+      "step": 7549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.142305387033082e-06,
+      "loss": 0.9336,
+      "step": 7550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.140077321997967e-06,
+      "loss": 0.9584,
+      "step": 7551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.137849299968961e-06,
+      "loss": 0.8992,
+      "step": 7552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.135621321057491e-06,
+      "loss": 0.9213,
+      "step": 7553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.133393385374982e-06,
+      "loss": 0.8911,
+      "step": 7554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.131165493032858e-06,
+      "loss": 0.8906,
+      "step": 7555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.128937644142533e-06,
+      "loss": 0.9369,
+      "step": 7556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.126709838815433e-06,
+      "loss": 0.6708,
+      "step": 7557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.124482077162968e-06,
+      "loss": 0.8862,
+      "step": 7558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.122254359296555e-06,
+      "loss": 0.8692,
+      "step": 7559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.12002668532761e-06,
+      "loss": 0.8891,
+      "step": 7560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.117799055367535e-06,
+      "loss": 1.012,
+      "step": 7561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.115571469527742e-06,
+      "loss": 0.9124,
+      "step": 7562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.113343927919634e-06,
+      "loss": 0.8482,
+      "step": 7563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.111116430654614e-06,
+      "loss": 1.004,
+      "step": 7564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.108888977844085e-06,
+      "loss": 1.0417,
+      "step": 7565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.106661569599442e-06,
+      "loss": 0.9767,
+      "step": 7566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.104434206032084e-06,
+      "loss": 0.9338,
+      "step": 7567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.102206887253404e-06,
+      "loss": 0.9765,
+      "step": 7568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.099979613374792e-06,
+      "loss": 0.775,
+      "step": 7569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.097752384507643e-06,
+      "loss": 1.0385,
+      "step": 7570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.095525200763337e-06,
+      "loss": 0.9342,
+      "step": 7571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.093298062253265e-06,
+      "loss": 0.7331,
+      "step": 7572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.091070969088804e-06,
+      "loss": 0.8322,
+      "step": 7573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.088843921381342e-06,
+      "loss": 0.9051,
+      "step": 7574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.086616919242253e-06,
+      "loss": 0.8476,
+      "step": 7575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.084389962782909e-06,
+      "loss": 0.8514,
+      "step": 7576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.082163052114691e-06,
+      "loss": 0.6233,
+      "step": 7577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.079936187348964e-06,
+      "loss": 0.987,
+      "step": 7578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.077709368597104e-06,
+      "loss": 0.9109,
+      "step": 7579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.075482595970473e-06,
+      "loss": 1.0277,
+      "step": 7580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.073255869580434e-06,
+      "loss": 0.9994,
+      "step": 7581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.071029189538355e-06,
+      "loss": 0.9103,
+      "step": 7582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.068802555955587e-06,
+      "loss": 1.0381,
+      "step": 7583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.066575968943498e-06,
+      "loss": 0.9243,
+      "step": 7584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.064349428613433e-06,
+      "loss": 0.8633,
+      "step": 7585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.062122935076755e-06,
+      "loss": 1.0426,
+      "step": 7586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.059896488444809e-06,
+      "loss": 0.8899,
+      "step": 7587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.057670088828939e-06,
+      "loss": 1.0421,
+      "step": 7588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.055443736340499e-06,
+      "loss": 0.9573,
+      "step": 7589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.053217431090825e-06,
+      "loss": 0.7794,
+      "step": 7590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.050991173191262e-06,
+      "loss": 0.9598,
+      "step": 7591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048764962753152e-06,
+      "loss": 0.917,
+      "step": 7592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046538799887828e-06,
+      "loss": 0.9483,
+      "step": 7593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044312684706622e-06,
+      "loss": 0.8767,
+      "step": 7594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042086617320869e-06,
+      "loss": 0.8481,
+      "step": 7595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039860597841897e-06,
+      "loss": 0.8287,
+      "step": 7596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037634626381032e-06,
+      "loss": 0.9081,
+      "step": 7597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035408703049596e-06,
+      "loss": 0.9016,
+      "step": 7598
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033182827958917e-06,
+      "loss": 0.8301,
+      "step": 7599
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030957001220311e-06,
+      "loss": 0.957,
+      "step": 7600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.028731222945094e-06,
+      "loss": 0.6482,
+      "step": 7601
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.026505493244584e-06,
+      "loss": 0.8546,
+      "step": 7602
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.024279812230092e-06,
+      "loss": 0.9411,
+      "step": 7603
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.022054180012926e-06,
+      "loss": 0.8413,
+      "step": 7604
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.019828596704394e-06,
+      "loss": 0.6327,
+      "step": 7605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.017603062415804e-06,
+      "loss": 0.7627,
+      "step": 7606
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.015377577258458e-06,
+      "loss": 0.8892,
+      "step": 7607
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.01315214134365e-06,
+      "loss": 0.985,
+      "step": 7608
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.010926754782685e-06,
+      "loss": 0.8919,
+      "step": 7609
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.008701417686852e-06,
+      "loss": 0.8856,
+      "step": 7610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.006476130167451e-06,
+      "loss": 0.9327,
+      "step": 7611
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.004250892335767e-06,
+      "loss": 0.6416,
+      "step": 7612
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.002025704303087e-06,
+      "loss": 0.8818,
+      "step": 7613
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.999800566180701e-06,
+      "loss": 0.8636,
+      "step": 7614
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.997575478079885e-06,
+      "loss": 0.936,
+      "step": 7615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.995350440111926e-06,
+      "loss": 0.959,
+      "step": 7616
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.993125452388097e-06,
+      "loss": 0.9113,
+      "step": 7617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.990900515019675e-06,
+      "loss": 1.0831,
+      "step": 7618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.988675628117935e-06,
+      "loss": 0.7934,
+      "step": 7619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.98645079179414e-06,
+      "loss": 0.7986,
+      "step": 7620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.984226006159562e-06,
+      "loss": 0.9363,
+      "step": 7621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.982001271325469e-06,
+      "loss": 0.9287,
+      "step": 7622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.979776587403121e-06,
+      "loss": 0.892,
+      "step": 7623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.977551954503777e-06,
+      "loss": 0.9257,
+      "step": 7624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.975327372738697e-06,
+      "loss": 0.9179,
+      "step": 7625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.973102842219133e-06,
+      "loss": 0.8565,
+      "step": 7626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.970878363056342e-06,
+      "loss": 0.8638,
+      "step": 7627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.968653935361566e-06,
+      "loss": 0.883,
+      "step": 7628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.966429559246058e-06,
+      "loss": 0.8693,
+      "step": 7629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.96420523482106e-06,
+      "loss": 0.9654,
+      "step": 7630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.961980962197819e-06,
+      "loss": 0.9094,
+      "step": 7631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.959756741487568e-06,
+      "loss": 0.8673,
+      "step": 7632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.957532572801546e-06,
+      "loss": 0.6562,
+      "step": 7633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.95530845625099e-06,
+      "loss": 0.8367,
+      "step": 7634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.953084391947124e-06,
+      "loss": 0.925,
+      "step": 7635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.950860380001188e-06,
+      "loss": 0.9187,
+      "step": 7636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.9486364205244e-06,
+      "loss": 0.9739,
+      "step": 7637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.946412513627986e-06,
+      "loss": 0.6096,
+      "step": 7638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.94418865942317e-06,
+      "loss": 0.9679,
+      "step": 7639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.941964858021165e-06,
+      "loss": 0.9024,
+      "step": 7640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.939741109533194e-06,
+      "loss": 0.9771,
+      "step": 7641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.93751741407046e-06,
+      "loss": 0.8519,
+      "step": 7642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.935293771744186e-06,
+      "loss": 0.907,
+      "step": 7643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.933070182665573e-06,
+      "loss": 0.859,
+      "step": 7644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.930846646945825e-06,
+      "loss": 0.9418,
+      "step": 7645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.928623164696147e-06,
+      "loss": 1.0007,
+      "step": 7646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.926399736027737e-06,
+      "loss": 0.5927,
+      "step": 7647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.924176361051796e-06,
+      "loss": 0.9016,
+      "step": 7648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.921953039879516e-06,
+      "loss": 0.9435,
+      "step": 7649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.919729772622087e-06,
+      "loss": 0.7648,
+      "step": 7650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.917506559390703e-06,
+      "loss": 0.8586,
+      "step": 7651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.915283400296544e-06,
+      "loss": 0.8763,
+      "step": 7652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.913060295450796e-06,
+      "loss": 0.9254,
+      "step": 7653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.910837244964646e-06,
+      "loss": 0.9296,
+      "step": 7654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.908614248949266e-06,
+      "loss": 0.9197,
+      "step": 7655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.906391307515834e-06,
+      "loss": 0.9594,
+      "step": 7656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.904168420775523e-06,
+      "loss": 0.9384,
+      "step": 7657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.9019455888395e-06,
+      "loss": 0.7573,
+      "step": 7658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.899722811818937e-06,
+      "loss": 0.8985,
+      "step": 7659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.897500089824994e-06,
+      "loss": 0.9373,
+      "step": 7660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.895277422968838e-06,
+      "loss": 0.8311,
+      "step": 7661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.893054811361624e-06,
+      "loss": 0.9768,
+      "step": 7662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.890832255114508e-06,
+      "loss": 1.0236,
+      "step": 7663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.888609754338648e-06,
+      "loss": 0.9847,
+      "step": 7664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.88638730914519e-06,
+      "loss": 0.6437,
+      "step": 7665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.884164919645286e-06,
+      "loss": 0.9026,
+      "step": 7666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.881942585950078e-06,
+      "loss": 0.9511,
+      "step": 7667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87972030817071e-06,
+      "loss": 0.9751,
+      "step": 7668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.877498086418323e-06,
+      "loss": 0.9666,
+      "step": 7669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87527592080405e-06,
+      "loss": 0.9405,
+      "step": 7670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87305381143903e-06,
+      "loss": 0.9917,
+      "step": 7671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.870831758434388e-06,
+      "loss": 0.939,
+      "step": 7672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86860976190126e-06,
+      "loss": 0.8867,
+      "step": 7673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.866387821950767e-06,
+      "loss": 0.9723,
+      "step": 7674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86416593869403e-06,
+      "loss": 0.9419,
+      "step": 7675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.861944112242173e-06,
+      "loss": 1.0527,
+      "step": 7676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.859722342706311e-06,
+      "loss": 1.0119,
+      "step": 7677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.857500630197559e-06,
+      "loss": 0.874,
+      "step": 7678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.855278974827028e-06,
+      "loss": 0.9965,
+      "step": 7679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.853057376705829e-06,
+      "loss": 0.9465,
+      "step": 7680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.850835835945063e-06,
+      "loss": 0.8738,
+      "step": 7681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.848614352655834e-06,
+      "loss": 0.9212,
+      "step": 7682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.846392926949246e-06,
+      "loss": 0.8396,
+      "step": 7683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.84417155893639e-06,
+      "loss": 0.678,
+      "step": 7684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.841950248728363e-06,
+      "loss": 0.9491,
+      "step": 7685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.839728996436258e-06,
+      "loss": 0.9328,
+      "step": 7686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.837507802171164e-06,
+      "loss": 0.864,
+      "step": 7687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.835286666044163e-06,
+      "loss": 0.9813,
+      "step": 7688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.833065588166339e-06,
+      "loss": 0.9743,
+      "step": 7689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.830844568648771e-06,
+      "loss": 0.9289,
+      "step": 7690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.828623607602539e-06,
+      "loss": 0.6335,
+      "step": 7691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.82640270513871e-06,
+      "loss": 0.9139,
+      "step": 7692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.824181861368364e-06,
+      "loss": 0.8316,
+      "step": 7693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.821961076402563e-06,
+      "loss": 1.0023,
+      "step": 7694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.81974035035237e-06,
+      "loss": 1.0067,
+      "step": 7695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.817519683328853e-06,
+      "loss": 0.9165,
+      "step": 7696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.815299075443067e-06,
+      "loss": 0.8456,
+      "step": 7697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.81307852680607e-06,
+      "loss": 0.8747,
+      "step": 7698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.810858037528914e-06,
+      "loss": 0.8934,
+      "step": 7699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.808637607722652e-06,
+      "loss": 1.0182,
+      "step": 7700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.806417237498327e-06,
+      "loss": 0.8968,
+      "step": 7701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.804196926966984e-06,
+      "loss": 0.9206,
+      "step": 7702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.801976676239669e-06,
+      "loss": 0.9085,
+      "step": 7703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.799756485427412e-06,
+      "loss": 0.8974,
+      "step": 7704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.797536354641259e-06,
+      "loss": 0.8958,
+      "step": 7705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.795316283992235e-06,
+      "loss": 0.8173,
+      "step": 7706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.793096273591367e-06,
+      "loss": 0.9392,
+      "step": 7707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.79087632354969e-06,
+      "loss": 0.6428,
+      "step": 7708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.788656433978217e-06,
+      "loss": 1.0087,
+      "step": 7709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.786436604987978e-06,
+      "loss": 0.97,
+      "step": 7710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.784216836689982e-06,
+      "loss": 0.9594,
+      "step": 7711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.78199712919525e-06,
+      "loss": 0.9362,
+      "step": 7712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.779777482614791e-06,
+      "loss": 0.9037,
+      "step": 7713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77755789705961e-06,
+      "loss": 0.5957,
+      "step": 7714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77533837264071e-06,
+      "loss": 0.9268,
+      "step": 7715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.773118909469103e-06,
+      "loss": 0.679,
+      "step": 7716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77089950765578e-06,
+      "loss": 0.9093,
+      "step": 7717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.76868016731174e-06,
+      "loss": 0.6113,
+      "step": 7718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.766460888547974e-06,
+      "loss": 0.9231,
+      "step": 7719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.764241671475471e-06,
+      "loss": 0.7997,
+      "step": 7720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.762022516205221e-06,
+      "loss": 0.8735,
+      "step": 7721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.759803422848204e-06,
+      "loss": 0.9179,
+      "step": 7722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.757584391515403e-06,
+      "loss": 0.9023,
+      "step": 7723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.75536542231779e-06,
+      "loss": 0.9972,
+      "step": 7724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.753146515366348e-06,
+      "loss": 0.8759,
+      "step": 7725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.750927670772044e-06,
+      "loss": 0.8537,
+      "step": 7726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.748708888645841e-06,
+      "loss": 0.9388,
+      "step": 7727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.746490169098711e-06,
+      "loss": 0.9739,
+      "step": 7728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.744271512241612e-06,
+      "loss": 0.9748,
+      "step": 7729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.742052918185505e-06,
+      "loss": 0.6242,
+      "step": 7730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.739834387041345e-06,
+      "loss": 0.9449,
+      "step": 7731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.737615918920078e-06,
+      "loss": 0.9473,
+      "step": 7732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.735397513932661e-06,
+      "loss": 1.0058,
+      "step": 7733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733179172190037e-06,
+      "loss": 0.9012,
+      "step": 7734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730960893803151e-06,
+      "loss": 0.9053,
+      "step": 7735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728742678882938e-06,
+      "loss": 0.8537,
+      "step": 7736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72652452754034e-06,
+      "loss": 0.8898,
+      "step": 7737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724306439886287e-06,
+      "loss": 0.8206,
+      "step": 7738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722088416031709e-06,
+      "loss": 1.0305,
+      "step": 7739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719870456087534e-06,
+      "loss": 0.6176,
+      "step": 7740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717652560164684e-06,
+      "loss": 0.8148,
+      "step": 7741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715434728374083e-06,
+      "loss": 0.8244,
+      "step": 7742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713216960826645e-06,
+      "loss": 0.9181,
+      "step": 7743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710999257633283e-06,
+      "loss": 0.9272,
+      "step": 7744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.708781618904914e-06,
+      "loss": 0.9986,
+      "step": 7745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.706564044752439e-06,
+      "loss": 0.9658,
+      "step": 7746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.704346535286764e-06,
+      "loss": 0.8604,
+      "step": 7747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.702129090618794e-06,
+      "loss": 0.8425,
+      "step": 7748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.699911710859424e-06,
+      "loss": 0.8361,
+      "step": 7749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.697694396119549e-06,
+      "loss": 0.9796,
+      "step": 7750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.695477146510063e-06,
+      "loss": 0.915,
+      "step": 7751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.693259962141848e-06,
+      "loss": 0.8301,
+      "step": 7752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.691042843125796e-06,
+      "loss": 0.9118,
+      "step": 7753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.688825789572782e-06,
+      "loss": 0.8555,
+      "step": 7754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.68660880159369e-06,
+      "loss": 1.0573,
+      "step": 7755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.684391879299395e-06,
+      "loss": 0.987,
+      "step": 7756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.682175022800761e-06,
+      "loss": 0.9462,
+      "step": 7757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.679958232208667e-06,
+      "loss": 0.9801,
+      "step": 7758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.677741507633971e-06,
+      "loss": 0.6624,
+      "step": 7759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.675524849187539e-06,
+      "loss": 1.0149,
+      "step": 7760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.673308256980227e-06,
+      "loss": 0.9589,
+      "step": 7761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.671091731122891e-06,
+      "loss": 0.8745,
+      "step": 7762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.668875271726387e-06,
+      "loss": 0.8534,
+      "step": 7763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.666658878901555e-06,
+      "loss": 0.8882,
+      "step": 7764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.664442552759249e-06,
+      "loss": 0.8717,
+      "step": 7765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.662226293410305e-06,
+      "loss": 0.923,
+      "step": 7766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.660010100965566e-06,
+      "loss": 0.8285,
+      "step": 7767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.657793975535865e-06,
+      "loss": 0.6381,
+      "step": 7768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.655577917232034e-06,
+      "loss": 1.0129,
+      "step": 7769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.653361926164902e-06,
+      "loss": 0.9938,
+      "step": 7770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.651146002445293e-06,
+      "loss": 0.9135,
+      "step": 7771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.648930146184034e-06,
+      "loss": 0.9474,
+      "step": 7772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.646714357491936e-06,
+      "loss": 0.9446,
+      "step": 7773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.64449863647982e-06,
+      "loss": 0.9483,
+      "step": 7774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.642282983258496e-06,
+      "loss": 0.8544,
+      "step": 7775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.640067397938769e-06,
+      "loss": 0.9027,
+      "step": 7776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.63785188063145e-06,
+      "loss": 1.0071,
+      "step": 7777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.635636431447332e-06,
+      "loss": 0.6515,
+      "step": 7778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.633421050497221e-06,
+      "loss": 0.8525,
+      "step": 7779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.631205737891912e-06,
+      "loss": 0.9394,
+      "step": 7780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.628990493742194e-06,
+      "loss": 0.7995,
+      "step": 7781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.626775318158853e-06,
+      "loss": 0.9906,
+      "step": 7782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.624560211252676e-06,
+      "loss": 0.6528,
+      "step": 7783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.622345173134443e-06,
+      "loss": 0.9191,
+      "step": 7784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.620130203914931e-06,
+      "loss": 0.9016,
+      "step": 7785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.617915303704914e-06,
+      "loss": 0.9283,
+      "step": 7786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.615700472615167e-06,
+      "loss": 0.8862,
+      "step": 7787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.613485710756453e-06,
+      "loss": 1.0039,
+      "step": 7788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.611271018239534e-06,
+      "loss": 0.9437,
+      "step": 7789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.609056395175175e-06,
+      "loss": 1.0134,
+      "step": 7790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.60684184167413e-06,
+      "loss": 0.9709,
+      "step": 7791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.604627357847152e-06,
+      "loss": 0.9007,
+      "step": 7792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.602412943804995e-06,
+      "loss": 0.8057,
+      "step": 7793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.6001985996584e-06,
+      "loss": 1.0052,
+      "step": 7794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.597984325518113e-06,
+      "loss": 0.8177,
+      "step": 7795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.595770121494873e-06,
+      "loss": 0.9539,
+      "step": 7796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.593555987699416e-06,
+      "loss": 0.9497,
+      "step": 7797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.591341924242473e-06,
+      "loss": 0.8187,
+      "step": 7798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.589127931234776e-06,
+      "loss": 1.1223,
+      "step": 7799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.586914008787046e-06,
+      "loss": 0.8929,
+      "step": 7800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.584700157010006e-06,
+      "loss": 0.9439,
+      "step": 7801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.582486376014378e-06,
+      "loss": 1.0135,
+      "step": 7802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.58027266591087e-06,
+      "loss": 0.9446,
+      "step": 7803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.578059026810202e-06,
+      "loss": 0.884,
+      "step": 7804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.575845458823072e-06,
+      "loss": 0.6489,
+      "step": 7805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.573631962060193e-06,
+      "loss": 0.9896,
+      "step": 7806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.57141853663226e-06,
+      "loss": 0.9476,
+      "step": 7807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56920518264997e-06,
+      "loss": 0.9095,
+      "step": 7808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.566991900224014e-06,
+      "loss": 0.9349,
+      "step": 7809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.564778689465091e-06,
+      "loss": 0.9807,
+      "step": 7810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.562565550483881e-06,
+      "loss": 0.8317,
+      "step": 7811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.560352483391067e-06,
+      "loss": 1.0426,
+      "step": 7812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.558139488297328e-06,
+      "loss": 1.0089,
+      "step": 7813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.55592656531334e-06,
+      "loss": 0.9609,
+      "step": 7814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.553713714549776e-06,
+      "loss": 0.6205,
+      "step": 7815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.551500936117299e-06,
+      "loss": 0.8713,
+      "step": 7816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.549288230126582e-06,
+      "loss": 0.9546,
+      "step": 7817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.547075596688277e-06,
+      "loss": 0.9215,
+      "step": 7818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.544863035913049e-06,
+      "loss": 0.8661,
+      "step": 7819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.542650547911548e-06,
+      "loss": 0.9612,
+      "step": 7820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.540438132794421e-06,
+      "loss": 0.9911,
+      "step": 7821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.538225790672323e-06,
+      "loss": 0.8864,
+      "step": 7822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.536013521655888e-06,
+      "loss": 0.924,
+      "step": 7823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.533801325855759e-06,
+      "loss": 0.9352,
+      "step": 7824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.531589203382573e-06,
+      "loss": 0.9263,
+      "step": 7825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.529377154346959e-06,
+      "loss": 0.8991,
+      "step": 7826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.527165178859546e-06,
+      "loss": 0.9777,
+      "step": 7827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.524953277030957e-06,
+      "loss": 0.9426,
+      "step": 7828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.522741448971817e-06,
+      "loss": 0.9045,
+      "step": 7829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.52052969479274e-06,
+      "loss": 0.9509,
+      "step": 7830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.518318014604337e-06,
+      "loss": 0.8691,
+      "step": 7831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.516106408517224e-06,
+      "loss": 0.8849,
+      "step": 7832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.513894876641999e-06,
+      "loss": 0.6094,
+      "step": 7833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.511683419089273e-06,
+      "loss": 0.93,
+      "step": 7834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.509472035969637e-06,
+      "loss": 0.8892,
+      "step": 7835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.50726072739369e-06,
+      "loss": 0.7811,
+      "step": 7836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.505049493472023e-06,
+      "loss": 0.9489,
+      "step": 7837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.50283833431522e-06,
+      "loss": 0.9122,
+      "step": 7838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.500627250033869e-06,
+      "loss": 0.9765,
+      "step": 7839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.498416240738546e-06,
+      "loss": 0.796,
+      "step": 7840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.496205306539826e-06,
+      "loss": 0.8518,
+      "step": 7841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.49399444754829e-06,
+      "loss": 0.9058,
+      "step": 7842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.491783663874497e-06,
+      "loss": 0.985,
+      "step": 7843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.48957295562902e-06,
+      "loss": 0.9506,
+      "step": 7844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.487362322922413e-06,
+      "loss": 0.6513,
+      "step": 7845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.485151765865236e-06,
+      "loss": 0.9501,
+      "step": 7846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.482941284568044e-06,
+      "loss": 0.6173,
+      "step": 7847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.480730879141382e-06,
+      "loss": 0.9538,
+      "step": 7848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.478520549695804e-06,
+      "loss": 0.8989,
+      "step": 7849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.476310296341844e-06,
+      "loss": 0.8788,
+      "step": 7850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.474100119190044e-06,
+      "loss": 0.616,
+      "step": 7851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.471890018350939e-06,
+      "loss": 0.8976,
+      "step": 7852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.469679993935056e-06,
+      "loss": 0.9531,
+      "step": 7853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.467470046052927e-06,
+      "loss": 1.065,
+      "step": 7854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.465260174815072e-06,
+      "loss": 0.929,
+      "step": 7855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.46305038033201e-06,
+      "loss": 0.8703,
+      "step": 7856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.46084066271426e-06,
+      "loss": 0.8573,
+      "step": 7857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.458631022072327e-06,
+      "loss": 0.9841,
+      "step": 7858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.456421458516724e-06,
+      "loss": 0.8476,
+      "step": 7859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.454211972157951e-06,
+      "loss": 0.9466,
+      "step": 7860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.452002563106514e-06,
+      "loss": 0.9894,
+      "step": 7861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.449793231472904e-06,
+      "loss": 0.9043,
+      "step": 7862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.447583977367613e-06,
+      "loss": 1.0125,
+      "step": 7863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.445374800901132e-06,
+      "loss": 0.9072,
+      "step": 7864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.443165702183942e-06,
+      "loss": 0.9425,
+      "step": 7865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.440956681326529e-06,
+      "loss": 1.0046,
+      "step": 7866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.438747738439364e-06,
+      "loss": 0.8597,
+      "step": 7867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.436538873632923e-06,
+      "loss": 0.6085,
+      "step": 7868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.434330087017675e-06,
+      "loss": 0.8579,
+      "step": 7869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.432121378704081e-06,
+      "loss": 0.9217,
+      "step": 7870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.429912748802607e-06,
+      "loss": 0.7995,
+      "step": 7871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.427704197423705e-06,
+      "loss": 0.8128,
+      "step": 7872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.425495724677834e-06,
+      "loss": 0.9359,
+      "step": 7873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.423287330675441e-06,
+      "loss": 0.8935,
+      "step": 7874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.42107901552697e-06,
+      "loss": 0.8452,
+      "step": 7875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418870779342865e-06,
+      "loss": 0.949,
+      "step": 7876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41666262223356e-06,
+      "loss": 0.8402,
+      "step": 7877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41445454430949e-06,
+      "loss": 0.8533,
+      "step": 7878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412246545681086e-06,
+      "loss": 0.8994,
+      "step": 7879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410038626458771e-06,
+      "loss": 0.916,
+      "step": 7880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407830786752968e-06,
+      "loss": 0.9088,
+      "step": 7881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405623026674097e-06,
+      "loss": 0.9182,
+      "step": 7882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.403415346332566e-06,
+      "loss": 0.865,
+      "step": 7883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40120774583879e-06,
+      "loss": 0.9967,
+      "step": 7884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399000225303168e-06,
+      "loss": 0.989,
+      "step": 7885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396792784836108e-06,
+      "loss": 0.8067,
+      "step": 7886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394585424548005e-06,
+      "loss": 0.896,
+      "step": 7887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.392378144549257e-06,
+      "loss": 0.8995,
+      "step": 7888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390170944950247e-06,
+      "loss": 1.0324,
+      "step": 7889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.387963825861361e-06,
+      "loss": 1.0555,
+      "step": 7890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.385756787392986e-06,
+      "loss": 0.9195,
+      "step": 7891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.383549829655495e-06,
+      "loss": 0.9331,
+      "step": 7892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.381342952759263e-06,
+      "loss": 0.8804,
+      "step": 7893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.37913615681466e-06,
+      "loss": 0.9671,
+      "step": 7894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.37692944193205e-06,
+      "loss": 0.894,
+      "step": 7895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.374722808221796e-06,
+      "loss": 0.8921,
+      "step": 7896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.372516255794252e-06,
+      "loss": 0.9115,
+      "step": 7897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.370309784759776e-06,
+      "loss": 0.6073,
+      "step": 7898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.368103395228711e-06,
+      "loss": 0.8793,
+      "step": 7899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.36589708731141e-06,
+      "loss": 1.0041,
+      "step": 7900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.36369086111821e-06,
+      "loss": 0.9599,
+      "step": 7901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.361484716759445e-06,
+      "loss": 0.9004,
+      "step": 7902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.359278654345447e-06,
+      "loss": 0.938,
+      "step": 7903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.357072673986554e-06,
+      "loss": 0.9633,
+      "step": 7904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.354866775793083e-06,
+      "loss": 0.8971,
+      "step": 7905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.352660959875357e-06,
+      "loss": 0.9194,
+      "step": 7906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.350455226343693e-06,
+      "loss": 0.9481,
+      "step": 7907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.348249575308398e-06,
+      "loss": 0.8412,
+      "step": 7908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.346044006879788e-06,
+      "loss": 0.9271,
+      "step": 7909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.34383852116816e-06,
+      "loss": 0.9736,
+      "step": 7910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.341633118283818e-06,
+      "loss": 0.9897,
+      "step": 7911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.339427798337056e-06,
+      "loss": 1.013,
+      "step": 7912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.337222561438168e-06,
+      "loss": 0.6001,
+      "step": 7913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.33501740769744e-06,
+      "loss": 0.9933,
+      "step": 7914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.332812337225152e-06,
+      "loss": 0.8952,
+      "step": 7915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.33060735013159e-06,
+      "loss": 0.9036,
+      "step": 7916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.32840244652702e-06,
+      "loss": 0.8994,
+      "step": 7917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.326197626521723e-06,
+      "loss": 1.0489,
+      "step": 7918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.323992890225958e-06,
+      "loss": 0.9021,
+      "step": 7919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.321788237749988e-06,
+      "loss": 0.9608,
+      "step": 7920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.319583669204076e-06,
+      "loss": 0.8688,
+      "step": 7921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.31737918469847e-06,
+      "loss": 0.9286,
+      "step": 7922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.315174784343426e-06,
+      "loss": 0.8925,
+      "step": 7923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.312970468249183e-06,
+      "loss": 0.9704,
+      "step": 7924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.310766236525987e-06,
+      "loss": 0.8916,
+      "step": 7925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.308562089284074e-06,
+      "loss": 0.8477,
+      "step": 7926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.306358026633674e-06,
+      "loss": 0.9277,
+      "step": 7927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.304154048685023e-06,
+      "loss": 0.8905,
+      "step": 7928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.301950155548337e-06,
+      "loss": 0.9817,
+      "step": 7929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.299746347333843e-06,
+      "loss": 0.8735,
+      "step": 7930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.297542624151755e-06,
+      "loss": 0.9,
+      "step": 7931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.295338986112278e-06,
+      "loss": 0.8586,
+      "step": 7932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.293135433325631e-06,
+      "loss": 1.0168,
+      "step": 7933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.290931965902007e-06,
+      "loss": 0.7643,
+      "step": 7934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.28872858395161e-06,
+      "loss": 0.95,
+      "step": 7935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.286525287584638e-06,
+      "loss": 0.9394,
+      "step": 7936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.284322076911272e-06,
+      "loss": 0.9331,
+      "step": 7937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.28211895204171e-06,
+      "loss": 0.9457,
+      "step": 7938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.279915913086124e-06,
+      "loss": 0.8709,
+      "step": 7939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.277712960154694e-06,
+      "loss": 0.8883,
+      "step": 7940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.275510093357594e-06,
+      "loss": 0.6418,
+      "step": 7941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.273307312804995e-06,
+      "loss": 0.8331,
+      "step": 7942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.271104618607059e-06,
+      "loss": 0.9852,
+      "step": 7943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.268902010873947e-06,
+      "loss": 0.9636,
+      "step": 7944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.266699489715812e-06,
+      "loss": 0.8913,
+      "step": 7945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.26449705524281e-06,
+      "loss": 0.9543,
+      "step": 7946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.262294707565085e-06,
+      "loss": 0.9724,
+      "step": 7947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.260092446792785e-06,
+      "loss": 0.8409,
+      "step": 7948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.257890273036043e-06,
+      "loss": 0.8702,
+      "step": 7949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.255688186404996e-06,
+      "loss": 0.9329,
+      "step": 7950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.253486187009774e-06,
+      "loss": 1.077,
+      "step": 7951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.2512842749605e-06,
+      "loss": 1.0065,
+      "step": 7952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.249082450367299e-06,
+      "loss": 0.9756,
+      "step": 7953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.246880713340282e-06,
+      "loss": 0.8921,
+      "step": 7954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.244679063989567e-06,
+      "loss": 0.9171,
+      "step": 7955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.242477502425258e-06,
+      "loss": 0.8775,
+      "step": 7956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.240276028757464e-06,
+      "loss": 0.9042,
+      "step": 7957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.238074643096281e-06,
+      "loss": 1.0109,
+      "step": 7958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.235873345551802e-06,
+      "loss": 0.8596,
+      "step": 7959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.233672136234118e-06,
+      "loss": 0.96,
+      "step": 7960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.231471015253317e-06,
+      "loss": 0.8823,
+      "step": 7961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.229269982719482e-06,
+      "loss": 0.9275,
+      "step": 7962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.227069038742685e-06,
+      "loss": 0.9911,
+      "step": 7963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.224868183433002e-06,
+      "loss": 1.0068,
+      "step": 7964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.222667416900502e-06,
+      "loss": 0.8826,
+      "step": 7965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.220466739255244e-06,
+      "loss": 0.8943,
+      "step": 7966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.218266150607292e-06,
+      "loss": 0.9715,
+      "step": 7967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.216065651066703e-06,
+      "loss": 0.8393,
+      "step": 7968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.213865240743523e-06,
+      "loss": 0.8108,
+      "step": 7969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.211664919747801e-06,
+      "loss": 0.8864,
+      "step": 7970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.209464688189577e-06,
+      "loss": 0.9761,
+      "step": 7971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.207264546178888e-06,
+      "loss": 0.6705,
+      "step": 7972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.205064493825767e-06,
+      "loss": 0.9535,
+      "step": 7973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.20286453124024e-06,
+      "loss": 0.9458,
+      "step": 7974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.200664658532335e-06,
+      "loss": 0.8069,
+      "step": 7975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.198464875812068e-06,
+      "loss": 0.8437,
+      "step": 7976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.196265183189454e-06,
+      "loss": 0.9932,
+      "step": 7977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.194065580774506e-06,
+      "loss": 0.8505,
+      "step": 7978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.191866068677224e-06,
+      "loss": 0.9415,
+      "step": 7979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.189666647007615e-06,
+      "loss": 0.8378,
+      "step": 7980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18746731587567e-06,
+      "loss": 0.9697,
+      "step": 7981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.185268075391389e-06,
+      "loss": 0.9322,
+      "step": 7982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.183068925664751e-06,
+      "loss": 0.8715,
+      "step": 7983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.180869866805742e-06,
+      "loss": 1.006,
+      "step": 7984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.178670898924344e-06,
+      "loss": 0.8676,
+      "step": 7985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.176472022130523e-06,
+      "loss": 0.8644,
+      "step": 7986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.174273236534259e-06,
+      "loss": 0.9111,
+      "step": 7987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.17207454224551e-06,
+      "loss": 0.8103,
+      "step": 7988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.169875939374234e-06,
+      "loss": 0.8599,
+      "step": 7989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.167677428030392e-06,
+      "loss": 0.9062,
+      "step": 7990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.165479008323933e-06,
+      "loss": 0.9377,
+      "step": 7991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163280680364804e-06,
+      "loss": 1.0457,
+      "step": 7992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.161082444262944e-06,
+      "loss": 0.9105,
+      "step": 7993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.158884300128295e-06,
+      "loss": 0.9907,
+      "step": 7994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15668624807079e-06,
+      "loss": 0.8938,
+      "step": 7995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.154488288200348e-06,
+      "loss": 0.9424,
+      "step": 7996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1522904206269e-06,
+      "loss": 0.8435,
+      "step": 7997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.150092645460366e-06,
+      "loss": 0.9566,
+      "step": 7998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.147894962810658e-06,
+      "loss": 0.9385,
+      "step": 7999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.145697372787686e-06,
+      "loss": 0.5905,
+      "step": 8000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.143499875501357e-06,
+      "loss": 1.0442,
+      "step": 8001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.141302471061565e-06,
+      "loss": 0.9133,
+      "step": 8002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.139105159578212e-06,
+      "loss": 0.8441,
+      "step": 8003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.136907941161185e-06,
+      "loss": 0.9309,
+      "step": 8004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.134710815920374e-06,
+      "loss": 0.911,
+      "step": 8005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.132513783965658e-06,
+      "loss": 0.9629,
+      "step": 8006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.130316845406916e-06,
+      "loss": 0.9736,
+      "step": 8007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.128120000354022e-06,
+      "loss": 0.9449,
+      "step": 8008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.125923248916838e-06,
+      "loss": 0.9578,
+      "step": 8009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.12372659120523e-06,
+      "loss": 0.9477,
+      "step": 8010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.121530027329058e-06,
+      "loss": 0.894,
+      "step": 8011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.119333557398175e-06,
+      "loss": 0.9498,
+      "step": 8012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.117137181522429e-06,
+      "loss": 0.8439,
+      "step": 8013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.114940899811662e-06,
+      "loss": 0.9715,
+      "step": 8014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.11274471237572e-06,
+      "loss": 0.8882,
+      "step": 8015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.11054861932443e-06,
+      "loss": 0.919,
+      "step": 8016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.108352620767629e-06,
+      "loss": 0.9498,
+      "step": 8017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106156716815137e-06,
+      "loss": 0.9616,
+      "step": 8018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10396090757678e-06,
+      "loss": 0.9982,
+      "step": 8019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10176519316237e-06,
+      "loss": 0.8625,
+      "step": 8020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09956957368172e-06,
+      "loss": 0.8994,
+      "step": 8021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097374049244636e-06,
+      "loss": 0.8701,
+      "step": 8022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095178619960915e-06,
+      "loss": 0.8294,
+      "step": 8023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092983285940363e-06,
+      "loss": 0.8374,
+      "step": 8024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090788047292768e-06,
+      "loss": 0.9331,
+      "step": 8025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088592904127915e-06,
+      "loss": 0.8756,
+      "step": 8026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08639785655559e-06,
+      "loss": 0.9864,
+      "step": 8027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084202904685566e-06,
+      "loss": 0.9146,
+      "step": 8028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08200804862762e-06,
+      "loss": 0.8914,
+      "step": 8029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079813288491521e-06,
+      "loss": 1.0313,
+      "step": 8030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077618624387032e-06,
+      "loss": 0.9019,
+      "step": 8031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07542405642391e-06,
+      "loss": 1.0046,
+      "step": 8032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07322958471191e-06,
+      "loss": 0.6547,
+      "step": 8033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071035209360781e-06,
+      "loss": 0.6567,
+      "step": 8034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.068840930480269e-06,
+      "loss": 0.9659,
+      "step": 8035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.066646748180107e-06,
+      "loss": 0.9104,
+      "step": 8036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.064452662570036e-06,
+      "loss": 0.935,
+      "step": 8037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.062258673759787e-06,
+      "loss": 1.0032,
+      "step": 8038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.060064781859075e-06,
+      "loss": 0.6271,
+      "step": 8039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.057870986977631e-06,
+      "loss": 0.8678,
+      "step": 8040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.055677289225165e-06,
+      "loss": 0.8789,
+      "step": 8041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.053483688711388e-06,
+      "loss": 0.8804,
+      "step": 8042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.051290185546004e-06,
+      "loss": 0.8756,
+      "step": 8043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.04909677983872e-06,
+      "loss": 0.8395,
+      "step": 8044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.046903471699224e-06,
+      "loss": 0.8599,
+      "step": 8045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.044710261237207e-06,
+      "loss": 1.0189,
+      "step": 8046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.042517148562362e-06,
+      "loss": 1.0351,
+      "step": 8047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.040324133784362e-06,
+      "loss": 0.8921,
+      "step": 8048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.03813121701289e-06,
+      "loss": 0.6362,
+      "step": 8049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.035938398357612e-06,
+      "loss": 0.9723,
+      "step": 8050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.033745677928198e-06,
+      "loss": 0.6312,
+      "step": 8051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.031553055834309e-06,
+      "loss": 0.9776,
+      "step": 8052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.029360532185599e-06,
+      "loss": 0.9585,
+      "step": 8053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.027168107091723e-06,
+      "loss": 1.0012,
+      "step": 8054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.024975780662322e-06,
+      "loss": 0.876,
+      "step": 8055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.022783553007046e-06,
+      "loss": 0.9328,
+      "step": 8056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.020591424235526e-06,
+      "loss": 0.9804,
+      "step": 8057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.018399394457394e-06,
+      "loss": 0.9422,
+      "step": 8058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.016207463782282e-06,
+      "loss": 1.01,
+      "step": 8059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.014015632319801e-06,
+      "loss": 0.9671,
+      "step": 8060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.01182390017958e-06,
+      "loss": 0.9681,
+      "step": 8061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.009632267471227e-06,
+      "loss": 0.9414,
+      "step": 8062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.007440734304346e-06,
+      "loss": 0.945,
+      "step": 8063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.005249300788543e-06,
+      "loss": 0.8966,
+      "step": 8064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.003057967033415e-06,
+      "loss": 0.8609,
+      "step": 8065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.00086673314855e-06,
+      "loss": 0.6445,
+      "step": 8066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.99867559924354e-06,
+      "loss": 0.863,
+      "step": 8067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.996484565427962e-06,
+      "loss": 0.8847,
+      "step": 8068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.994293631811398e-06,
+      "loss": 0.9314,
+      "step": 8069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.992102798503419e-06,
+      "loss": 1.0906,
+      "step": 8070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.989912065613587e-06,
+      "loss": 0.8431,
+      "step": 8071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.987721433251471e-06,
+      "loss": 0.8366,
+      "step": 8072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.985530901526626e-06,
+      "loss": 0.9688,
+      "step": 8073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.983340470548601e-06,
+      "loss": 0.8767,
+      "step": 8074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.981150140426947e-06,
+      "loss": 0.8925,
+      "step": 8075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.978959911271203e-06,
+      "loss": 0.9657,
+      "step": 8076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.976769783190908e-06,
+      "loss": 0.9241,
+      "step": 8077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.97457975629559e-06,
+      "loss": 1.0526,
+      "step": 8078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.972389830694783e-06,
+      "loss": 0.9112,
+      "step": 8079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.970200006498e-06,
+      "loss": 0.9904,
+      "step": 8080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.968010283814765e-06,
+      "loss": 1.0101,
+      "step": 8081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.965820662754584e-06,
+      "loss": 0.8445,
+      "step": 8082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.963631143426967e-06,
+      "loss": 0.7991,
+      "step": 8083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.961441725941414e-06,
+      "loss": 0.8043,
+      "step": 8084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.959252410407418e-06,
+      "loss": 1.039,
+      "step": 8085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.957063196934478e-06,
+      "loss": 0.8706,
+      "step": 8086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.954874085632074e-06,
+      "loss": 0.6549,
+      "step": 8087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.952685076609688e-06,
+      "loss": 0.8933,
+      "step": 8088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.950496169976799e-06,
+      "loss": 0.9873,
+      "step": 8089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.948307365842872e-06,
+      "loss": 0.8512,
+      "step": 8090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.946118664317377e-06,
+      "loss": 0.9152,
+      "step": 8091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.943930065509771e-06,
+      "loss": 0.9134,
+      "step": 8092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.941741569529512e-06,
+      "loss": 1.0639,
+      "step": 8093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.939553176486053e-06,
+      "loss": 0.9983,
+      "step": 8094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.937364886488834e-06,
+      "loss": 0.9366,
+      "step": 8095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.935176699647296e-06,
+      "loss": 0.8524,
+      "step": 8096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.932988616070875e-06,
+      "loss": 0.9371,
+      "step": 8097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.930800635869001e-06,
+      "loss": 0.8978,
+      "step": 8098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.928612759151097e-06,
+      "loss": 1.0054,
+      "step": 8099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.926424986026581e-06,
+      "loss": 0.9224,
+      "step": 8100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.92423731660487e-06,
+      "loss": 0.9379,
+      "step": 8101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.922049750995371e-06,
+      "loss": 0.8269,
+      "step": 8102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.919862289307487e-06,
+      "loss": 0.8757,
+      "step": 8103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.917674931650621e-06,
+      "loss": 0.8977,
+      "step": 8104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.915487678134158e-06,
+      "loss": 0.8392,
+      "step": 8105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.913300528867495e-06,
+      "loss": 0.9452,
+      "step": 8106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.91111348396001e-06,
+      "loss": 0.9101,
+      "step": 8107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.908926543521079e-06,
+      "loss": 0.9319,
+      "step": 8108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.906739707660078e-06,
+      "loss": 0.8718,
+      "step": 8109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90455297648637e-06,
+      "loss": 0.92,
+      "step": 8110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.902366350109325e-06,
+      "loss": 0.8947,
+      "step": 8111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90017982863829e-06,
+      "loss": 0.9647,
+      "step": 8112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.897993412182624e-06,
+      "loss": 0.9516,
+      "step": 8113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.89580710085167e-06,
+      "loss": 0.9297,
+      "step": 8114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.893620894754766e-06,
+      "loss": 1.0338,
+      "step": 8115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.891434794001253e-06,
+      "loss": 1.0079,
+      "step": 8116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.889248798700458e-06,
+      "loss": 0.9023,
+      "step": 8117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.887062908961708e-06,
+      "loss": 1.0286,
+      "step": 8118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.884877124894324e-06,
+      "loss": 0.9693,
+      "step": 8119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.882691446607614e-06,
+      "loss": 0.9318,
+      "step": 8120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.880505874210896e-06,
+      "loss": 0.9109,
+      "step": 8121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.878320407813467e-06,
+      "loss": 0.6295,
+      "step": 8122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.876135047524625e-06,
+      "loss": 0.9383,
+      "step": 8123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.873949793453673e-06,
+      "loss": 0.825,
+      "step": 8124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.87176464570989e-06,
+      "loss": 0.9325,
+      "step": 8125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.869579604402563e-06,
+      "loss": 0.9056,
+      "step": 8126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.867394669640969e-06,
+      "loss": 0.8454,
+      "step": 8127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.865209841534375e-06,
+      "loss": 0.8513,
+      "step": 8128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.863025120192057e-06,
+      "loss": 0.9491,
+      "step": 8129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.860840505723268e-06,
+      "loss": 0.8716,
+      "step": 8130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85865599823727e-06,
+      "loss": 0.8535,
+      "step": 8131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.856471597843309e-06,
+      "loss": 0.9958,
+      "step": 8132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.854287304650632e-06,
+      "loss": 0.8759,
+      "step": 8133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.852103118768482e-06,
+      "loss": 0.791,
+      "step": 8134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.849919040306086e-06,
+      "loss": 0.8711,
+      "step": 8135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.847735069372685e-06,
+      "loss": 0.8505,
+      "step": 8136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.84555120607749e-06,
+      "loss": 0.8988,
+      "step": 8137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.84336745052973e-06,
+      "loss": 0.8538,
+      "step": 8138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.841183802838612e-06,
+      "loss": 0.8883,
+      "step": 8139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.839000263113345e-06,
+      "loss": 0.9092,
+      "step": 8140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.836816831463133e-06,
+      "loss": 0.8217,
+      "step": 8141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.83463350799717e-06,
+      "loss": 0.928,
+      "step": 8142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.832450292824651e-06,
+      "loss": 0.8778,
+      "step": 8143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.83026718605476e-06,
+      "loss": 0.9308,
+      "step": 8144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.828084187796678e-06,
+      "loss": 0.839,
+      "step": 8145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.825901298159582e-06,
+      "loss": 0.8258,
+      "step": 8146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.823718517252636e-06,
+      "loss": 0.9203,
+      "step": 8147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.821535845185011e-06,
+      "loss": 0.8796,
+      "step": 8148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.819353282065863e-06,
+      "loss": 0.9878,
+      "step": 8149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.817170828004345e-06,
+      "loss": 0.8143,
+      "step": 8150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.814988483109606e-06,
+      "loss": 0.914,
+      "step": 8151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.812806247490788e-06,
+      "loss": 0.8818,
+      "step": 8152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.81062412125703e-06,
+      "loss": 0.9467,
+      "step": 8153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.808442104517458e-06,
+      "loss": 0.8828,
+      "step": 8154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.806260197381205e-06,
+      "loss": 0.9449,
+      "step": 8155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.804078399957389e-06,
+      "loss": 0.9565,
+      "step": 8156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.801896712355123e-06,
+      "loss": 0.8505,
+      "step": 8157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.799715134683523e-06,
+      "loss": 0.9607,
+      "step": 8158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.797533667051688e-06,
+      "loss": 0.9211,
+      "step": 8159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795352309568716e-06,
+      "loss": 0.8649,
+      "step": 8160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793171062343703e-06,
+      "loss": 0.9725,
+      "step": 8161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790989925485732e-06,
+      "loss": 0.9001,
+      "step": 8162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788808899103894e-06,
+      "loss": 0.6377,
+      "step": 8163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786627983307258e-06,
+      "loss": 0.9171,
+      "step": 8164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784447178204897e-06,
+      "loss": 0.9825,
+      "step": 8165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782266483905876e-06,
+      "loss": 0.8076,
+      "step": 8166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780085900519255e-06,
+      "loss": 0.8754,
+      "step": 8167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777905428154092e-06,
+      "loss": 0.9317,
+      "step": 8168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.775725066919432e-06,
+      "loss": 0.6018,
+      "step": 8169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77354481692432e-06,
+      "loss": 0.8501,
+      "step": 8170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771364678277794e-06,
+      "loss": 0.8499,
+      "step": 8171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769184651088883e-06,
+      "loss": 0.886,
+      "step": 8172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76700473546662e-06,
+      "loss": 0.8866,
+      "step": 8173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764824931520018e-06,
+      "loss": 0.9228,
+      "step": 8174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762645239358101e-06,
+      "loss": 0.8543,
+      "step": 8175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760465659089874e-06,
+      "loss": 0.9219,
+      "step": 8176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758286190824341e-06,
+      "loss": 0.8665,
+      "step": 8177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756106834670501e-06,
+      "loss": 0.8561,
+      "step": 8178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753927590737346e-06,
+      "loss": 0.9357,
+      "step": 8179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75174845913387e-06,
+      "loss": 0.9818,
+      "step": 8180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.749569439969046e-06,
+      "loss": 0.9476,
+      "step": 8181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.747390533351857e-06,
+      "loss": 0.8487,
+      "step": 8182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.745211739391272e-06,
+      "loss": 0.8396,
+      "step": 8183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.743033058196251e-06,
+      "loss": 0.9793,
+      "step": 8184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.74085448987576e-06,
+      "loss": 0.8659,
+      "step": 8185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.738676034538746e-06,
+      "loss": 0.9449,
+      "step": 8186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.736497692294163e-06,
+      "loss": 0.6183,
+      "step": 8187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.734319463250954e-06,
+      "loss": 0.9018,
+      "step": 8188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.732141347518052e-06,
+      "loss": 0.9257,
+      "step": 8189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.729963345204386e-06,
+      "loss": 0.8975,
+      "step": 8190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.727785456418889e-06,
+      "loss": 0.878,
+      "step": 8191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.725607681270473e-06,
+      "loss": 0.8648,
+      "step": 8192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.723430019868057e-06,
+      "loss": 0.8701,
+      "step": 8193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.721252472320546e-06,
+      "loss": 0.9303,
+      "step": 8194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.719075038736848e-06,
+      "loss": 0.9309,
+      "step": 8195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.716897719225853e-06,
+      "loss": 0.8752,
+      "step": 8196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.714720513896455e-06,
+      "loss": 0.9525,
+      "step": 8197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.712543422857542e-06,
+      "loss": 0.9329,
+      "step": 8198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.710366446217989e-06,
+      "loss": 0.8889,
+      "step": 8199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.708189584086675e-06,
+      "loss": 0.8983,
+      "step": 8200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.706012836572466e-06,
+      "loss": 0.9413,
+      "step": 8201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.703836203784224e-06,
+      "loss": 0.9132,
+      "step": 8202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.701659685830806e-06,
+      "loss": 0.8869,
+      "step": 8203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.699483282821064e-06,
+      "loss": 1.0236,
+      "step": 8204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.697306994863845e-06,
+      "loss": 0.6136,
+      "step": 8205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.695130822067985e-06,
+      "loss": 0.8407,
+      "step": 8206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.69295476454232e-06,
+      "loss": 0.8698,
+      "step": 8207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.690778822395678e-06,
+      "loss": 0.811,
+      "step": 8208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.688602995736881e-06,
+      "loss": 1.0157,
+      "step": 8209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.686427284674747e-06,
+      "loss": 0.9095,
+      "step": 8210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.684251689318084e-06,
+      "loss": 0.8127,
+      "step": 8211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6820762097757e-06,
+      "loss": 0.9078,
+      "step": 8212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.679900846156391e-06,
+      "loss": 0.8802,
+      "step": 8213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.677725598568952e-06,
+      "loss": 0.9501,
+      "step": 8214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.675550467122173e-06,
+      "loss": 1.0172,
+      "step": 8215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.673375451924832e-06,
+      "loss": 0.8646,
+      "step": 8216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.671200553085703e-06,
+      "loss": 0.968,
+      "step": 8217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.669025770713565e-06,
+      "loss": 0.874,
+      "step": 8218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.666851104917174e-06,
+      "loss": 0.8252,
+      "step": 8219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.664676555805296e-06,
+      "loss": 0.8497,
+      "step": 8220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.662502123486677e-06,
+      "loss": 0.909,
+      "step": 8221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.660327808070064e-06,
+      "loss": 0.9384,
+      "step": 8222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.658153609664202e-06,
+      "loss": 0.8712,
+      "step": 8223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.655979528377824e-06,
+      "loss": 0.6217,
+      "step": 8224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.653805564319662e-06,
+      "loss": 0.882,
+      "step": 8225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.651631717598432e-06,
+      "loss": 0.8516,
+      "step": 8226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.64945798832286e-06,
+      "loss": 0.9946,
+      "step": 8227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.647284376601655e-06,
+      "loss": 0.878,
+      "step": 8228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.645110882543517e-06,
+      "loss": 0.8946,
+      "step": 8229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.642937506257156e-06,
+      "loss": 0.7892,
+      "step": 8230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.640764247851259e-06,
+      "loss": 0.8867,
+      "step": 8231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.638591107434517e-06,
+      "loss": 0.8213,
+      "step": 8232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.636418085115611e-06,
+      "loss": 0.9891,
+      "step": 8233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.634245181003216e-06,
+      "loss": 0.9303,
+      "step": 8234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.632072395206008e-06,
+      "loss": 0.5984,
+      "step": 8235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.629899727832642e-06,
+      "loss": 0.8373,
+      "step": 8236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.627727178991786e-06,
+      "loss": 0.933,
+      "step": 8237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.625554748792086e-06,
+      "loss": 0.998,
+      "step": 8238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.623382437342194e-06,
+      "loss": 0.9734,
+      "step": 8239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.621210244750748e-06,
+      "loss": 0.9599,
+      "step": 8240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.619038171126381e-06,
+      "loss": 0.876,
+      "step": 8241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.616866216577727e-06,
+      "loss": 1.0221,
+      "step": 8242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6146943812134014e-06,
+      "loss": 0.9842,
+      "step": 8243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6125226651420295e-06,
+      "loss": 0.8592,
+      "step": 8244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6103510684722174e-06,
+      "loss": 0.7952,
+      "step": 8245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.60817959131257e-06,
+      "loss": 0.6191,
+      "step": 8246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.606008233771689e-06,
+      "loss": 0.8738,
+      "step": 8247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6038369959581615e-06,
+      "loss": 0.8225,
+      "step": 8248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6016658779805814e-06,
+      "loss": 0.8254,
+      "step": 8249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.59949487994753e-06,
+      "loss": 0.9542,
+      "step": 8250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.597324001967576e-06,
+      "loss": 0.9926,
+      "step": 8251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.595153244149296e-06,
+      "loss": 0.9224,
+      "step": 8252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.592982606601249e-06,
+      "loss": 0.8981,
+      "step": 8253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5908120894319906e-06,
+      "loss": 0.9073,
+      "step": 8254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.588641692750075e-06,
+      "loss": 0.8204,
+      "step": 8255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.586471416664044e-06,
+      "loss": 0.8541,
+      "step": 8256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.58430126128244e-06,
+      "loss": 0.8336,
+      "step": 8257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.582131226713795e-06,
+      "loss": 0.8521,
+      "step": 8258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.579961313066633e-06,
+      "loss": 0.9382,
+      "step": 8259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.577791520449479e-06,
+      "loss": 1.0013,
+      "step": 8260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.575621848970844e-06,
+      "loss": 1.0183,
+      "step": 8261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.573452298739241e-06,
+      "loss": 0.8947,
+      "step": 8262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.571282869863168e-06,
+      "loss": 0.8483,
+      "step": 8263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.569113562451125e-06,
+      "loss": 0.9734,
+      "step": 8264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.566944376611602e-06,
+      "loss": 0.8899,
+      "step": 8265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.564775312453082e-06,
+      "loss": 0.8935,
+      "step": 8266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5626063700840446e-06,
+      "loss": 0.9637,
+      "step": 8267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5604375496129615e-06,
+      "loss": 0.8054,
+      "step": 8268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5582688511483e-06,
+      "loss": 0.8754,
+      "step": 8269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5561002747985195e-06,
+      "loss": 0.9435,
+      "step": 8270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.553931820672071e-06,
+      "loss": 0.9105,
+      "step": 8271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.551763488877409e-06,
+      "loss": 0.906,
+      "step": 8272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5495952795229675e-06,
+      "loss": 0.8838,
+      "step": 8273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.547427192717189e-06,
+      "loss": 0.9059,
+      "step": 8274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.545259228568499e-06,
+      "loss": 0.9074,
+      "step": 8275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.543091387185323e-06,
+      "loss": 0.8501,
+      "step": 8276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.540923668676077e-06,
+      "loss": 0.9105,
+      "step": 8277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.538756073149169e-06,
+      "loss": 0.9653,
+      "step": 8278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5365886007130104e-06,
+      "loss": 0.8805,
+      "step": 8279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.534421251475992e-06,
+      "loss": 0.9497,
+      "step": 8280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.532254025546513e-06,
+      "loss": 0.9186,
+      "step": 8281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.53008692303296e-06,
+      "loss": 0.9059,
+      "step": 8282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.527919944043711e-06,
+      "loss": 0.6521,
+      "step": 8283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.525753088687139e-06,
+      "loss": 0.8065,
+      "step": 8284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.523586357071615e-06,
+      "loss": 0.8582,
+      "step": 8285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.521419749305496e-06,
+      "loss": 0.8595,
+      "step": 8286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.519253265497143e-06,
+      "loss": 0.9107,
+      "step": 8287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5170869057549e-06,
+      "loss": 0.8764,
+      "step": 8288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.514920670187116e-06,
+      "loss": 0.9335,
+      "step": 8289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.512754558902123e-06,
+      "loss": 0.9123,
+      "step": 8290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.510588572008253e-06,
+      "loss": 0.9095,
+      "step": 8291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.508422709613832e-06,
+      "loss": 0.9509,
+      "step": 8292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.506256971827175e-06,
+      "loss": 0.9093,
+      "step": 8293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.504091358756599e-06,
+      "loss": 0.9441,
+      "step": 8294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.501925870510406e-06,
+      "loss": 0.9371,
+      "step": 8295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.499760507196893e-06,
+      "loss": 0.9332,
+      "step": 8296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.497595268924362e-06,
+      "loss": 0.8694,
+      "step": 8297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.495430155801091e-06,
+      "loss": 0.9417,
+      "step": 8298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.493265167935366e-06,
+      "loss": 0.8994,
+      "step": 8299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4911003054354595e-06,
+      "loss": 0.9877,
+      "step": 8300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.488935568409641e-06,
+      "loss": 0.8479,
+      "step": 8301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486770956966172e-06,
+      "loss": 0.8539,
+      "step": 8302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484606471213305e-06,
+      "loss": 0.8825,
+      "step": 8303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.482442111259296e-06,
+      "loss": 0.9369,
+      "step": 8304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480277877212381e-06,
+      "loss": 0.8664,
+      "step": 8305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478113769180803e-06,
+      "loss": 0.8388,
+      "step": 8306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475949787272787e-06,
+      "loss": 0.9242,
+      "step": 8307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473785931596562e-06,
+      "loss": 0.8936,
+      "step": 8308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4716222022603445e-06,
+      "loss": 0.8836,
+      "step": 8309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4694585993723435e-06,
+      "loss": 0.9185,
+      "step": 8310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.467295123040763e-06,
+      "loss": 0.9072,
+      "step": 8311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465131773373809e-06,
+      "loss": 1.0392,
+      "step": 8312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462968550479669e-06,
+      "loss": 0.9355,
+      "step": 8313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460805454466531e-06,
+      "loss": 0.6517,
+      "step": 8314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4586424854425755e-06,
+      "loss": 0.812,
+      "step": 8315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.456479643515972e-06,
+      "loss": 0.9634,
+      "step": 8316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454316928794893e-06,
+      "loss": 0.9323,
+      "step": 8317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4521543413874946e-06,
+      "loss": 0.9035,
+      "step": 8318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449991881401934e-06,
+      "loss": 0.8337,
+      "step": 8319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4478295489463595e-06,
+      "loss": 0.9227,
+      "step": 8320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445667344128913e-06,
+      "loss": 0.9937,
+      "step": 8321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443505267057727e-06,
+      "loss": 0.8688,
+      "step": 8322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441343317840932e-06,
+      "loss": 0.8526,
+      "step": 8323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.439181496586651e-06,
+      "loss": 0.8829,
+      "step": 8324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4370198034030005e-06,
+      "loss": 0.86,
+      "step": 8325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.43485823839809e-06,
+      "loss": 0.9724,
+      "step": 8326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.432696801680022e-06,
+      "loss": 0.841,
+      "step": 8327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.430535493356894e-06,
+      "loss": 0.9394,
+      "step": 8328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.428374313536797e-06,
+      "loss": 0.8436,
+      "step": 8329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.426213262327813e-06,
+      "loss": 0.598,
+      "step": 8330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.424052339838021e-06,
+      "loss": 0.9464,
+      "step": 8331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.421891546175492e-06,
+      "loss": 0.9578,
+      "step": 8332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.419730881448294e-06,
+      "loss": 0.9072,
+      "step": 8333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.417570345764481e-06,
+      "loss": 1.0634,
+      "step": 8334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.415409939232103e-06,
+      "loss": 0.9013,
+      "step": 8335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.413249661959213e-06,
+      "loss": 0.9455,
+      "step": 8336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.411089514053842e-06,
+      "loss": 0.7789,
+      "step": 8337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.408929495624029e-06,
+      "loss": 0.8501,
+      "step": 8338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.406769606777795e-06,
+      "loss": 0.9896,
+      "step": 8339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.404609847623161e-06,
+      "loss": 0.8631,
+      "step": 8340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.402450218268142e-06,
+      "loss": 0.8963,
+      "step": 8341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.400290718820738e-06,
+      "loss": 0.9972,
+      "step": 8342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.398131349388957e-06,
+      "loss": 0.6638,
+      "step": 8343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3959721100807915e-06,
+      "loss": 0.8741,
+      "step": 8344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3938130010042245e-06,
+      "loss": 0.9475,
+      "step": 8345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.391654022267239e-06,
+      "loss": 0.9867,
+      "step": 8346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.38949517397781e-06,
+      "loss": 0.7821,
+      "step": 8347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.387336456243901e-06,
+      "loss": 0.88,
+      "step": 8348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.385177869173477e-06,
+      "loss": 0.9201,
+      "step": 8349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3830194128744895e-06,
+      "loss": 0.97,
+      "step": 8350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.380861087454889e-06,
+      "loss": 1.0122,
+      "step": 8351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.378702893022615e-06,
+      "loss": 1.054,
+      "step": 8352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3765448296856015e-06,
+      "loss": 0.8247,
+      "step": 8353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3743868975517796e-06,
+      "loss": 0.9155,
+      "step": 8354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.372229096729067e-06,
+      "loss": 0.9367,
+      "step": 8355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.370071427325382e-06,
+      "loss": 0.934,
+      "step": 8356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.367913889448631e-06,
+      "loss": 0.9934,
+      "step": 8357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.365756483206719e-06,
+      "loss": 0.7873,
+      "step": 8358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.363599208707539e-06,
+      "loss": 0.8205,
+      "step": 8359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.361442066058979e-06,
+      "loss": 0.958,
+      "step": 8360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.359285055368923e-06,
+      "loss": 0.8403,
+      "step": 8361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357128176745245e-06,
+      "loss": 0.9009,
+      "step": 8362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.354971430295815e-06,
+      "loss": 0.9786,
+      "step": 8363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.352814816128497e-06,
+      "loss": 1.0251,
+      "step": 8364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.350658334351141e-06,
+      "loss": 0.728,
+      "step": 8365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.348501985071603e-06,
+      "loss": 0.9194,
+      "step": 8366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.34634576839772e-06,
+      "loss": 1.0097,
+      "step": 8367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.344189684437333e-06,
+      "loss": 0.8888,
+      "step": 8368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.342033733298265e-06,
+      "loss": 0.9177,
+      "step": 8369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.339877915088345e-06,
+      "loss": 0.948,
+      "step": 8370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.337722229915385e-06,
+      "loss": 1.0362,
+      "step": 8371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.335566677887194e-06,
+      "loss": 0.7934,
+      "step": 8372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3334112591115765e-06,
+      "loss": 0.8797,
+      "step": 8373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.331255973696325e-06,
+      "loss": 0.8638,
+      "step": 8374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.329100821749233e-06,
+      "loss": 0.9045,
+      "step": 8375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.326945803378084e-06,
+      "loss": 1.0196,
+      "step": 8376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.32479091869065e-06,
+      "loss": 0.9068,
+      "step": 8377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.322636167794701e-06,
+      "loss": 0.9636,
+      "step": 8378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.320481550798002e-06,
+      "loss": 0.9565,
+      "step": 8379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3183270678083065e-06,
+      "loss": 0.9269,
+      "step": 8380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.316172718933364e-06,
+      "loss": 0.8774,
+      "step": 8381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.314018504280917e-06,
+      "loss": 0.9024,
+      "step": 8382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.311864423958702e-06,
+      "loss": 0.7741,
+      "step": 8383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.309710478074448e-06,
+      "loss": 0.9717,
+      "step": 8384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3075566667358745e-06,
+      "loss": 0.8672,
+      "step": 8385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3054029900507004e-06,
+      "loss": 1.0013,
+      "step": 8386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.303249448126633e-06,
+      "loss": 0.9037,
+      "step": 8387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.301096041071376e-06,
+      "loss": 0.7914,
+      "step": 8388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2989427689926225e-06,
+      "loss": 0.9189,
+      "step": 8389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.296789631998063e-06,
+      "loss": 0.8128,
+      "step": 8390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.294636630195377e-06,
+      "loss": 0.8991,
+      "step": 8391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.29248376369224e-06,
+      "loss": 0.8649,
+      "step": 8392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.290331032596324e-06,
+      "loss": 0.912,
+      "step": 8393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.288178437015285e-06,
+      "loss": 0.8242,
+      "step": 8394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.286025977056783e-06,
+      "loss": 1.027,
+      "step": 8395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.283873652828464e-06,
+      "loss": 0.7868,
+      "step": 8396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.281721464437965e-06,
+      "loss": 0.8234,
+      "step": 8397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.279569411992926e-06,
+      "loss": 0.8786,
+      "step": 8398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.277417495600971e-06,
+      "loss": 0.9403,
+      "step": 8399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2752657153697255e-06,
+      "loss": 0.9677,
+      "step": 8400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2731140714067975e-06,
+      "loss": 0.8689,
+      "step": 8401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.270962563819799e-06,
+      "loss": 0.8865,
+      "step": 8402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2688111927163295e-06,
+      "loss": 0.8741,
+      "step": 8403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.266659958203979e-06,
+      "loss": 0.8516,
+      "step": 8404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.264508860390332e-06,
+      "loss": 1.002,
+      "step": 8405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.262357899382979e-06,
+      "loss": 0.8253,
+      "step": 8406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.260207075289485e-06,
+      "loss": 0.8382,
+      "step": 8407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2580563882174204e-06,
+      "loss": 0.8781,
+      "step": 8408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.255905838274342e-06,
+      "loss": 0.8115,
+      "step": 8409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.253755425567799e-06,
+      "loss": 0.8986,
+      "step": 8410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.251605150205343e-06,
+      "loss": 0.8911,
+      "step": 8411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.249455012294509e-06,
+      "loss": 0.9448,
+      "step": 8412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.247305011942831e-06,
+      "loss": 0.8956,
+      "step": 8413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.245155149257831e-06,
+      "loss": 0.7775,
+      "step": 8414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.243005424347031e-06,
+      "loss": 0.912,
+      "step": 8415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.240855837317939e-06,
+      "loss": 0.6288,
+      "step": 8416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.23870638827806e-06,
+      "loss": 0.787,
+      "step": 8417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2365570773348935e-06,
+      "loss": 0.636,
+      "step": 8418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.234407904595926e-06,
+      "loss": 0.9362,
+      "step": 8419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.232258870168645e-06,
+      "loss": 0.8741,
+      "step": 8420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.230109974160526e-06,
+      "loss": 0.8207,
+      "step": 8421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.227961216679037e-06,
+      "loss": 0.9745,
+      "step": 8422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2258125978316426e-06,
+      "loss": 0.8141,
+      "step": 8423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.223664117725797e-06,
+      "loss": 0.8757,
+      "step": 8424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.221515776468953e-06,
+      "loss": 0.8829,
+      "step": 8425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.219367574168545e-06,
+      "loss": 1.024,
+      "step": 8426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2172195109320166e-06,
+      "loss": 0.9008,
+      "step": 8427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.215071586866793e-06,
+      "loss": 1.0642,
+      "step": 8428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.212923802080292e-06,
+      "loss": 0.9136,
+      "step": 8429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.21077615667993e-06,
+      "loss": 0.8129,
+      "step": 8430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.208628650773115e-06,
+      "loss": 0.8745,
+      "step": 8431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.206481284467246e-06,
+      "loss": 0.9289,
+      "step": 8432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.20433405786972e-06,
+      "loss": 0.9498,
+      "step": 8433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.202186971087916e-06,
+      "loss": 0.973,
+      "step": 8434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.200040024229219e-06,
+      "loss": 0.9012,
+      "step": 8435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.197893217400995e-06,
+      "loss": 0.9487,
+      "step": 8436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.195746550710618e-06,
+      "loss": 0.9642,
+      "step": 8437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1936000242654425e-06,
+      "loss": 0.9418,
+      "step": 8438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.191453638172817e-06,
+      "loss": 0.9536,
+      "step": 8439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.18930739254009e-06,
+      "loss": 0.8477,
+      "step": 8440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.187161287474598e-06,
+      "loss": 0.9315,
+      "step": 8441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.185015323083666e-06,
+      "loss": 0.9969,
+      "step": 8442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1828694994746255e-06,
+      "loss": 0.9538,
+      "step": 8443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180723816754782e-06,
+      "loss": 0.9326,
+      "step": 8444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178578275031456e-06,
+      "loss": 0.6703,
+      "step": 8445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176432874411942e-06,
+      "loss": 0.9461,
+      "step": 8446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174287615003534e-06,
+      "loss": 0.8741,
+      "step": 8447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172142496913527e-06,
+      "loss": 0.9046,
+      "step": 8448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169997520249195e-06,
+      "loss": 0.8963,
+      "step": 8449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167852685117815e-06,
+      "loss": 0.8138,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165707991626652e-06,
+      "loss": 0.8974,
+      "step": 8451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163563439882968e-06,
+      "loss": 0.6571,
+      "step": 8452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161419029994013e-06,
+      "loss": 0.6503,
+      "step": 8453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1592747620670324e-06,
+      "loss": 1.0252,
+      "step": 8454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1571306362092665e-06,
+      "loss": 0.8822,
+      "step": 8455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154986652527943e-06,
+      "loss": 0.9039,
+      "step": 8456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152842811130291e-06,
+      "loss": 0.6225,
+      "step": 8457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150699112123525e-06,
+      "loss": 0.896,
+      "step": 8458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148555555614851e-06,
+      "loss": 0.6158,
+      "step": 8459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146412141711479e-06,
+      "loss": 1.0564,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144268870520596e-06,
+      "loss": 0.8065,
+      "step": 8461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1421257421494e-06,
+      "loss": 0.947,
+      "step": 8462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139982756705064e-06,
+      "loss": 0.9261,
+      "step": 8463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1378399142947686e-06,
+      "loss": 0.9146,
+      "step": 8464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135697215025677e-06,
+      "loss": 0.913,
+      "step": 8465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.133554659004946e-06,
+      "loss": 0.8817,
+      "step": 8466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131412246339736e-06,
+      "loss": 0.9394,
+      "step": 8467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129269977137184e-06,
+      "loss": 0.8519,
+      "step": 8468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1271278515044355e-06,
+      "loss": 0.9331,
+      "step": 8469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.12498586954862e-06,
+      "loss": 0.8823,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.122844031376861e-06,
+      "loss": 0.878,
+      "step": 8471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.120702337096273e-06,
+      "loss": 0.7884,
+      "step": 8472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.118560786813967e-06,
+      "loss": 1.0014,
+      "step": 8473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.116419380637046e-06,
+      "loss": 0.953,
+      "step": 8474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.114278118672604e-06,
+      "loss": 0.8941,
+      "step": 8475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.11213700102773e-06,
+      "loss": 0.8708,
+      "step": 8476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.109996027809505e-06,
+      "loss": 0.9524,
+      "step": 8477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1078551991250025e-06,
+      "loss": 1.0176,
+      "step": 8478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.105714515081285e-06,
+      "loss": 0.6134,
+      "step": 8479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.103573975785418e-06,
+      "loss": 0.8692,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.101433581344448e-06,
+      "loss": 0.8879,
+      "step": 8481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0992933318654225e-06,
+      "loss": 0.8175,
+      "step": 8482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.097153227455379e-06,
+      "loss": 1.0022,
+      "step": 8483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0950132682213425e-06,
+      "loss": 0.9111,
+      "step": 8484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.092873454270344e-06,
+      "loss": 1.0124,
+      "step": 8485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.090733785709391e-06,
+      "loss": 0.9021,
+      "step": 8486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.088594262645497e-06,
+      "loss": 0.661,
+      "step": 8487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.08645488518566e-06,
+      "loss": 0.9365,
+      "step": 8488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.084315653436878e-06,
+      "loss": 0.8728,
+      "step": 8489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.082176567506132e-06,
+      "loss": 1.0058,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.080037627500403e-06,
+      "loss": 0.8315,
+      "step": 8491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0778988335266655e-06,
+      "loss": 0.8378,
+      "step": 8492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.075760185691879e-06,
+      "loss": 0.9268,
+      "step": 8493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.073621684103007e-06,
+      "loss": 0.6525,
+      "step": 8494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.071483328866993e-06,
+      "loss": 0.9737,
+      "step": 8495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.069345120090784e-06,
+      "loss": 0.8878,
+      "step": 8496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.067207057881314e-06,
+      "loss": 0.9035,
+      "step": 8497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0650691423455085e-06,
+      "loss": 0.9322,
+      "step": 8498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.06293137359029e-06,
+      "loss": 0.7953,
+      "step": 8499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0607937517225734e-06,
+      "loss": 0.8867,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.058656276849262e-06,
+      "loss": 0.9527,
+      "step": 8501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.056518949077258e-06,
+      "loss": 0.9497,
+      "step": 8502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.05438176851345e-06,
+      "loss": 1.0522,
+      "step": 8503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.052244735264719e-06,
+      "loss": 0.808,
+      "step": 8504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.050107849437948e-06,
+      "loss": 0.9313,
+      "step": 8505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.047971111139999e-06,
+      "loss": 0.8496,
+      "step": 8506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.045834520477741e-06,
+      "loss": 0.8197,
+      "step": 8507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.043698077558021e-06,
+      "loss": 1.0888,
+      "step": 8508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.041561782487691e-06,
+      "loss": 0.8148,
+      "step": 8509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.039425635373589e-06,
+      "loss": 0.8208,
+      "step": 8510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.037289636322546e-06,
+      "loss": 0.9455,
+      "step": 8511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.035153785441388e-06,
+      "loss": 0.9065,
+      "step": 8512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.033018082836933e-06,
+      "loss": 0.8211,
+      "step": 8513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0308825286159895e-06,
+      "loss": 0.9172,
+      "step": 8514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.028747122885362e-06,
+      "loss": 0.8459,
+      "step": 8515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.026611865751841e-06,
+      "loss": 0.9044,
+      "step": 8516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0244767573222195e-06,
+      "loss": 0.8508,
+      "step": 8517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.022341797703274e-06,
+      "loss": 0.8933,
+      "step": 8518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.020206987001779e-06,
+      "loss": 0.9867,
+      "step": 8519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.018072325324498e-06,
+      "loss": 0.8553,
+      "step": 8520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.015937812778193e-06,
+      "loss": 0.899,
+      "step": 8521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.013803449469612e-06,
+      "loss": 0.851,
+      "step": 8522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.011669235505494e-06,
+      "loss": 0.9497,
+      "step": 8523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.009535170992581e-06,
+      "loss": 0.9112,
+      "step": 8524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.007401256037596e-06,
+      "loss": 0.9408,
+      "step": 8525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.005267490747263e-06,
+      "loss": 0.9864,
+      "step": 8526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.003133875228293e-06,
+      "loss": 0.871,
+      "step": 8527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.001000409587391e-06,
+      "loss": 0.8693,
+      "step": 8528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.998867093931258e-06,
+      "loss": 0.8194,
+      "step": 8529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9967339283665795e-06,
+      "loss": 0.8101,
+      "step": 8530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.99460091300004e-06,
+      "loss": 0.9282,
+      "step": 8531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.992468047938321e-06,
+      "loss": 0.8851,
+      "step": 8532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.990335333288084e-06,
+      "loss": 0.91,
+      "step": 8533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.988202769155994e-06,
+      "loss": 0.9072,
+      "step": 8534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.986070355648701e-06,
+      "loss": 0.9583,
+      "step": 8535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.983938092872849e-06,
+      "loss": 0.9745,
+      "step": 8536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.98180598093508e-06,
+      "loss": 0.9649,
+      "step": 8537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.97967401994202e-06,
+      "loss": 0.9938,
+      "step": 8538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.977542210000297e-06,
+      "loss": 0.8347,
+      "step": 8539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.975410551216522e-06,
+      "loss": 0.9148,
+      "step": 8540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9732790436973005e-06,
+      "loss": 0.8737,
+      "step": 8541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.97114768754924e-06,
+      "loss": 0.6126,
+      "step": 8542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.969016482878925e-06,
+      "loss": 0.8418,
+      "step": 8543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.966885429792949e-06,
+      "loss": 0.9003,
+      "step": 8544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.964754528397881e-06,
+      "loss": 0.9219,
+      "step": 8545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.962623778800297e-06,
+      "loss": 0.9435,
+      "step": 8546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.960493181106755e-06,
+      "loss": 1.0554,
+      "step": 8547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9583627354238114e-06,
+      "loss": 0.9449,
+      "step": 8548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.956232441858015e-06,
+      "loss": 0.8593,
+      "step": 8549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.954102300515901e-06,
+      "loss": 0.911,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.951972311504006e-06,
+      "loss": 0.9201,
+      "step": 8551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.949842474928852e-06,
+      "loss": 0.608,
+      "step": 8552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.947712790896952e-06,
+      "loss": 0.9852,
+      "step": 8553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.945583259514821e-06,
+      "loss": 0.9027,
+      "step": 8554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.943453880888954e-06,
+      "loss": 1.0435,
+      "step": 8555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.941324655125851e-06,
+      "loss": 0.8747,
+      "step": 8556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.939195582331993e-06,
+      "loss": 0.9973,
+      "step": 8557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937066662613863e-06,
+      "loss": 0.9236,
+      "step": 8558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.934937896077927e-06,
+      "loss": 0.843,
+      "step": 8559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.932809282830649e-06,
+      "loss": 0.9639,
+      "step": 8560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.930680822978489e-06,
+      "loss": 0.8245,
+      "step": 8561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9285525166278845e-06,
+      "loss": 0.9041,
+      "step": 8562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.926424363885285e-06,
+      "loss": 0.8844,
+      "step": 8563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.924296364857123e-06,
+      "loss": 0.9784,
+      "step": 8564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.922168519649819e-06,
+      "loss": 0.9681,
+      "step": 8565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.920040828369789e-06,
+      "loss": 0.9143,
+      "step": 8566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.917913291123446e-06,
+      "loss": 0.8504,
+      "step": 8567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.915785908017189e-06,
+      "loss": 0.9471,
+      "step": 8568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9136586791574136e-06,
+      "loss": 1.0962,
+      "step": 8569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.911531604650503e-06,
+      "loss": 0.894,
+      "step": 8570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.909404684602838e-06,
+      "loss": 0.9136,
+      "step": 8571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.907277919120792e-06,
+      "loss": 0.9951,
+      "step": 8572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90515130831072e-06,
+      "loss": 0.8928,
+      "step": 8573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.903024852278985e-06,
+      "loss": 0.9323,
+      "step": 8574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.900898551131929e-06,
+      "loss": 0.7801,
+      "step": 8575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.898772404975897e-06,
+      "loss": 0.9688,
+      "step": 8576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.896646413917216e-06,
+      "loss": 0.9176,
+      "step": 8577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.894520578062213e-06,
+      "loss": 0.8864,
+      "step": 8578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.892394897517205e-06,
+      "loss": 0.9362,
+      "step": 8579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.890269372388497e-06,
+      "loss": 1.0081,
+      "step": 8580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.888144002782395e-06,
+      "loss": 0.6964,
+      "step": 8581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.886018788805187e-06,
+      "loss": 0.8701,
+      "step": 8582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.883893730563163e-06,
+      "loss": 0.8925,
+      "step": 8583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.881768828162598e-06,
+      "loss": 0.8626,
+      "step": 8584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.87964408170976e-06,
+      "loss": 0.6345,
+      "step": 8585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877519491310916e-06,
+      "loss": 0.8918,
+      "step": 8586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875395057072314e-06,
+      "loss": 0.8981,
+      "step": 8587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873270779100207e-06,
+      "loss": 0.8624,
+      "step": 8588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871146657500827e-06,
+      "loss": 0.7895,
+      "step": 8589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869022692380411e-06,
+      "loss": 0.8333,
+      "step": 8590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8668988838451785e-06,
+      "loss": 0.779,
+      "step": 8591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864775232001342e-06,
+      "loss": 0.6125,
+      "step": 8592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862651736955111e-06,
+      "loss": 0.931,
+      "step": 8593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860528398812689e-06,
+      "loss": 0.9506,
+      "step": 8594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8584052176802615e-06,
+      "loss": 0.903,
+      "step": 8595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856282193664017e-06,
+      "loss": 0.6049,
+      "step": 8596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8541593268701286e-06,
+      "loss": 0.9545,
+      "step": 8597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852036617404764e-06,
+      "loss": 0.8949,
+      "step": 8598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849914065374084e-06,
+      "loss": 0.972,
+      "step": 8599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847791670884241e-06,
+      "loss": 0.9198,
+      "step": 8600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8456694340413795e-06,
+      "loss": 0.8969,
+      "step": 8601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843547354951634e-06,
+      "loss": 0.9059,
+      "step": 8602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841425433721138e-06,
+      "loss": 0.936,
+      "step": 8603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839303670456008e-06,
+      "loss": 0.8682,
+      "step": 8604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837182065262356e-06,
+      "loss": 0.8838,
+      "step": 8605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83506061824629e-06,
+      "loss": 0.9966,
+      "step": 8606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832939329513903e-06,
+      "loss": 0.9478,
+      "step": 8607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830818199171288e-06,
+      "loss": 0.913,
+      "step": 8608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828697227324525e-06,
+      "loss": 0.9441,
+      "step": 8609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8265764140796865e-06,
+      "loss": 1.0278,
+      "step": 8610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824455759542839e-06,
+      "loss": 0.8971,
+      "step": 8611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822335263820037e-06,
+      "loss": 0.8832,
+      "step": 8612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820214927017334e-06,
+      "loss": 0.9683,
+      "step": 8613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818094749240767e-06,
+      "loss": 0.9654,
+      "step": 8614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.815974730596373e-06,
+      "loss": 0.8543,
+      "step": 8615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.813854871190178e-06,
+      "loss": 0.7576,
+      "step": 8616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8117351711281935e-06,
+      "loss": 0.8911,
+      "step": 8617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.809615630516437e-06,
+      "loss": 0.8724,
+      "step": 8618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8074962494609035e-06,
+      "loss": 0.8249,
+      "step": 8619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.805377028067593e-06,
+      "loss": 0.943,
+      "step": 8620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8032579664424866e-06,
+      "loss": 0.9267,
+      "step": 8621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.801139064691562e-06,
+      "loss": 0.8555,
+      "step": 8622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.79902032292079e-06,
+      "loss": 0.588,
+      "step": 8623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.796901741236133e-06,
+      "loss": 0.9392,
+      "step": 8624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.794783319743541e-06,
+      "loss": 0.9877,
+      "step": 8625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.792665058548967e-06,
+      "loss": 0.9581,
+      "step": 8626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.790546957758341e-06,
+      "loss": 0.8326,
+      "step": 8627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.788429017477598e-06,
+      "loss": 0.977,
+      "step": 8628
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.786311237812659e-06,
+      "loss": 0.9254,
+      "step": 8629
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7841936188694324e-06,
+      "loss": 0.9478,
+      "step": 8630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.78207616075383e-06,
+      "loss": 0.8884,
+      "step": 8631
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.779958863571745e-06,
+      "loss": 0.8342,
+      "step": 8632
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.777841727429069e-06,
+      "loss": 0.9816,
+      "step": 8633
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.775724752431683e-06,
+      "loss": 0.6353,
+      "step": 8634
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.773607938685459e-06,
+      "loss": 0.6257,
+      "step": 8635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.771491286296264e-06,
+      "loss": 0.8931,
+      "step": 8636
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7693747953699544e-06,
+      "loss": 0.8393,
+      "step": 8637
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.76725846601238e-06,
+      "loss": 0.9056,
+      "step": 8638
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.765142298329379e-06,
+      "loss": 0.8768,
+      "step": 8639
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763026292426789e-06,
+      "loss": 0.9296,
+      "step": 8640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.760910448410431e-06,
+      "loss": 0.8995,
+      "step": 8641
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.758794766386123e-06,
+      "loss": 0.9494,
+      "step": 8642
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.756679246459675e-06,
+      "loss": 0.8206,
+      "step": 8643
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.754563888736885e-06,
+      "loss": 0.93,
+      "step": 8644
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.752448693323547e-06,
+      "loss": 0.8812,
+      "step": 8645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.750333660325445e-06,
+      "loss": 0.8715,
+      "step": 8646
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7482187898483555e-06,
+      "loss": 0.9054,
+      "step": 8647
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.746104081998046e-06,
+      "loss": 0.838,
+      "step": 8648
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.743989536880275e-06,
+      "loss": 0.992,
+      "step": 8649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.741875154600799e-06,
+      "loss": 0.8465,
+      "step": 8650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.739760935265356e-06,
+      "loss": 0.641,
+      "step": 8651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7376468789796864e-06,
+      "loss": 0.933,
+      "step": 8652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7355329858495154e-06,
+      "loss": 0.9698,
+      "step": 8653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.733419255980559e-06,
+      "loss": 0.8987,
+      "step": 8654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.731305689478533e-06,
+      "loss": 0.8408,
+      "step": 8655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.729192286449136e-06,
+      "loss": 0.945,
+      "step": 8656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.727079046998065e-06,
+      "loss": 0.9207,
+      "step": 8657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.72496597123101e-06,
+      "loss": 0.9648,
+      "step": 8658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.722853059253645e-06,
+      "loss": 1.0939,
+      "step": 8659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.72074031117164e-06,
+      "loss": 0.8734,
+      "step": 8660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7186277270906594e-06,
+      "loss": 0.6064,
+      "step": 8661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.716515307116353e-06,
+      "loss": 0.8364,
+      "step": 8662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.714403051354372e-06,
+      "loss": 0.9195,
+      "step": 8663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.712290959910348e-06,
+      "loss": 0.9729,
+      "step": 8664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.710179032889914e-06,
+      "loss": 0.8883,
+      "step": 8665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.708067270398689e-06,
+      "loss": 1.011,
+      "step": 8666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.705955672542285e-06,
+      "loss": 0.8695,
+      "step": 8667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7038442394263085e-06,
+      "loss": 0.9077,
+      "step": 8668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.701732971156351e-06,
+      "loss": 0.9061,
+      "step": 8669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.699621867838009e-06,
+      "loss": 0.8936,
+      "step": 8670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.697510929576854e-06,
+      "loss": 0.8493,
+      "step": 8671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.695400156478462e-06,
+      "loss": 0.9178,
+      "step": 8672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.693289548648394e-06,
+      "loss": 0.865,
+      "step": 8673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.691179106192205e-06,
+      "loss": 0.9124,
+      "step": 8674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.689068829215442e-06,
+      "loss": 0.9402,
+      "step": 8675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.686958717823643e-06,
+      "loss": 0.9145,
+      "step": 8676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.68484877212234e-06,
+      "loss": 0.916,
+      "step": 8677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.682738992217053e-06,
+      "loss": 0.8464,
+      "step": 8678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.680629378213294e-06,
+      "loss": 0.9207,
+      "step": 8679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.678519930216572e-06,
+      "loss": 0.6252,
+      "step": 8680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.676410648332379e-06,
+      "loss": 0.9435,
+      "step": 8681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.674301532666207e-06,
+      "loss": 0.6082,
+      "step": 8682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.672192583323535e-06,
+      "loss": 0.8721,
+      "step": 8683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.670083800409839e-06,
+      "loss": 0.8515,
+      "step": 8684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.667975184030577e-06,
+      "loss": 1.1134,
+      "step": 8685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6658667342912055e-06,
+      "loss": 0.9507,
+      "step": 8686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.663758451297169e-06,
+      "loss": 0.9639,
+      "step": 8687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.661650335153914e-06,
+      "loss": 0.8579,
+      "step": 8688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.659542385966864e-06,
+      "loss": 0.6433,
+      "step": 8689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.657434603841446e-06,
+      "loss": 0.859,
+      "step": 8690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.655326988883069e-06,
+      "loss": 0.8846,
+      "step": 8691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.653219541197139e-06,
+      "loss": 0.9219,
+      "step": 8692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.651112260889054e-06,
+      "loss": 0.8827,
+      "step": 8693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.649005148064202e-06,
+      "loss": 0.8663,
+      "step": 8694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.646898202827965e-06,
+      "loss": 0.9991,
+      "step": 8695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.64479142528571e-06,
+      "loss": 0.6123,
+      "step": 8696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.642684815542806e-06,
+      "loss": 0.9116,
+      "step": 8697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.640578373704605e-06,
+      "loss": 0.8609,
+      "step": 8698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.638472099876449e-06,
+      "loss": 0.9459,
+      "step": 8699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.636365994163687e-06,
+      "loss": 0.8394,
+      "step": 8700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.634260056671638e-06,
+      "loss": 0.9075,
+      "step": 8701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.63215428750563e-06,
+      "loss": 1.0405,
+      "step": 8702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6300486867709735e-06,
+      "loss": 0.8643,
+      "step": 8703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.627943254572971e-06,
+      "loss": 0.9199,
+      "step": 8704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.625837991016923e-06,
+      "loss": 0.8371,
+      "step": 8705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.623732896208114e-06,
+      "loss": 0.9779,
+      "step": 8706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.621627970251825e-06,
+      "loss": 0.8097,
+      "step": 8707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.619523213253322e-06,
+      "loss": 0.9452,
+      "step": 8708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.617418625317875e-06,
+      "loss": 0.905,
+      "step": 8709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.615314206550733e-06,
+      "loss": 0.9375,
+      "step": 8710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.613209957057141e-06,
+      "loss": 0.9652,
+      "step": 8711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6111058769423375e-06,
+      "loss": 0.8902,
+      "step": 8712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.609001966311549e-06,
+      "loss": 1.0193,
+      "step": 8713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6068982252699986e-06,
+      "loss": 0.9614,
+      "step": 8714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.604794653922897e-06,
+      "loss": 0.9406,
+      "step": 8715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.602691252375442e-06,
+      "loss": 0.9672,
+      "step": 8716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.600588020732835e-06,
+      "loss": 0.9448,
+      "step": 8717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.598484959100257e-06,
+      "loss": 0.893,
+      "step": 8718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.596382067582887e-06,
+      "loss": 0.951,
+      "step": 8719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5942793462858964e-06,
+      "loss": 0.9275,
+      "step": 8720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.592176795314445e-06,
+      "loss": 0.956,
+      "step": 8721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.590074414773683e-06,
+      "loss": 0.8134,
+      "step": 8722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.587972204768756e-06,
+      "loss": 0.8746,
+      "step": 8723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.585870165404793e-06,
+      "loss": 0.9695,
+      "step": 8724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.583768296786927e-06,
+      "loss": 0.935,
+      "step": 8725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.581666599020273e-06,
+      "loss": 0.9268,
+      "step": 8726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.579565072209944e-06,
+      "loss": 0.8458,
+      "step": 8727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577463716461034e-06,
+      "loss": 1.0654,
+      "step": 8728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5753625318786395e-06,
+      "loss": 0.8729,
+      "step": 8729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573261518567844e-06,
+      "loss": 0.9339,
+      "step": 8730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571160676633719e-06,
+      "loss": 0.891,
+      "step": 8731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569060006181336e-06,
+      "loss": 0.8874,
+      "step": 8732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56695950731575e-06,
+      "loss": 0.9609,
+      "step": 8733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564859180142011e-06,
+      "loss": 1.0378,
+      "step": 8734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562759024765162e-06,
+      "loss": 0.9608,
+      "step": 8735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56065904129023e-06,
+      "loss": 0.9274,
+      "step": 8736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5585592298222416e-06,
+      "loss": 0.9564,
+      "step": 8737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556459590466211e-06,
+      "loss": 0.7556,
+      "step": 8738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554360123327146e-06,
+      "loss": 0.9583,
+      "step": 8739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.552260828510043e-06,
+      "loss": 0.8653,
+      "step": 8740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.550161706119891e-06,
+      "loss": 0.883,
+      "step": 8741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5480627562616705e-06,
+      "loss": 0.837,
+      "step": 8742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545963979040353e-06,
+      "loss": 0.9062,
+      "step": 8743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5438653745609045e-06,
+      "loss": 0.8506,
+      "step": 8744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541766942928276e-06,
+      "loss": 0.8457,
+      "step": 8745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5396686842474155e-06,
+      "loss": 0.7966,
+      "step": 8746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5375705986232606e-06,
+      "loss": 0.8798,
+      "step": 8747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535472686160738e-06,
+      "loss": 1.0158,
+      "step": 8748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5333749469647675e-06,
+      "loss": 0.9329,
+      "step": 8749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53127738114026e-06,
+      "loss": 0.8874,
+      "step": 8750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5291799887921206e-06,
+      "loss": 0.9761,
+      "step": 8751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527082770025244e-06,
+      "loss": 0.6271,
+      "step": 8752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5249857249445126e-06,
+      "loss": 0.8748,
+      "step": 8753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5228888536548056e-06,
+      "loss": 0.8838,
+      "step": 8754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52079215626099e-06,
+      "loss": 0.8528,
+      "step": 8755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518695632867922e-06,
+      "loss": 1.0866,
+      "step": 8756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.516599283580456e-06,
+      "loss": 0.992,
+      "step": 8757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514503108503431e-06,
+      "loss": 0.9998,
+      "step": 8758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5124071077416825e-06,
+      "loss": 0.9329,
+      "step": 8759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.510311281400035e-06,
+      "loss": 0.9014,
+      "step": 8760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5082156295832996e-06,
+      "loss": 0.8808,
+      "step": 8761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.506120152396289e-06,
+      "loss": 0.9126,
+      "step": 8762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.504024849943796e-06,
+      "loss": 0.9429,
+      "step": 8763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.501929722330615e-06,
+      "loss": 0.8795,
+      "step": 8764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.499834769661524e-06,
+      "loss": 0.9761,
+      "step": 8765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.497739992041296e-06,
+      "loss": 0.9032,
+      "step": 8766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.495645389574694e-06,
+      "loss": 1.0186,
+      "step": 8767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.49355096236647e-06,
+      "loss": 0.9617,
+      "step": 8768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.491456710521374e-06,
+      "loss": 0.877,
+      "step": 8769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.489362634144138e-06,
+      "loss": 0.9803,
+      "step": 8770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.487268733339495e-06,
+      "loss": 0.8288,
+      "step": 8771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.485175008212162e-06,
+      "loss": 0.8345,
+      "step": 8772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.483081458866847e-06,
+      "loss": 0.943,
+      "step": 8773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4809880854082565e-06,
+      "loss": 1.0085,
+      "step": 8774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.478894887941079e-06,
+      "loss": 0.706,
+      "step": 8775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.476801866570004e-06,
+      "loss": 0.6356,
+      "step": 8776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.474709021399701e-06,
+      "loss": 0.9853,
+      "step": 8777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.472616352534841e-06,
+      "loss": 0.886,
+      "step": 8778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.47052386008008e-06,
+      "loss": 0.8609,
+      "step": 8779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.468431544140066e-06,
+      "loss": 0.9437,
+      "step": 8780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.466339404819437e-06,
+      "loss": 1.0179,
+      "step": 8781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4642474422228305e-06,
+      "loss": 0.8343,
+      "step": 8782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.462155656454865e-06,
+      "loss": 0.9013,
+      "step": 8783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.460064047620156e-06,
+      "loss": 0.9203,
+      "step": 8784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.457972615823307e-06,
+      "loss": 0.8148,
+      "step": 8785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.455881361168912e-06,
+      "loss": 0.8288,
+      "step": 8786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.453790283761561e-06,
+      "loss": 0.5975,
+      "step": 8787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4516993837058296e-06,
+      "loss": 0.928,
+      "step": 8788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.44960866110629e-06,
+      "loss": 0.9182,
+      "step": 8789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.447518116067498e-06,
+      "loss": 0.843,
+      "step": 8790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.445427748694011e-06,
+      "loss": 0.9046,
+      "step": 8791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.443337559090368e-06,
+      "loss": 0.6065,
+      "step": 8792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.441247547361102e-06,
+      "loss": 0.8412,
+      "step": 8793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.439157713610741e-06,
+      "loss": 0.9266,
+      "step": 8794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.437068057943796e-06,
+      "loss": 0.8612,
+      "step": 8795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4349785804647806e-06,
+      "loss": 0.8477,
+      "step": 8796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.432889281278189e-06,
+      "loss": 0.8903,
+      "step": 8797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4308001604885085e-06,
+      "loss": 0.9672,
+      "step": 8798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.428711218200223e-06,
+      "loss": 0.9949,
+      "step": 8799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.426622454517801e-06,
+      "loss": 0.8899,
+      "step": 8800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.424533869545709e-06,
+      "loss": 0.7461,
+      "step": 8801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.422445463388395e-06,
+      "loss": 0.8427,
+      "step": 8802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.420357236150308e-06,
+      "loss": 0.9554,
+      "step": 8803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4182691879358835e-06,
+      "loss": 0.9877,
+      "step": 8804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.416181318849543e-06,
+      "loss": 0.7663,
+      "step": 8805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.414093628995712e-06,
+      "loss": 0.947,
+      "step": 8806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.412006118478791e-06,
+      "loss": 0.6332,
+      "step": 8807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.409918787403187e-06,
+      "loss": 0.9846,
+      "step": 8808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.407831635873287e-06,
+      "loss": 1.0014,
+      "step": 8809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.405744663993472e-06,
+      "loss": 0.8676,
+      "step": 8810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.403657871868116e-06,
+      "loss": 0.8644,
+      "step": 8811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.401571259601583e-06,
+      "loss": 0.889,
+      "step": 8812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3994848272982255e-06,
+      "loss": 0.6361,
+      "step": 8813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.397398575062395e-06,
+      "loss": 0.9109,
+      "step": 8814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.395312502998423e-06,
+      "loss": 0.9205,
+      "step": 8815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.393226611210642e-06,
+      "loss": 0.8351,
+      "step": 8816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3911408998033675e-06,
+      "loss": 0.8533,
+      "step": 8817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.389055368880908e-06,
+      "loss": 0.6016,
+      "step": 8818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.386970018547567e-06,
+      "loss": 1.0226,
+      "step": 8819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.384884848907633e-06,
+      "loss": 0.8529,
+      "step": 8820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.382799860065394e-06,
+      "loss": 0.8433,
+      "step": 8821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.38071505212512e-06,
+      "loss": 0.9995,
+      "step": 8822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3786304251910745e-06,
+      "loss": 0.8868,
+      "step": 8823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3765459793675165e-06,
+      "loss": 0.9044,
+      "step": 8824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.374461714758688e-06,
+      "loss": 0.8139,
+      "step": 8825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.372377631468831e-06,
+      "loss": 0.9566,
+      "step": 8826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3702937296021695e-06,
+      "loss": 0.9462,
+      "step": 8827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.368210009262927e-06,
+      "loss": 0.846,
+      "step": 8828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.366126470555313e-06,
+      "loss": 0.928,
+      "step": 8829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.364043113583525e-06,
+      "loss": 0.6068,
+      "step": 8830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.361959938451759e-06,
+      "loss": 0.8486,
+      "step": 8831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.359876945264194e-06,
+      "loss": 0.8794,
+      "step": 8832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.357794134125008e-06,
+      "loss": 0.8358,
+      "step": 8833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3557115051383625e-06,
+      "loss": 0.8557,
+      "step": 8834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.353629058408416e-06,
+      "loss": 0.9615,
+      "step": 8835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3515467940393136e-06,
+      "loss": 0.6061,
+      "step": 8836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3494647121351905e-06,
+      "loss": 0.8037,
+      "step": 8837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.347382812800178e-06,
+      "loss": 0.8777,
+      "step": 8838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.345301096138394e-06,
+      "loss": 0.8996,
+      "step": 8839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.343219562253949e-06,
+      "loss": 0.9057,
+      "step": 8840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3411382112509445e-06,
+      "loss": 0.9813,
+      "step": 8841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3390570432334695e-06,
+      "loss": 0.7726,
+      "step": 8842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3369760583056085e-06,
+      "loss": 0.9953,
+      "step": 8843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.334895256571432e-06,
+      "loss": 0.8194,
+      "step": 8844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.332814638135008e-06,
+      "loss": 0.7863,
+      "step": 8845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.330734203100394e-06,
+      "loss": 0.8136,
+      "step": 8846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.328653951571628e-06,
+      "loss": 0.8627,
+      "step": 8847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.326573883652755e-06,
+      "loss": 0.9459,
+      "step": 8848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3244939994477985e-06,
+      "loss": 0.9515,
+      "step": 8849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.322414299060773e-06,
+      "loss": 0.9655,
+      "step": 8850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.320334782595694e-06,
+      "loss": 0.5882,
+      "step": 8851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.318255450156557e-06,
+      "loss": 0.816,
+      "step": 8852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3161763018473575e-06,
+      "loss": 1.0686,
+      "step": 8853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.314097337772074e-06,
+      "loss": 0.9494,
+      "step": 8854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.312018558034676e-06,
+      "loss": 0.8816,
+      "step": 8855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.309939962739133e-06,
+      "loss": 1.004,
+      "step": 8856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.307861551989392e-06,
+      "loss": 0.9722,
+      "step": 8857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.305783325889405e-06,
+      "loss": 0.9314,
+      "step": 8858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3037052845431e-06,
+      "loss": 0.9747,
+      "step": 8859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.301627428054409e-06,
+      "loss": 0.8973,
+      "step": 8860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.299549756527248e-06,
+      "loss": 0.9733,
+      "step": 8861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.297472270065519e-06,
+      "loss": 0.9936,
+      "step": 8862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.295394968773129e-06,
+      "loss": 1.0323,
+      "step": 8863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.29331785275396e-06,
+      "loss": 0.9193,
+      "step": 8864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2912409221118964e-06,
+      "loss": 0.953,
+      "step": 8865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2891641769508075e-06,
+      "loss": 0.9172,
+      "step": 8866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.287087617374552e-06,
+      "loss": 0.8816,
+      "step": 8867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.285011243486987e-06,
+      "loss": 0.9245,
+      "step": 8868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.28293505539195e-06,
+      "loss": 0.9661,
+      "step": 8869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28085905319328e-06,
+      "loss": 0.5951,
+      "step": 8870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278783236994794e-06,
+      "loss": 0.9842,
+      "step": 8871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276707606900317e-06,
+      "loss": 0.8171,
+      "step": 8872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274632163013646e-06,
+      "loss": 0.9707,
+      "step": 8873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272556905438578e-06,
+      "loss": 0.8936,
+      "step": 8874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270481834278901e-06,
+      "loss": 0.9316,
+      "step": 8875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268406949638397e-06,
+      "loss": 0.9415,
+      "step": 8876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266332251620829e-06,
+      "loss": 0.8931,
+      "step": 8877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26425774032996e-06,
+      "loss": 1.0074,
+      "step": 8878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262183415869538e-06,
+      "loss": 0.9467,
+      "step": 8879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2601092783433e-06,
+      "loss": 0.746,
+      "step": 8880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258035327854983e-06,
+      "loss": 0.6207,
+      "step": 8881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255961564508302e-06,
+      "loss": 0.937,
+      "step": 8882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.253887988406974e-06,
+      "loss": 0.8743,
+      "step": 8883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2518145996547e-06,
+      "loss": 0.8422,
+      "step": 8884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249741398355175e-06,
+      "loss": 0.879,
+      "step": 8885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247668384612083e-06,
+      "loss": 0.8905,
+      "step": 8886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245595558529095e-06,
+      "loss": 0.9028,
+      "step": 8887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243522920209881e-06,
+      "loss": 0.6126,
+      "step": 8888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241450469758094e-06,
+      "loss": 0.9306,
+      "step": 8889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239378207277382e-06,
+      "loss": 0.8959,
+      "step": 8890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237306132871383e-06,
+      "loss": 0.967,
+      "step": 8891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23523424664372e-06,
+      "loss": 0.8314,
+      "step": 8892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233162548698019e-06,
+      "loss": 0.872,
+      "step": 8893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231091039137881e-06,
+      "loss": 0.9256,
+      "step": 8894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229019718066912e-06,
+      "loss": 0.9899,
+      "step": 8895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226948585588697e-06,
+      "loss": 0.9199,
+      "step": 8896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224877641806821e-06,
+      "loss": 0.9001,
+      "step": 8897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222806886824855e-06,
+      "loss": 0.9114,
+      "step": 8898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220736320746355e-06,
+      "loss": 0.8655,
+      "step": 8899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2186659436748795e-06,
+      "loss": 0.8398,
+      "step": 8900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216595755713968e-06,
+      "loss": 0.6222,
+      "step": 8901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214525756967156e-06,
+      "loss": 0.8978,
+      "step": 8902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212455947537967e-06,
+      "loss": 1.0711,
+      "step": 8903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.210386327529914e-06,
+      "loss": 0.8527,
+      "step": 8904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.208316897046505e-06,
+      "loss": 0.9752,
+      "step": 8905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.20624765619123e-06,
+      "loss": 0.9439,
+      "step": 8906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.204178605067576e-06,
+      "loss": 0.7731,
+      "step": 8907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.202109743779029e-06,
+      "loss": 0.9892,
+      "step": 8908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.200041072429046e-06,
+      "loss": 0.9006,
+      "step": 8909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.19797259112109e-06,
+      "loss": 0.9082,
+      "step": 8910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.195904299958606e-06,
+      "loss": 0.8462,
+      "step": 8911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.193836199045032e-06,
+      "loss": 0.9366,
+      "step": 8912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.191768288483801e-06,
+      "loss": 0.9455,
+      "step": 8913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.189700568378327e-06,
+      "loss": 0.967,
+      "step": 8914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.187633038832026e-06,
+      "loss": 0.9115,
+      "step": 8915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.185565699948295e-06,
+      "loss": 0.9063,
+      "step": 8916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.183498551830522e-06,
+      "loss": 0.5977,
+      "step": 8917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.181431594582096e-06,
+      "loss": 0.9236,
+      "step": 8918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.17936482830638e-06,
+      "loss": 0.931,
+      "step": 8919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.177298253106745e-06,
+      "loss": 1.0024,
+      "step": 8920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.175231869086537e-06,
+      "loss": 0.8583,
+      "step": 8921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.173165676349103e-06,
+      "loss": 0.9452,
+      "step": 8922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1710996749977756e-06,
+      "loss": 0.9021,
+      "step": 8923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1690338651358785e-06,
+      "loss": 0.9913,
+      "step": 8924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.166968246866725e-06,
+      "loss": 0.8596,
+      "step": 8925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.164902820293621e-06,
+      "loss": 0.9439,
+      "step": 8926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.162837585519866e-06,
+      "loss": 0.9204,
+      "step": 8927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.160772542648737e-06,
+      "loss": 0.8983,
+      "step": 8928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1587076917835165e-06,
+      "loss": 0.9741,
+      "step": 8929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1566430330274695e-06,
+      "loss": 0.9301,
+      "step": 8930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.154578566483852e-06,
+      "loss": 0.9879,
+      "step": 8931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.152514292255913e-06,
+      "loss": 0.8931,
+      "step": 8932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.150450210446888e-06,
+      "loss": 1.0182,
+      "step": 8933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.148386321160007e-06,
+      "loss": 0.8308,
+      "step": 8934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.146322624498489e-06,
+      "loss": 0.867,
+      "step": 8935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1442591205655365e-06,
+      "loss": 0.7209,
+      "step": 8936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1421958094643585e-06,
+      "loss": 0.8547,
+      "step": 8937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.140132691298134e-06,
+      "loss": 0.7834,
+      "step": 8938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.138069766170051e-06,
+      "loss": 0.9476,
+      "step": 8939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.136007034183278e-06,
+      "loss": 0.9627,
+      "step": 8940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.133944495440973e-06,
+      "loss": 0.8659,
+      "step": 8941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.131882150046291e-06,
+      "loss": 0.7563,
+      "step": 8942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.129819998102369e-06,
+      "loss": 0.9256,
+      "step": 8943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.127758039712339e-06,
+      "loss": 0.9693,
+      "step": 8944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.125696274979326e-06,
+      "loss": 0.8416,
+      "step": 8945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.123634704006438e-06,
+      "loss": 0.8047,
+      "step": 8946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.121573326896782e-06,
+      "loss": 0.8583,
+      "step": 8947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.119512143753448e-06,
+      "loss": 0.8374,
+      "step": 8948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.117451154679517e-06,
+      "loss": 0.9321,
+      "step": 8949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.115390359778066e-06,
+      "loss": 0.8866,
+      "step": 8950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1133297591521555e-06,
+      "loss": 0.8786,
+      "step": 8951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.111269352904845e-06,
+      "loss": 0.9481,
+      "step": 8952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.109209141139172e-06,
+      "loss": 0.9016,
+      "step": 8953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.107149123958175e-06,
+      "loss": 0.8116,
+      "step": 8954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.105089301464879e-06,
+      "loss": 0.8772,
+      "step": 8955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.103029673762295e-06,
+      "loss": 0.9155,
+      "step": 8956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.100970240953433e-06,
+      "loss": 0.7958,
+      "step": 8957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0989110031412855e-06,
+      "loss": 0.9135,
+      "step": 8958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.09685196042884e-06,
+      "loss": 0.8996,
+      "step": 8959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.094793112919073e-06,
+      "loss": 0.6841,
+      "step": 8960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0927344607149485e-06,
+      "loss": 0.8378,
+      "step": 8961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.090676003919425e-06,
+      "loss": 0.8044,
+      "step": 8962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.088617742635446e-06,
+      "loss": 0.9262,
+      "step": 8963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.086559676965953e-06,
+      "loss": 0.9213,
+      "step": 8964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.084501807013869e-06,
+      "loss": 0.9464,
+      "step": 8965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0824441328821145e-06,
+      "loss": 0.9083,
+      "step": 8966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.080386654673598e-06,
+      "loss": 0.8316,
+      "step": 8967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.078329372491211e-06,
+      "loss": 0.8792,
+      "step": 8968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.076272286437848e-06,
+      "loss": 0.9204,
+      "step": 8969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0742153966163805e-06,
+      "loss": 0.9128,
+      "step": 8970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.072158703129685e-06,
+      "loss": 0.8966,
+      "step": 8971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.070102206080617e-06,
+      "loss": 0.8921,
+      "step": 8972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.068045905572024e-06,
+      "loss": 0.946,
+      "step": 8973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.065989801706745e-06,
+      "loss": 0.8535,
+      "step": 8974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.063933894587608e-06,
+      "loss": 0.8928,
+      "step": 8975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.061878184317433e-06,
+      "loss": 0.8964,
+      "step": 8976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.059822670999034e-06,
+      "loss": 0.9132,
+      "step": 8977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.057767354735202e-06,
+      "loss": 0.9312,
+      "step": 8978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.055712235628734e-06,
+      "loss": 1.0258,
+      "step": 8979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.053657313782406e-06,
+      "loss": 0.9899,
+      "step": 8980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.051602589298987e-06,
+      "loss": 0.9879,
+      "step": 8981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.049548062281241e-06,
+      "loss": 0.8608,
+      "step": 8982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.047493732831914e-06,
+      "loss": 0.866,
+      "step": 8983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.045439601053751e-06,
+      "loss": 0.9785,
+      "step": 8984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.043385667049478e-06,
+      "loss": 0.9021,
+      "step": 8985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.041331930921816e-06,
+      "loss": 0.9238,
+      "step": 8986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.03927839277348e-06,
+      "loss": 0.9971,
+      "step": 8987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.037225052707164e-06,
+      "loss": 0.8962,
+      "step": 8988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.035171910825566e-06,
+      "loss": 0.8852,
+      "step": 8989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.033118967231359e-06,
+      "loss": 0.8818,
+      "step": 8990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.031066222027222e-06,
+      "loss": 0.9965,
+      "step": 8991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.029013675315813e-06,
+      "loss": 0.9004,
+      "step": 8992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.026961327199779e-06,
+      "loss": 0.6427,
+      "step": 8993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.024909177781767e-06,
+      "loss": 0.6394,
+      "step": 8994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.022857227164405e-06,
+      "loss": 0.9215,
+      "step": 8995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.020805475450315e-06,
+      "loss": 0.9752,
+      "step": 8996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.018753922742111e-06,
+      "loss": 0.8602,
+      "step": 8997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0167025691423894e-06,
+      "loss": 1.0338,
+      "step": 8998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.014651414753747e-06,
+      "loss": 0.9589,
+      "step": 8999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.012600459678761e-06,
+      "loss": 0.85,
+      "step": 9000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.010549704020001e-06,
+      "loss": 0.6027,
+      "step": 9001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.008499147880038e-06,
+      "loss": 0.6342,
+      "step": 9002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.006448791361415e-06,
+      "loss": 0.8928,
+      "step": 9003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.004398634566679e-06,
+      "loss": 0.8178,
+      "step": 9004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.002348677598359e-06,
+      "loss": 0.8528,
+      "step": 9005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.000298920558973e-06,
+      "loss": 0.9724,
+      "step": 9006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.998249363551041e-06,
+      "loss": 0.9007,
+      "step": 9007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.996200006677057e-06,
+      "loss": 0.9904,
+      "step": 9008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.994150850039517e-06,
+      "loss": 0.8135,
+      "step": 9009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.992101893740901e-06,
+      "loss": 0.883,
+      "step": 9010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99005313788368e-06,
+      "loss": 0.799,
+      "step": 9011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9880045825703174e-06,
+      "loss": 0.9913,
+      "step": 9012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985956227903262e-06,
+      "loss": 0.9415,
+      "step": 9013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983908073984958e-06,
+      "loss": 0.8257,
+      "step": 9014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981860120917835e-06,
+      "loss": 0.8759,
+      "step": 9015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979812368804316e-06,
+      "loss": 0.897,
+      "step": 9016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977764817746814e-06,
+      "loss": 0.8917,
+      "step": 9017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9757174678477235e-06,
+      "loss": 0.8619,
+      "step": 9018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973670319209444e-06,
+      "loss": 0.8743,
+      "step": 9019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97162337193435e-06,
+      "loss": 0.8651,
+      "step": 9020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9695766261248155e-06,
+      "loss": 0.8913,
+      "step": 9021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9675300818832015e-06,
+      "loss": 0.8063,
+      "step": 9022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965483739311862e-06,
+      "loss": 0.9247,
+      "step": 9023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963437598513132e-06,
+      "loss": 1.0023,
+      "step": 9024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961391659589344e-06,
+      "loss": 0.8568,
+      "step": 9025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959345922642823e-06,
+      "loss": 1.0006,
+      "step": 9026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957300387775873e-06,
+      "loss": 0.9332,
+      "step": 9027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9552550550908e-06,
+      "loss": 0.8909,
+      "step": 9028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953209924689891e-06,
+      "loss": 0.8913,
+      "step": 9029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951164996675427e-06,
+      "loss": 0.9141,
+      "step": 9030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94912027114968e-06,
+      "loss": 1.0641,
+      "step": 9031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947075748214904e-06,
+      "loss": 0.8617,
+      "step": 9032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945031427973356e-06,
+      "loss": 0.9087,
+      "step": 9033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942987310527275e-06,
+      "loss": 0.8138,
+      "step": 9034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9409433959788865e-06,
+      "loss": 0.9252,
+      "step": 9035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938899684430414e-06,
+      "loss": 0.8918,
+      "step": 9036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936856175984065e-06,
+      "loss": 0.8972,
+      "step": 9037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934812870742036e-06,
+      "loss": 0.7708,
+      "step": 9038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932769768806521e-06,
+      "loss": 0.8342,
+      "step": 9039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930726870279694e-06,
+      "loss": 0.9493,
+      "step": 9040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928684175263728e-06,
+      "loss": 0.9237,
+      "step": 9041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9266416838607784e-06,
+      "loss": 0.9523,
+      "step": 9042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9245993961729945e-06,
+      "loss": 0.7971,
+      "step": 9043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9225573123025144e-06,
+      "loss": 1.0617,
+      "step": 9044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920515432351465e-06,
+      "loss": 0.8255,
+      "step": 9045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918473756421967e-06,
+      "loss": 1.0253,
+      "step": 9046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916432284616123e-06,
+      "loss": 0.9498,
+      "step": 9047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914391017036034e-06,
+      "loss": 0.8853,
+      "step": 9048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912349953783787e-06,
+      "loss": 0.9024,
+      "step": 9049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.910309094961456e-06,
+      "loss": 0.9457,
+      "step": 9050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.908268440671112e-06,
+      "loss": 0.9028,
+      "step": 9051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.906227991014806e-06,
+      "loss": 0.868,
+      "step": 9052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.904187746094589e-06,
+      "loss": 0.8145,
+      "step": 9053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.902147706012495e-06,
+      "loss": 0.923,
+      "step": 9054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.900107870870549e-06,
+      "loss": 0.9589,
+      "step": 9055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.898068240770766e-06,
+      "loss": 0.9768,
+      "step": 9056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.896028815815152e-06,
+      "loss": 0.8822,
+      "step": 9057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.893989596105704e-06,
+      "loss": 0.9855,
+      "step": 9058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.891950581744404e-06,
+      "loss": 0.8853,
+      "step": 9059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.889911772833228e-06,
+      "loss": 0.8725,
+      "step": 9060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.887873169474139e-06,
+      "loss": 0.8435,
+      "step": 9061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.885834771769088e-06,
+      "loss": 0.8064,
+      "step": 9062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.883796579820024e-06,
+      "loss": 0.8846,
+      "step": 9063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.881758593728875e-06,
+      "loss": 0.9179,
+      "step": 9064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.879720813597569e-06,
+      "loss": 0.9491,
+      "step": 9065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.877683239528017e-06,
+      "loss": 0.9357,
+      "step": 9066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.875645871622121e-06,
+      "loss": 0.7549,
+      "step": 9067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.873608709981771e-06,
+      "loss": 0.8471,
+      "step": 9068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.871571754708853e-06,
+      "loss": 0.5878,
+      "step": 9069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.869535005905233e-06,
+      "loss": 0.8313,
+      "step": 9070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8674984636727764e-06,
+      "loss": 0.93,
+      "step": 9071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.86546212811333e-06,
+      "loss": 0.8247,
+      "step": 9072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8634259993287416e-06,
+      "loss": 0.8541,
+      "step": 9073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8613900774208346e-06,
+      "loss": 0.8806,
+      "step": 9074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.859354362491427e-06,
+      "loss": 0.7929,
+      "step": 9075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.857318854642335e-06,
+      "loss": 0.9674,
+      "step": 9076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.855283553975353e-06,
+      "loss": 0.7918,
+      "step": 9077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.853248460592271e-06,
+      "loss": 0.9001,
+      "step": 9078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.851213574594868e-06,
+      "loss": 0.8361,
+      "step": 9079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.849178896084909e-06,
+      "loss": 0.8709,
+      "step": 9080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.847144425164155e-06,
+      "loss": 0.9153,
+      "step": 9081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.845110161934351e-06,
+      "loss": 0.8717,
+      "step": 9082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.843076106497235e-06,
+      "loss": 0.8922,
+      "step": 9083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.841042258954531e-06,
+      "loss": 1.0472,
+      "step": 9084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.83900861940796e-06,
+      "loss": 0.8501,
+      "step": 9085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8369751879592225e-06,
+      "loss": 0.9155,
+      "step": 9086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.834941964710017e-06,
+      "loss": 0.8112,
+      "step": 9087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.832908949762022e-06,
+      "loss": 0.8211,
+      "step": 9088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.83087614321692e-06,
+      "loss": 0.9502,
+      "step": 9089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8288435451763725e-06,
+      "loss": 0.977,
+      "step": 9090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.826811155742031e-06,
+      "loss": 0.9021,
+      "step": 9091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.824778975015541e-06,
+      "loss": 0.6623,
+      "step": 9092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.822747003098527e-06,
+      "loss": 0.8413,
+      "step": 9093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.820715240092625e-06,
+      "loss": 0.8652,
+      "step": 9094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.818683686099433e-06,
+      "loss": 0.8225,
+      "step": 9095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8166523412205635e-06,
+      "loss": 0.952,
+      "step": 9096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.814621205557603e-06,
+      "loss": 0.8493,
+      "step": 9097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.812590279212131e-06,
+      "loss": 0.9304,
+      "step": 9098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.810559562285714e-06,
+      "loss": 0.8826,
+      "step": 9099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.808529054879919e-06,
+      "loss": 0.8931,
+      "step": 9100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.806498757096291e-06,
+      "loss": 0.8922,
+      "step": 9101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.80446866903637e-06,
+      "loss": 0.9577,
+      "step": 9102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.802438790801677e-06,
+      "loss": 0.9687,
+      "step": 9103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.800409122493738e-06,
+      "loss": 0.9513,
+      "step": 9104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.798379664214059e-06,
+      "loss": 0.8553,
+      "step": 9105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.796350416064134e-06,
+      "loss": 0.8345,
+      "step": 9106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.794321378145449e-06,
+      "loss": 0.9573,
+      "step": 9107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.792292550559475e-06,
+      "loss": 0.9574,
+      "step": 9108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.790263933407685e-06,
+      "loss": 0.9782,
+      "step": 9109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.788235526791532e-06,
+      "loss": 0.6413,
+      "step": 9110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.786207330812457e-06,
+      "loss": 0.833,
+      "step": 9111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.784179345571894e-06,
+      "loss": 0.9037,
+      "step": 9112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7821515711712625e-06,
+      "loss": 0.8483,
+      "step": 9113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.780124007711982e-06,
+      "loss": 0.8703,
+      "step": 9114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.778096655295451e-06,
+      "loss": 0.8272,
+      "step": 9115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.776069514023059e-06,
+      "loss": 0.8945,
+      "step": 9116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.774042583996189e-06,
+      "loss": 0.9539,
+      "step": 9117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.772015865316205e-06,
+      "loss": 0.9762,
+      "step": 9118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.769989358084476e-06,
+      "loss": 0.925,
+      "step": 9119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.767963062402344e-06,
+      "loss": 0.9402,
+      "step": 9120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7659369783711514e-06,
+      "loss": 0.8361,
+      "step": 9121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7639111060922184e-06,
+      "loss": 0.6609,
+      "step": 9122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.761885445666873e-06,
+      "loss": 0.9101,
+      "step": 9123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.759859997196414e-06,
+      "loss": 0.8877,
+      "step": 9124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.757834760782141e-06,
+      "loss": 0.8329,
+      "step": 9125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.755809736525333e-06,
+      "loss": 0.8537,
+      "step": 9126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7537849245272744e-06,
+      "loss": 0.8918,
+      "step": 9127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.75176032488922e-06,
+      "loss": 0.8208,
+      "step": 9128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.749735937712431e-06,
+      "loss": 0.9532,
+      "step": 9129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7477117630981496e-06,
+      "loss": 0.9091,
+      "step": 9130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.745687801147602e-06,
+      "loss": 0.9329,
+      "step": 9131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.743664051962015e-06,
+      "loss": 0.9233,
+      "step": 9132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.741640515642594e-06,
+      "loss": 0.64,
+      "step": 9133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.739617192290546e-06,
+      "loss": 0.8887,
+      "step": 9134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.737594082007059e-06,
+      "loss": 0.9317,
+      "step": 9135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.735571184893308e-06,
+      "loss": 0.9084,
+      "step": 9136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.733548501050465e-06,
+      "loss": 0.9225,
+      "step": 9137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.731526030579684e-06,
+      "loss": 1.0143,
+      "step": 9138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.729503773582117e-06,
+      "loss": 0.8498,
+      "step": 9139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.727481730158898e-06,
+      "loss": 0.9056,
+      "step": 9140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.725459900411152e-06,
+      "loss": 1.0525,
+      "step": 9141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.723438284439996e-06,
+      "loss": 0.8801,
+      "step": 9142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.721416882346528e-06,
+      "loss": 0.8574,
+      "step": 9143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.71939569423185e-06,
+      "loss": 0.8514,
+      "step": 9144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7173747201970405e-06,
+      "loss": 0.848,
+      "step": 9145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.715353960343173e-06,
+      "loss": 0.8441,
+      "step": 9146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.713333414771305e-06,
+      "loss": 0.8367,
+      "step": 9147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.711313083582493e-06,
+      "loss": 0.8781,
+      "step": 9148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.709292966877774e-06,
+      "loss": 0.8946,
+      "step": 9149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7072730647581784e-06,
+      "loss": 0.9003,
+      "step": 9150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7052533773247245e-06,
+      "loss": 0.8034,
+      "step": 9151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.703233904678413e-06,
+      "loss": 0.8177,
+      "step": 9152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701214646920254e-06,
+      "loss": 0.9317,
+      "step": 9153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699195604151226e-06,
+      "loss": 0.859,
+      "step": 9154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697176776472307e-06,
+      "loss": 0.8783,
+      "step": 9155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69515816398446e-06,
+      "loss": 0.9078,
+      "step": 9156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.693139766788635e-06,
+      "loss": 0.6863,
+      "step": 9157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691121584985783e-06,
+      "loss": 0.9876,
+      "step": 9158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689103618676835e-06,
+      "loss": 0.8314,
+      "step": 9159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687085867962712e-06,
+      "loss": 0.9444,
+      "step": 9160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685068332944324e-06,
+      "loss": 0.9468,
+      "step": 9161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683051013722571e-06,
+      "loss": 0.834,
+      "step": 9162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.681033910398338e-06,
+      "loss": 0.9523,
+      "step": 9163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6790170230725126e-06,
+      "loss": 0.9099,
+      "step": 9164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677000351845958e-06,
+      "loss": 0.6189,
+      "step": 9165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674983896819532e-06,
+      "loss": 0.8498,
+      "step": 9166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6729676580940816e-06,
+      "loss": 0.9252,
+      "step": 9167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6709516357704345e-06,
+      "loss": 0.6076,
+      "step": 9168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668935829949426e-06,
+      "loss": 0.827,
+      "step": 9169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666920240731865e-06,
+      "loss": 0.8812,
+      "step": 9170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664904868218555e-06,
+      "loss": 0.8599,
+      "step": 9171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662889712510282e-06,
+      "loss": 0.895,
+      "step": 9172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660874773707838e-06,
+      "loss": 0.8095,
+      "step": 9173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658860051911987e-06,
+      "loss": 0.7739,
+      "step": 9174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65684554722349e-06,
+      "loss": 0.9456,
+      "step": 9175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.654831259743094e-06,
+      "loss": 0.5759,
+      "step": 9176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6528171895715345e-06,
+      "loss": 0.8823,
+      "step": 9177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650803336809545e-06,
+      "loss": 0.8773,
+      "step": 9178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64878970155784e-06,
+      "loss": 0.7911,
+      "step": 9179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64677628391712e-06,
+      "loss": 0.9411,
+      "step": 9180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644763083988083e-06,
+      "loss": 0.845,
+      "step": 9181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642750101871408e-06,
+      "loss": 0.8946,
+      "step": 9182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640737337667774e-06,
+      "loss": 0.8727,
+      "step": 9183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63872479147784e-06,
+      "loss": 0.8833,
+      "step": 9184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636712463402255e-06,
+      "loss": 0.9803,
+      "step": 9185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63470035354166e-06,
+      "loss": 0.9866,
+      "step": 9186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632688461996681e-06,
+      "loss": 0.8552,
+      "step": 9187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6306767888679405e-06,
+      "loss": 0.8085,
+      "step": 9188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62866533425604e-06,
+      "loss": 0.9751,
+      "step": 9189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626654098261583e-06,
+      "loss": 0.8802,
+      "step": 9190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62464308098515e-06,
+      "loss": 0.8681,
+      "step": 9191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622632282527314e-06,
+      "loss": 0.905,
+      "step": 9192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620621702988643e-06,
+      "loss": 0.9073,
+      "step": 9193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6186113424696846e-06,
+      "loss": 0.8914,
+      "step": 9194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.616601201070983e-06,
+      "loss": 0.5866,
+      "step": 9195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614591278893069e-06,
+      "loss": 0.6337,
+      "step": 9196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.612581576036456e-06,
+      "loss": 0.8571,
+      "step": 9197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6105720926016595e-06,
+      "loss": 0.9124,
+      "step": 9198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.608562828689175e-06,
+      "loss": 0.891,
+      "step": 9199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6065537843994886e-06,
+      "loss": 0.8741,
+      "step": 9200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.604544959833076e-06,
+      "loss": 0.8893,
+      "step": 9201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.602536355090398e-06,
+      "loss": 0.9557,
+      "step": 9202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.600527970271914e-06,
+      "loss": 0.8996,
+      "step": 9203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.598519805478064e-06,
+      "loss": 0.9685,
+      "step": 9204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5965118608092815e-06,
+      "loss": 0.8128,
+      "step": 9205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.594504136365984e-06,
+      "loss": 0.8937,
+      "step": 9206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.592496632248579e-06,
+      "loss": 0.9559,
+      "step": 9207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5904893485574705e-06,
+      "loss": 0.9305,
+      "step": 9208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.588482285393044e-06,
+      "loss": 0.8718,
+      "step": 9209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.586475442855676e-06,
+      "loss": 0.9374,
+      "step": 9210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.584468821045732e-06,
+      "loss": 0.7654,
+      "step": 9211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.582462420063561e-06,
+      "loss": 0.897,
+      "step": 9212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.580456240009516e-06,
+      "loss": 0.9006,
+      "step": 9213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.578450280983925e-06,
+      "loss": 0.9411,
+      "step": 9214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.576444543087107e-06,
+      "loss": 0.8749,
+      "step": 9215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.574439026419371e-06,
+      "loss": 1.0087,
+      "step": 9216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.572433731081022e-06,
+      "loss": 0.9382,
+      "step": 9217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.570428657172345e-06,
+      "loss": 0.8979,
+      "step": 9218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.568423804793618e-06,
+      "loss": 0.6177,
+      "step": 9219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.566419174045101e-06,
+      "loss": 0.7835,
+      "step": 9220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.564414765027057e-06,
+      "loss": 0.9866,
+      "step": 9221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.562410577839723e-06,
+      "loss": 0.8433,
+      "step": 9222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.560406612583338e-06,
+      "loss": 0.8273,
+      "step": 9223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5584028693581214e-06,
+      "loss": 0.8422,
+      "step": 9224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.556399348264282e-06,
+      "loss": 1.0654,
+      "step": 9225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55439604940202e-06,
+      "loss": 0.9343,
+      "step": 9226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.552392972871519e-06,
+      "loss": 0.7746,
+      "step": 9227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5503901187729635e-06,
+      "loss": 0.8477,
+      "step": 9228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.548387487206518e-06,
+      "loss": 1.0011,
+      "step": 9229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.546385078272335e-06,
+      "loss": 0.8945,
+      "step": 9230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.544382892070559e-06,
+      "loss": 1.0026,
+      "step": 9231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.542380928701316e-06,
+      "loss": 0.7878,
+      "step": 9232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.540379188264739e-06,
+      "loss": 0.9066,
+      "step": 9233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.538377670860935e-06,
+      "loss": 0.8726,
+      "step": 9234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.536376376589998e-06,
+      "loss": 0.8908,
+      "step": 9235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.53437530555202e-06,
+      "loss": 0.8648,
+      "step": 9236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.532374457847074e-06,
+      "loss": 1.0087,
+      "step": 9237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.530373833575231e-06,
+      "loss": 0.9472,
+      "step": 9238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.528373432836542e-06,
+      "loss": 0.934,
+      "step": 9239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.526373255731052e-06,
+      "loss": 0.8912,
+      "step": 9240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.524373302358788e-06,
+      "loss": 0.9852,
+      "step": 9241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.522373572819778e-06,
+      "loss": 0.9233,
+      "step": 9242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.52037406721403e-06,
+      "loss": 0.945,
+      "step": 9243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.518374785641539e-06,
+      "loss": 0.8725,
+      "step": 9244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5163757282022955e-06,
+      "loss": 0.8868,
+      "step": 9245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.51437689499627e-06,
+      "loss": 0.8889,
+      "step": 9246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.512378286123437e-06,
+      "loss": 0.8691,
+      "step": 9247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.510379901683742e-06,
+      "loss": 0.9107,
+      "step": 9248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5083817417771335e-06,
+      "loss": 0.9014,
+      "step": 9249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.506383806503537e-06,
+      "loss": 0.8877,
+      "step": 9250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.504386095962872e-06,
+      "loss": 0.9098,
+      "step": 9251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.502388610255049e-06,
+      "loss": 0.8497,
+      "step": 9252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.500391349479971e-06,
+      "loss": 0.9675,
+      "step": 9253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.498394313737521e-06,
+      "loss": 0.8182,
+      "step": 9254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.49639750312757e-06,
+      "loss": 0.6153,
+      "step": 9255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.494400917749986e-06,
+      "loss": 0.8266,
+      "step": 9256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.492404557704616e-06,
+      "loss": 0.9123,
+      "step": 9257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.490408423091308e-06,
+      "loss": 0.9372,
+      "step": 9258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.488412514009888e-06,
+      "loss": 0.8136,
+      "step": 9259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4864168305601776e-06,
+      "loss": 0.5957,
+      "step": 9260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.484421372841982e-06,
+      "loss": 0.9523,
+      "step": 9261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.48242614095509e-06,
+      "loss": 0.9119,
+      "step": 9262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4804311349993e-06,
+      "loss": 0.9308,
+      "step": 9263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4784363550743794e-06,
+      "loss": 0.8086,
+      "step": 9264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.476441801280088e-06,
+      "loss": 0.9284,
+      "step": 9265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.474447473716176e-06,
+      "loss": 0.8775,
+      "step": 9266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.472453372482389e-06,
+      "loss": 0.905,
+      "step": 9267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.470459497678451e-06,
+      "loss": 0.8455,
+      "step": 9268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.468465849404081e-06,
+      "loss": 0.9243,
+      "step": 9269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.466472427758982e-06,
+      "loss": 0.8333,
+      "step": 9270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.464479232842845e-06,
+      "loss": 0.8149,
+      "step": 9271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.462486264755361e-06,
+      "loss": 0.9341,
+      "step": 9272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.460493523596198e-06,
+      "loss": 0.888,
+      "step": 9273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.458501009465016e-06,
+      "loss": 0.9299,
+      "step": 9274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.456508722461462e-06,
+      "loss": 0.6644,
+      "step": 9275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.454516662685173e-06,
+      "loss": 0.96,
+      "step": 9276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.452524830235781e-06,
+      "loss": 0.9478,
+      "step": 9277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.450533225212896e-06,
+      "loss": 0.8217,
+      "step": 9278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.448541847716123e-06,
+      "loss": 0.8072,
+      "step": 9279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4465506978450524e-06,
+      "loss": 0.8805,
+      "step": 9280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.444559775699263e-06,
+      "loss": 0.9262,
+      "step": 9281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.442569081378328e-06,
+      "loss": 0.8211,
+      "step": 9282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.440578614981802e-06,
+      "loss": 0.9092,
+      "step": 9283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.438588376609237e-06,
+      "loss": 0.8417,
+      "step": 9284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.436598366360164e-06,
+      "loss": 0.8451,
+      "step": 9285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.434608584334102e-06,
+      "loss": 0.8895,
+      "step": 9286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.432619030630573e-06,
+      "loss": 0.8522,
+      "step": 9287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.430629705349073e-06,
+      "loss": 0.8042,
+      "step": 9288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.428640608589091e-06,
+      "loss": 0.951,
+      "step": 9289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.426651740450104e-06,
+      "loss": 0.9527,
+      "step": 9290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4246631010315755e-06,
+      "loss": 0.9131,
+      "step": 9291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4226746904329694e-06,
+      "loss": 0.9097,
+      "step": 9292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.420686508753724e-06,
+      "loss": 0.8204,
+      "step": 9293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.418698556093272e-06,
+      "loss": 0.6263,
+      "step": 9294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416710832551034e-06,
+      "loss": 0.9868,
+      "step": 9295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414723338226415e-06,
+      "loss": 1.0027,
+      "step": 9296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412736073218821e-06,
+      "loss": 0.7705,
+      "step": 9297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410749037627635e-06,
+      "loss": 1.0461,
+      "step": 9298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4087622315522304e-06,
+      "loss": 0.9,
+      "step": 9299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406775655091972e-06,
+      "loss": 0.9165,
+      "step": 9300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404789308346205e-06,
+      "loss": 0.9663,
+      "step": 9301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402803191414281e-06,
+      "loss": 0.9796,
+      "step": 9302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400817304395525e-06,
+      "loss": 0.7968,
+      "step": 9303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398831647389251e-06,
+      "loss": 0.8457,
+      "step": 9304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396846220494767e-06,
+      "loss": 0.884,
+      "step": 9305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394861023811364e-06,
+      "loss": 0.9371,
+      "step": 9306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392876057438332e-06,
+      "loss": 0.8086,
+      "step": 9307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390891321474938e-06,
+      "loss": 0.9749,
+      "step": 9308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388906816020442e-06,
+      "loss": 0.8367,
+      "step": 9309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3869225411740885e-06,
+      "loss": 0.7882,
+      "step": 9310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384938497035124e-06,
+      "loss": 0.8395,
+      "step": 9311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3829546837027655e-06,
+      "loss": 0.9211,
+      "step": 9312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380971101276231e-06,
+      "loss": 1.0408,
+      "step": 9313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378987749854716e-06,
+      "loss": 0.8252,
+      "step": 9314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3770046295374205e-06,
+      "loss": 0.943,
+      "step": 9315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375021740423514e-06,
+      "loss": 1.078,
+      "step": 9316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373039082612174e-06,
+      "loss": 0.8708,
+      "step": 9317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3710566562025514e-06,
+      "loss": 0.7286,
+      "step": 9318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369074461293789e-06,
+      "loss": 0.834,
+      "step": 9319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367092497985021e-06,
+      "loss": 0.8258,
+      "step": 9320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365110766375365e-06,
+      "loss": 0.9218,
+      "step": 9321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363129266563939e-06,
+      "loss": 0.8971,
+      "step": 9322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361147998649836e-06,
+      "loss": 0.8845,
+      "step": 9323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359166962732141e-06,
+      "loss": 0.8405,
+      "step": 9324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35718615890993e-06,
+      "loss": 0.912,
+      "step": 9325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355205587282264e-06,
+      "loss": 0.8917,
+      "step": 9326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3532252479482005e-06,
+      "loss": 0.9153,
+      "step": 9327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351245141006775e-06,
+      "loss": 0.9659,
+      "step": 9328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349265266557018e-06,
+      "loss": 0.9685,
+      "step": 9329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347285624697943e-06,
+      "loss": 0.8997,
+      "step": 9330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345306215528554e-06,
+      "loss": 0.9043,
+      "step": 9331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343327039147852e-06,
+      "loss": 0.8811,
+      "step": 9332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341348095654813e-06,
+      "loss": 0.9649,
+      "step": 9333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339369385148407e-06,
+      "loss": 0.6243,
+      "step": 9334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337390907727593e-06,
+      "loss": 0.889,
+      "step": 9335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335412663491319e-06,
+      "loss": 0.8962,
+      "step": 9336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333434652538523e-06,
+      "loss": 0.991,
+      "step": 9337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331456874968124e-06,
+      "loss": 0.9992,
+      "step": 9338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329479330879034e-06,
+      "loss": 0.9995,
+      "step": 9339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.32750202037015e-06,
+      "loss": 0.8832,
+      "step": 9340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.32552494354037e-06,
+      "loss": 0.9579,
+      "step": 9341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.323548100488565e-06,
+      "loss": 0.9004,
+      "step": 9342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3215714913136e-06,
+      "loss": 0.9121,
+      "step": 9343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.319595116114329e-06,
+      "loss": 0.9272,
+      "step": 9344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.317618974989589e-06,
+      "loss": 0.8276,
+      "step": 9345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.315643068038217e-06,
+      "loss": 0.8655,
+      "step": 9346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.31366739535903e-06,
+      "loss": 0.7702,
+      "step": 9347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.311691957050834e-06,
+      "loss": 0.9328,
+      "step": 9348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.309716753212424e-06,
+      "loss": 0.9192,
+      "step": 9349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.307741783942581e-06,
+      "loss": 1.046,
+      "step": 9350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.305767049340074e-06,
+      "loss": 0.9211,
+      "step": 9351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.30379254950367e-06,
+      "loss": 0.9378,
+      "step": 9352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.301818284532114e-06,
+      "loss": 0.971,
+      "step": 9353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.299844254524143e-06,
+      "loss": 0.8417,
+      "step": 9354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2978704595784754e-06,
+      "loss": 0.8173,
+      "step": 9355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.295896899793828e-06,
+      "loss": 0.9211,
+      "step": 9356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.293923575268904e-06,
+      "loss": 1.0224,
+      "step": 9357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.291950486102391e-06,
+      "loss": 0.9916,
+      "step": 9358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.289977632392965e-06,
+      "loss": 0.8387,
+      "step": 9359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.288005014239289e-06,
+      "loss": 0.9082,
+      "step": 9360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.286032631740023e-06,
+      "loss": 0.6738,
+      "step": 9361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.284060484993807e-06,
+      "loss": 0.8025,
+      "step": 9362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2820885740992715e-06,
+      "loss": 0.8756,
+      "step": 9363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2801168991550325e-06,
+      "loss": 0.8961,
+      "step": 9364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.278145460259692e-06,
+      "loss": 0.836,
+      "step": 9365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.276174257511857e-06,
+      "loss": 0.8546,
+      "step": 9366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.274203291010103e-06,
+      "loss": 0.8793,
+      "step": 9367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.272232560853003e-06,
+      "loss": 0.8147,
+      "step": 9368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.270262067139116e-06,
+      "loss": 0.8688,
+      "step": 9369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.268291809966984e-06,
+      "loss": 0.8535,
+      "step": 9370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.266321789435152e-06,
+      "loss": 0.5989,
+      "step": 9371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.264352005642142e-06,
+      "loss": 0.9247,
+      "step": 9372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.262382458686461e-06,
+      "loss": 0.807,
+      "step": 9373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.260413148666614e-06,
+      "loss": 0.9223,
+      "step": 9374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2584440756810825e-06,
+      "loss": 0.876,
+      "step": 9375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2564752398283525e-06,
+      "loss": 0.8717,
+      "step": 9376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.25450664120688e-06,
+      "loss": 0.9547,
+      "step": 9377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.252538279915125e-06,
+      "loss": 0.8643,
+      "step": 9378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.250570156051527e-06,
+      "loss": 0.8673,
+      "step": 9379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.248602269714507e-06,
+      "loss": 0.9356,
+      "step": 9380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.246634621002493e-06,
+      "loss": 0.861,
+      "step": 9381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.244667210013885e-06,
+      "loss": 0.9318,
+      "step": 9382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.242700036847078e-06,
+      "loss": 0.8147,
+      "step": 9383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2407331016004525e-06,
+      "loss": 0.9084,
+      "step": 9384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.238766404372374e-06,
+      "loss": 0.8638,
+      "step": 9385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2367999452612086e-06,
+      "loss": 0.954,
+      "step": 9386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.234833724365296e-06,
+      "loss": 0.7821,
+      "step": 9387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.232867741782973e-06,
+      "loss": 0.8174,
+      "step": 9388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.23090199761256e-06,
+      "loss": 0.8874,
+      "step": 9389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2289364919523636e-06,
+      "loss": 0.9268,
+      "step": 9390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.22697122490069e-06,
+      "loss": 0.8565,
+      "step": 9391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.22500619655582e-06,
+      "loss": 0.8558,
+      "step": 9392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.223041407016028e-06,
+      "loss": 0.9538,
+      "step": 9393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.221076856379579e-06,
+      "loss": 0.8479,
+      "step": 9394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.219112544744716e-06,
+      "loss": 0.9582,
+      "step": 9395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2171484722096865e-06,
+      "loss": 0.8831,
+      "step": 9396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.215184638872713e-06,
+      "loss": 0.9073,
+      "step": 9397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.213221044832011e-06,
+      "loss": 0.8819,
+      "step": 9398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2112576901857815e-06,
+      "loss": 0.8339,
+      "step": 9399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.209294575032211e-06,
+      "loss": 0.9523,
+      "step": 9400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2073316994694875e-06,
+      "loss": 0.9156,
+      "step": 9401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.205369063595774e-06,
+      "loss": 1.0415,
+      "step": 9402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.203406667509221e-06,
+      "loss": 0.9613,
+      "step": 9403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.201444511307969e-06,
+      "loss": 0.819,
+      "step": 9404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1994825950901595e-06,
+      "loss": 0.7516,
+      "step": 9405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1975209189539044e-06,
+      "loss": 0.9231,
+      "step": 9406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.19555948299731e-06,
+      "loss": 0.8911,
+      "step": 9407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.193598287318473e-06,
+      "loss": 0.8111,
+      "step": 9408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.191637332015469e-06,
+      "loss": 0.9033,
+      "step": 9409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.189676617186374e-06,
+      "loss": 0.8675,
+      "step": 9410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.187716142929251e-06,
+      "loss": 0.8727,
+      "step": 9411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1857559093421425e-06,
+      "loss": 0.9175,
+      "step": 9412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.183795916523082e-06,
+      "loss": 0.9148,
+      "step": 9413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.181836164570092e-06,
+      "loss": 0.8608,
+      "step": 9414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.17987665358118e-06,
+      "loss": 0.8628,
+      "step": 9415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.177917383654349e-06,
+      "loss": 0.8442,
+      "step": 9416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.175958354887587e-06,
+      "loss": 0.8436,
+      "step": 9417
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.173999567378862e-06,
+      "loss": 1.0152,
+      "step": 9418
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.172041021226139e-06,
+      "loss": 1.0089,
+      "step": 9419
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.170082716527365e-06,
+      "loss": 0.9311,
+      "step": 9420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.168124653380484e-06,
+      "loss": 0.8145,
+      "step": 9421
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.16616683188342e-06,
+      "loss": 0.88,
+      "step": 9422
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.164209252134085e-06,
+      "loss": 0.9538,
+      "step": 9423
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.162251914230381e-06,
+      "loss": 0.9855,
+      "step": 9424
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1602948182701925e-06,
+      "loss": 0.804,
+      "step": 9425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.158337964351407e-06,
+      "loss": 0.9088,
+      "step": 9426
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1563813525718866e-06,
+      "loss": 0.8111,
+      "step": 9427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.154424983029482e-06,
+      "loss": 0.8848,
+      "step": 9428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.152468855822031e-06,
+      "loss": 0.8106,
+      "step": 9429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.150512971047373e-06,
+      "loss": 0.9379,
+      "step": 9430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.148557328803316e-06,
+      "loss": 0.8353,
+      "step": 9431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.146601929187669e-06,
+      "loss": 0.8773,
+      "step": 9432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1446467722982256e-06,
+      "loss": 0.9124,
+      "step": 9433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142691858232758e-06,
+      "loss": 0.913,
+      "step": 9434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.140737187089047e-06,
+      "loss": 0.9033,
+      "step": 9435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.13878275896484e-06,
+      "loss": 0.9368,
+      "step": 9436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136828573957886e-06,
+      "loss": 0.8669,
+      "step": 9437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1348746321659135e-06,
+      "loss": 0.6111,
+      "step": 9438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1329209336866405e-06,
+      "loss": 0.7382,
+      "step": 9439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130967478617776e-06,
+      "loss": 0.8648,
+      "step": 9440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.129014267057022e-06,
+      "loss": 0.9381,
+      "step": 9441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127061299102056e-06,
+      "loss": 0.9699,
+      "step": 9442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12510857485055e-06,
+      "loss": 0.8221,
+      "step": 9443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1231560944001635e-06,
+      "loss": 0.6111,
+      "step": 9444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1212038578485365e-06,
+      "loss": 0.8271,
+      "step": 9445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119251865293315e-06,
+      "loss": 0.9503,
+      "step": 9446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117300116832114e-06,
+      "loss": 0.8254,
+      "step": 9447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1153486125625465e-06,
+      "loss": 0.8989,
+      "step": 9448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113397352582208e-06,
+      "loss": 0.98,
+      "step": 9449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111446336988682e-06,
+      "loss": 0.8415,
+      "step": 9450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109495565879548e-06,
+      "loss": 0.8167,
+      "step": 9451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107545039352364e-06,
+      "loss": 0.8863,
+      "step": 9452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105594757504679e-06,
+      "loss": 0.8809,
+      "step": 9453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103644720434027e-06,
+      "loss": 0.8871,
+      "step": 9454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101694928237938e-06,
+      "loss": 0.9034,
+      "step": 9455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099745381013922e-06,
+      "loss": 0.9247,
+      "step": 9456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0977960788594785e-06,
+      "loss": 0.8271,
+      "step": 9457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0958470218720956e-06,
+      "loss": 0.8512,
+      "step": 9458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.093898210149244e-06,
+      "loss": 0.913,
+      "step": 9459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091949643788394e-06,
+      "loss": 0.9548,
+      "step": 9460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090001322886995e-06,
+      "loss": 0.8754,
+      "step": 9461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088053247542485e-06,
+      "loss": 0.8998,
+      "step": 9462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08610541785229e-06,
+      "loss": 0.8393,
+      "step": 9463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084157833913818e-06,
+      "loss": 0.8896,
+      "step": 9464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082210495824482e-06,
+      "loss": 0.7647,
+      "step": 9465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080263403681666e-06,
+      "loss": 0.9053,
+      "step": 9466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078316557582748e-06,
+      "loss": 0.8357,
+      "step": 9467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07636995762509e-06,
+      "loss": 0.9074,
+      "step": 9468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074423603906044e-06,
+      "loss": 0.838,
+      "step": 9469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072477496522959e-06,
+      "loss": 0.8718,
+      "step": 9470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0705316355731504e-06,
+      "loss": 0.8913,
+      "step": 9471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068586021153945e-06,
+      "loss": 0.915,
+      "step": 9472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066640653362642e-06,
+      "loss": 0.9594,
+      "step": 9473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064695532296527e-06,
+      "loss": 0.8636,
+      "step": 9474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062750658052888e-06,
+      "loss": 0.7778,
+      "step": 9475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060806030728988e-06,
+      "loss": 0.8261,
+      "step": 9476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058861650422078e-06,
+      "loss": 0.8772,
+      "step": 9477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0569175172294025e-06,
+      "loss": 0.8474,
+      "step": 9478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0549736312481845e-06,
+      "loss": 0.9259,
+      "step": 9479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05302999257565e-06,
+      "loss": 0.9983,
+      "step": 9480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051086601308999e-06,
+      "loss": 0.8831,
+      "step": 9481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049143457545425e-06,
+      "loss": 0.9478,
+      "step": 9482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047200561382106e-06,
+      "loss": 0.8626,
+      "step": 9483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.045257912916206e-06,
+      "loss": 0.6141,
+      "step": 9484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.043315512244887e-06,
+      "loss": 0.9386,
+      "step": 9485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.041373359465289e-06,
+      "loss": 0.9028,
+      "step": 9486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.039431454674543e-06,
+      "loss": 0.6291,
+      "step": 9487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0374897979697636e-06,
+      "loss": 1.0452,
+      "step": 9488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0355483894480554e-06,
+      "loss": 0.8543,
+      "step": 9489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.033607229206517e-06,
+      "loss": 0.8229,
+      "step": 9490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.031666317342226e-06,
+      "loss": 0.8509,
+      "step": 9491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.029725653952251e-06,
+      "loss": 0.8431,
+      "step": 9492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0277852391336464e-06,
+      "loss": 0.9223,
+      "step": 9493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.025845072983453e-06,
+      "loss": 0.8322,
+      "step": 9494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.023905155598707e-06,
+      "loss": 0.9787,
+      "step": 9495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0219654870764255e-06,
+      "loss": 0.9051,
+      "step": 9496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.020026067513614e-06,
+      "loss": 0.6357,
+      "step": 9497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0180868970072604e-06,
+      "loss": 0.8905,
+      "step": 9498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.016147975654355e-06,
+      "loss": 0.8405,
+      "step": 9499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.014209303551862e-06,
+      "loss": 0.8932,
+      "step": 9500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.012270880796738e-06,
+      "loss": 0.8388,
+      "step": 9501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.010332707485925e-06,
+      "loss": 1.0151,
+      "step": 9502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.008394783716352e-06,
+      "loss": 0.8701,
+      "step": 9503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.006457109584943e-06,
+      "loss": 0.8661,
+      "step": 9504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.004519685188603e-06,
+      "loss": 0.8692,
+      "step": 9505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.002582510624227e-06,
+      "loss": 0.9003,
+      "step": 9506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.000645585988693e-06,
+      "loss": 0.9586,
+      "step": 9507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.998708911378871e-06,
+      "loss": 0.9673,
+      "step": 9508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.996772486891614e-06,
+      "loss": 0.9663,
+      "step": 9509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.994836312623771e-06,
+      "loss": 0.9699,
+      "step": 9510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.992900388672172e-06,
+      "loss": 0.8791,
+      "step": 9511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.990964715133632e-06,
+      "loss": 0.8503,
+      "step": 9512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.989029292104962e-06,
+      "loss": 0.5864,
+      "step": 9513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9870941196829496e-06,
+      "loss": 0.874,
+      "step": 9514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.985159197964382e-06,
+      "loss": 0.929,
+      "step": 9515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.983224527046026e-06,
+      "loss": 0.9362,
+      "step": 9516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.981290107024637e-06,
+      "loss": 0.9669,
+      "step": 9517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.979355937996957e-06,
+      "loss": 0.8436,
+      "step": 9518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.977422020059716e-06,
+      "loss": 0.8685,
+      "step": 9519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.975488353309637e-06,
+      "loss": 1.0157,
+      "step": 9520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.973554937843423e-06,
+      "loss": 0.5726,
+      "step": 9521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.971621773757768e-06,
+      "loss": 0.9136,
+      "step": 9522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.969688861149349e-06,
+      "loss": 0.8227,
+      "step": 9523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9677562001148395e-06,
+      "loss": 0.8866,
+      "step": 9524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.965823790750893e-06,
+      "loss": 0.8532,
+      "step": 9525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.963891633154152e-06,
+      "loss": 0.925,
+      "step": 9526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9619597274212475e-06,
+      "loss": 0.9769,
+      "step": 9527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.960028073648791e-06,
+      "loss": 0.9313,
+      "step": 9528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.958096671933398e-06,
+      "loss": 0.829,
+      "step": 9529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.956165522371656e-06,
+      "loss": 0.9383,
+      "step": 9530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.954234625060145e-06,
+      "loss": 0.9663,
+      "step": 9531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.952303980095432e-06,
+      "loss": 0.8303,
+      "step": 9532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.950373587574067e-06,
+      "loss": 0.8863,
+      "step": 9533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.948443447592598e-06,
+      "loss": 0.9434,
+      "step": 9534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9465135602475564e-06,
+      "loss": 1.0154,
+      "step": 9535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.944583925635457e-06,
+      "loss": 0.9815,
+      "step": 9536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.942654543852801e-06,
+      "loss": 0.8734,
+      "step": 9537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.940725414996083e-06,
+      "loss": 0.9024,
+      "step": 9538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.938796539161775e-06,
+      "loss": 0.9558,
+      "step": 9539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.936867916446354e-06,
+      "loss": 0.879,
+      "step": 9540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.934939546946267e-06,
+      "loss": 0.6516,
+      "step": 9541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.933011430757957e-06,
+      "loss": 0.999,
+      "step": 9542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.931083567977846e-06,
+      "loss": 0.9281,
+      "step": 9543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.929155958702359e-06,
+      "loss": 0.6052,
+      "step": 9544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.927228603027893e-06,
+      "loss": 0.8892,
+      "step": 9545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.925301501050842e-06,
+      "loss": 1.0012,
+      "step": 9546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9233746528675795e-06,
+      "loss": 0.8006,
+      "step": 9547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.921448058574468e-06,
+      "loss": 0.9093,
+      "step": 9548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.919521718267867e-06,
+      "loss": 0.9298,
+      "step": 9549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.917595632044113e-06,
+      "loss": 0.9287,
+      "step": 9550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.915669799999531e-06,
+      "loss": 0.9259,
+      "step": 9551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.913744222230435e-06,
+      "loss": 0.8456,
+      "step": 9552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.911818898833125e-06,
+      "loss": 0.9384,
+      "step": 9553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.909893829903895e-06,
+      "loss": 0.9682,
+      "step": 9554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9079690155390156e-06,
+      "loss": 0.9568,
+      "step": 9555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.906044455834753e-06,
+      "loss": 0.6038,
+      "step": 9556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.904120150887356e-06,
+      "loss": 1.0238,
+      "step": 9557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9021961007930576e-06,
+      "loss": 0.8296,
+      "step": 9558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.900272305648091e-06,
+      "loss": 0.8874,
+      "step": 9559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.898348765548665e-06,
+      "loss": 1.0305,
+      "step": 9560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.896425480590979e-06,
+      "loss": 0.9916,
+      "step": 9561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.894502450871218e-06,
+      "loss": 0.9374,
+      "step": 9562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.892579676485552e-06,
+      "loss": 0.8764,
+      "step": 9563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.890657157530153e-06,
+      "loss": 0.8606,
+      "step": 9564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888734894101158e-06,
+      "loss": 0.9661,
+      "step": 9565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8868128862947115e-06,
+      "loss": 0.8814,
+      "step": 9566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.884891134206932e-06,
+      "loss": 0.9188,
+      "step": 9567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.882969637933925e-06,
+      "loss": 0.8983,
+      "step": 9568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.881048397571797e-06,
+      "loss": 0.8268,
+      "step": 9569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.879127413216628e-06,
+      "loss": 0.8557,
+      "step": 9570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.877206684964487e-06,
+      "loss": 0.9111,
+      "step": 9571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.875286212911436e-06,
+      "loss": 0.634,
+      "step": 9572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.873365997153516e-06,
+      "loss": 0.8499,
+      "step": 9573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.871446037786767e-06,
+      "loss": 0.7746,
+      "step": 9574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.869526334907205e-06,
+      "loss": 0.8636,
+      "step": 9575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.86760688861084e-06,
+      "loss": 0.7553,
+      "step": 9576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.865687698993664e-06,
+      "loss": 0.9339,
+      "step": 9577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.863768766151656e-06,
+      "loss": 0.8683,
+      "step": 9578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8618500901807916e-06,
+      "loss": 1.0275,
+      "step": 9579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859931671177024e-06,
+      "loss": 0.84,
+      "step": 9580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8580135092362965e-06,
+      "loss": 0.5961,
+      "step": 9581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85609560445454e-06,
+      "loss": 0.8623,
+      "step": 9582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854177956927665e-06,
+      "loss": 0.8624,
+      "step": 9583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852260566751588e-06,
+      "loss": 0.9209,
+      "step": 9584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850343434022194e-06,
+      "loss": 0.8997,
+      "step": 9585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848426558835362e-06,
+      "loss": 0.8434,
+      "step": 9586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84650994128696e-06,
+      "loss": 0.6057,
+      "step": 9587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844593581472836e-06,
+      "loss": 0.8184,
+      "step": 9588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8426774794888375e-06,
+      "loss": 0.9356,
+      "step": 9589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840761635430788e-06,
+      "loss": 0.8984,
+      "step": 9590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8388460493945046e-06,
+      "loss": 0.9269,
+      "step": 9591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836930721475781e-06,
+      "loss": 0.9566,
+      "step": 9592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835015651770416e-06,
+      "loss": 0.9535,
+      "step": 9593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833100840374179e-06,
+      "loss": 0.9105,
+      "step": 9594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831186287382837e-06,
+      "loss": 0.9428,
+      "step": 9595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829271992892136e-06,
+      "loss": 0.9133,
+      "step": 9596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827357956997812e-06,
+      "loss": 0.8658,
+      "step": 9597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825444179795588e-06,
+      "loss": 0.8671,
+      "step": 9598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823530661381185e-06,
+      "loss": 0.7931,
+      "step": 9599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821617401850294e-06,
+      "loss": 0.9196,
+      "step": 9600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8197044012986e-06,
+      "loss": 0.8341,
+      "step": 9601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817791659821776e-06,
+      "loss": 0.9286,
+      "step": 9602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815879177515478e-06,
+      "loss": 0.924,
+      "step": 9603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8139669544753585e-06,
+      "loss": 0.9328,
+      "step": 9604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812054990797048e-06,
+      "loss": 0.9067,
+      "step": 9605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8101432865761675e-06,
+      "loss": 0.8856,
+      "step": 9606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808231841908323e-06,
+      "loss": 0.847,
+      "step": 9607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8063206568891055e-06,
+      "loss": 0.9065,
+      "step": 9608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804409731614104e-06,
+      "loss": 0.8836,
+      "step": 9609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8024990661788815e-06,
+      "loss": 0.8481,
+      "step": 9610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800588660678998e-06,
+      "loss": 0.9635,
+      "step": 9611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798678515209991e-06,
+      "loss": 0.8651,
+      "step": 9612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79676862986739e-06,
+      "loss": 0.8913,
+      "step": 9613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794859004746716e-06,
+      "loss": 0.9185,
+      "step": 9614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792949639943471e-06,
+      "loss": 0.8737,
+      "step": 9615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791040535553143e-06,
+      "loss": 0.8803,
+      "step": 9616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789131691671209e-06,
+      "loss": 0.6228,
+      "step": 9617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.787223108393136e-06,
+      "loss": 0.9704,
+      "step": 9618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785314785814376e-06,
+      "loss": 0.8145,
+      "step": 9619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783406724030365e-06,
+      "loss": 0.8114,
+      "step": 9620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781498923136529e-06,
+      "loss": 0.9471,
+      "step": 9621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779591383228276e-06,
+      "loss": 0.9253,
+      "step": 9622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777684104401013e-06,
+      "loss": 0.5867,
+      "step": 9623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7757770867501215e-06,
+      "loss": 0.8525,
+      "step": 9624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773870330370977e-06,
+      "loss": 0.8308,
+      "step": 9625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771963835358936e-06,
+      "loss": 0.8169,
+      "step": 9626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770057601809343e-06,
+      "loss": 0.996,
+      "step": 9627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768151629817536e-06,
+      "loss": 0.9164,
+      "step": 9628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766245919478839e-06,
+      "loss": 0.9023,
+      "step": 9629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.764340470888557e-06,
+      "loss": 0.9545,
+      "step": 9630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.762435284141983e-06,
+      "loss": 0.9254,
+      "step": 9631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7605303593343986e-06,
+      "loss": 0.8837,
+      "step": 9632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.75862569656107e-06,
+      "loss": 0.8495,
+      "step": 9633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.756721295917258e-06,
+      "loss": 0.8698,
+      "step": 9634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.754817157498202e-06,
+      "loss": 0.8927,
+      "step": 9635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.75291328139913e-06,
+      "loss": 0.8602,
+      "step": 9636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.751009667715255e-06,
+      "loss": 0.9154,
+      "step": 9637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.749106316541787e-06,
+      "loss": 0.8808,
+      "step": 9638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.747203227973913e-06,
+      "loss": 0.914,
+      "step": 9639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.745300402106808e-06,
+      "loss": 0.8922,
+      "step": 9640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.743397839035636e-06,
+      "loss": 0.8026,
+      "step": 9641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.741495538855543e-06,
+      "loss": 0.9406,
+      "step": 9642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7395935016616745e-06,
+      "loss": 0.8913,
+      "step": 9643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.73769172754915e-06,
+      "loss": 0.8548,
+      "step": 9644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.73579021661308e-06,
+      "loss": 0.8606,
+      "step": 9645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.733888968948562e-06,
+      "loss": 0.8359,
+      "step": 9646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.731987984650677e-06,
+      "loss": 0.912,
+      "step": 9647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.730087263814506e-06,
+      "loss": 0.77,
+      "step": 9648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7281868065351e-06,
+      "loss": 0.8475,
+      "step": 9649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7262866129075044e-06,
+      "loss": 0.9128,
+      "step": 9650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.724386683026753e-06,
+      "loss": 0.8439,
+      "step": 9651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7224870169878585e-06,
+      "loss": 0.8717,
+      "step": 9652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.720587614885835e-06,
+      "loss": 0.9166,
+      "step": 9653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.71868847681567e-06,
+      "loss": 0.8538,
+      "step": 9654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.716789602872344e-06,
+      "loss": 0.8269,
+      "step": 9655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.71489099315082e-06,
+      "loss": 0.8537,
+      "step": 9656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7129926477460485e-06,
+      "loss": 0.872,
+      "step": 9657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.711094566752977e-06,
+      "loss": 0.8798,
+      "step": 9658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.709196750266522e-06,
+      "loss": 0.8089,
+      "step": 9659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.707299198381607e-06,
+      "loss": 0.8758,
+      "step": 9660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.705401911193124e-06,
+      "loss": 0.8512,
+      "step": 9661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.703504888795958e-06,
+      "loss": 1.0308,
+      "step": 9662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.701608131284989e-06,
+      "loss": 0.811,
+      "step": 9663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.699711638755073e-06,
+      "loss": 0.886,
+      "step": 9664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.697815411301058e-06,
+      "loss": 0.8496,
+      "step": 9665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.695919449017775e-06,
+      "loss": 0.9052,
+      "step": 9666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.694023752000043e-06,
+      "loss": 0.9686,
+      "step": 9667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.692128320342674e-06,
+      "loss": 0.8961,
+      "step": 9668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.690233154140459e-06,
+      "loss": 0.8026,
+      "step": 9669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6883382534881775e-06,
+      "loss": 1.0575,
+      "step": 9670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.686443618480597e-06,
+      "loss": 1.0441,
+      "step": 9671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.684549249212469e-06,
+      "loss": 0.9437,
+      "step": 9672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.682655145778539e-06,
+      "loss": 0.8598,
+      "step": 9673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6807613082735315e-06,
+      "loss": 1.0174,
+      "step": 9674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.67886773679216e-06,
+      "loss": 0.8413,
+      "step": 9675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.676974431429125e-06,
+      "loss": 0.8723,
+      "step": 9676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.675081392279111e-06,
+      "loss": 0.9469,
+      "step": 9677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.673188619436798e-06,
+      "loss": 0.9543,
+      "step": 9678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671296112996843e-06,
+      "loss": 0.9039,
+      "step": 9679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669403873053895e-06,
+      "loss": 0.7834,
+      "step": 9680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.667511899702586e-06,
+      "loss": 0.9819,
+      "step": 9681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665620193037534e-06,
+      "loss": 0.9066,
+      "step": 9682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663728753153354e-06,
+      "loss": 0.7955,
+      "step": 9683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.661837580144634e-06,
+      "loss": 0.8323,
+      "step": 9684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.659946674105958e-06,
+      "loss": 0.9111,
+      "step": 9685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.658056035131885e-06,
+      "loss": 0.8441,
+      "step": 9686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.65616566331698e-06,
+      "loss": 0.9021,
+      "step": 9687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.65427555875578e-06,
+      "loss": 0.8633,
+      "step": 9688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.65238572154281e-06,
+      "loss": 0.9184,
+      "step": 9689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6504961517725846e-06,
+      "loss": 0.8586,
+      "step": 9690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.648606849539601e-06,
+      "loss": 0.8448,
+      "step": 9691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.646717814938349e-06,
+      "loss": 0.8141,
+      "step": 9692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6448290480633076e-06,
+      "loss": 0.9526,
+      "step": 9693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.642940549008931e-06,
+      "loss": 1.0555,
+      "step": 9694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.641052317869667e-06,
+      "loss": 0.9828,
+      "step": 9695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6391643547399515e-06,
+      "loss": 0.8789,
+      "step": 9696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.637276659714197e-06,
+      "loss": 1.0021,
+      "step": 9697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6353892328868185e-06,
+      "loss": 0.8553,
+      "step": 9698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6335020743522085e-06,
+      "loss": 0.6035,
+      "step": 9699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.631615184204743e-06,
+      "loss": 0.8452,
+      "step": 9700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.62972856253879e-06,
+      "loss": 0.8774,
+      "step": 9701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6278422094487e-06,
+      "loss": 0.9906,
+      "step": 9702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.625956125028818e-06,
+      "loss": 0.8909,
+      "step": 9703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.624070309373466e-06,
+      "loss": 0.8142,
+      "step": 9704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.622184762576959e-06,
+      "loss": 0.9061,
+      "step": 9705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.620299484733595e-06,
+      "loss": 0.821,
+      "step": 9706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.618414475937656e-06,
+      "loss": 0.7681,
+      "step": 9707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6165297362834215e-06,
+      "loss": 0.9514,
+      "step": 9708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.614645265865147e-06,
+      "loss": 0.9127,
+      "step": 9709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.612761064777077e-06,
+      "loss": 0.8778,
+      "step": 9710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6108771331134416e-06,
+      "loss": 0.942,
+      "step": 9711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.608993470968465e-06,
+      "loss": 0.8046,
+      "step": 9712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607110078436349e-06,
+      "loss": 0.9447,
+      "step": 9713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.605226955611285e-06,
+      "loss": 0.9202,
+      "step": 9714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.603344102587452e-06,
+      "loss": 0.6125,
+      "step": 9715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.601461519459009e-06,
+      "loss": 0.9854,
+      "step": 9716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.599579206320114e-06,
+      "loss": 0.9517,
+      "step": 9717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.597697163264903e-06,
+      "loss": 0.9798,
+      "step": 9718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.595815390387499e-06,
+      "loss": 0.8116,
+      "step": 9719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.593933887782013e-06,
+      "loss": 0.9201,
+      "step": 9720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592052655542537e-06,
+      "loss": 0.8641,
+      "step": 9721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590171693763159e-06,
+      "loss": 0.9643,
+      "step": 9722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588291002537952e-06,
+      "loss": 0.9868,
+      "step": 9723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586410581960971e-06,
+      "loss": 0.6407,
+      "step": 9724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5845304321262555e-06,
+      "loss": 0.6422,
+      "step": 9725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582650553127836e-06,
+      "loss": 0.9257,
+      "step": 9726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5807709450597245e-06,
+      "loss": 0.9213,
+      "step": 9727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578891608015932e-06,
+      "loss": 0.7892,
+      "step": 9728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577012542090442e-06,
+      "loss": 0.7094,
+      "step": 9729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5751337473772305e-06,
+      "loss": 0.8637,
+      "step": 9730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573255223970254e-06,
+      "loss": 0.7898,
+      "step": 9731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571376971963469e-06,
+      "loss": 0.7197,
+      "step": 9732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569498991450806e-06,
+      "loss": 0.8831,
+      "step": 9733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567621282526185e-06,
+      "loss": 0.9518,
+      "step": 9734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5657438452835145e-06,
+      "loss": 0.9686,
+      "step": 9735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563866679816684e-06,
+      "loss": 0.9668,
+      "step": 9736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56198978621958e-06,
+      "loss": 0.9579,
+      "step": 9737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560113164586068e-06,
+      "loss": 0.8788,
+      "step": 9738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558236815009997e-06,
+      "loss": 0.8875,
+      "step": 9739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556360737585208e-06,
+      "loss": 0.9546,
+      "step": 9740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554484932405524e-06,
+      "loss": 0.8636,
+      "step": 9741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552609399564763e-06,
+      "loss": 0.9129,
+      "step": 9742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55073413915672e-06,
+      "loss": 0.9673,
+      "step": 9743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54885915127518e-06,
+      "loss": 0.8926,
+      "step": 9744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546984436013915e-06,
+      "loss": 0.8063,
+      "step": 9745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545109993466676e-06,
+      "loss": 0.9249,
+      "step": 9746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543235823727218e-06,
+      "loss": 0.8931,
+      "step": 9747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541361926889265e-06,
+      "loss": 0.9501,
+      "step": 9748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539488303046535e-06,
+      "loss": 0.9561,
+      "step": 9749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537614952292729e-06,
+      "loss": 0.8761,
+      "step": 9750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5357418747215345e-06,
+      "loss": 0.7884,
+      "step": 9751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533869070426634e-06,
+      "loss": 0.8313,
+      "step": 9752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531996539501681e-06,
+      "loss": 0.8365,
+      "step": 9753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530124282040332e-06,
+      "loss": 0.9684,
+      "step": 9754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528252298136219e-06,
+      "loss": 0.9116,
+      "step": 9755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526380587882957e-06,
+      "loss": 0.9245,
+      "step": 9756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524509151374163e-06,
+      "loss": 0.9657,
+      "step": 9757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522637988703424e-06,
+      "loss": 0.9096,
+      "step": 9758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520767099964322e-06,
+      "loss": 0.9741,
+      "step": 9759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518896485250423e-06,
+      "loss": 0.8373,
+      "step": 9760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517026144655274e-06,
+      "loss": 0.9395,
+      "step": 9761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515156078272423e-06,
+      "loss": 0.9044,
+      "step": 9762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513286286195391e-06,
+      "loss": 0.8945,
+      "step": 9763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511416768517689e-06,
+      "loss": 0.6132,
+      "step": 9764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509547525332815e-06,
+      "loss": 0.9506,
+      "step": 9765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507678556734248e-06,
+      "loss": 0.8616,
+      "step": 9766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505809862815465e-06,
+      "loss": 0.9564,
+      "step": 9767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503941443669922e-06,
+      "loss": 0.8928,
+      "step": 9768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5020732993910585e-06,
+      "loss": 0.8708,
+      "step": 9769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5002054300723046e-06,
+      "loss": 0.8864,
+      "step": 9770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498337835807071e-06,
+      "loss": 0.888,
+      "step": 9771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496470516688768e-06,
+      "loss": 0.9563,
+      "step": 9772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494603472810778e-06,
+      "loss": 0.7721,
+      "step": 9773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.492736704266475e-06,
+      "loss": 0.8371,
+      "step": 9774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.490870211149219e-06,
+      "loss": 0.8451,
+      "step": 9775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4890039935523545e-06,
+      "loss": 0.9984,
+      "step": 9776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.487138051569219e-06,
+      "loss": 0.8403,
+      "step": 9777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.485272385293129e-06,
+      "loss": 0.8463,
+      "step": 9778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.483406994817388e-06,
+      "loss": 0.9517,
+      "step": 9779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.481541880235286e-06,
+      "loss": 0.9431,
+      "step": 9780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.479677041640105e-06,
+      "loss": 0.8979,
+      "step": 9781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4778124791251065e-06,
+      "loss": 0.5988,
+      "step": 9782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.47594819278354e-06,
+      "loss": 1.0027,
+      "step": 9783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.474084182708641e-06,
+      "loss": 0.9221,
+      "step": 9784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.472220448993628e-06,
+      "loss": 1.109,
+      "step": 9785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.470356991731714e-06,
+      "loss": 0.8495,
+      "step": 9786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4684938110160955e-06,
+      "loss": 0.9415,
+      "step": 9787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.466630906939951e-06,
+      "loss": 0.9577,
+      "step": 9788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4647682795964465e-06,
+      "loss": 0.7949,
+      "step": 9789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.462905929078735e-06,
+      "loss": 0.8348,
+      "step": 9790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.461043855479951e-06,
+      "loss": 0.7824,
+      "step": 9791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.459182058893229e-06,
+      "loss": 0.9392,
+      "step": 9792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.457320539411677e-06,
+      "loss": 0.9586,
+      "step": 9793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.455459297128391e-06,
+      "loss": 0.8704,
+      "step": 9794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.453598332136454e-06,
+      "loss": 0.8855,
+      "step": 9795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.451737644528934e-06,
+      "loss": 0.8474,
+      "step": 9796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.449877234398894e-06,
+      "loss": 0.9104,
+      "step": 9797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.448017101839371e-06,
+      "loss": 0.8374,
+      "step": 9798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.446157246943396e-06,
+      "loss": 0.9575,
+      "step": 9799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.444297669803981e-06,
+      "loss": 0.8274,
+      "step": 9800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.442438370514123e-06,
+      "loss": 0.8281,
+      "step": 9801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.440579349166816e-06,
+      "loss": 0.8402,
+      "step": 9802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.43872060585503e-06,
+      "loss": 0.9239,
+      "step": 9803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.436862140671723e-06,
+      "loss": 0.9602,
+      "step": 9804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.435003953709836e-06,
+      "loss": 0.9007,
+      "step": 9805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.433146045062307e-06,
+      "loss": 0.7973,
+      "step": 9806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.431288414822049e-06,
+      "loss": 0.9305,
+      "step": 9807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.429431063081966e-06,
+      "loss": 0.8931,
+      "step": 9808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4275739899349475e-06,
+      "loss": 0.8732,
+      "step": 9809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.425717195473865e-06,
+      "loss": 0.8125,
+      "step": 9810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.423860679791585e-06,
+      "loss": 0.8636,
+      "step": 9811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.422004442980953e-06,
+      "loss": 0.9404,
+      "step": 9812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.420148485134802e-06,
+      "loss": 0.8392,
+      "step": 9813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.418292806345952e-06,
+      "loss": 0.7553,
+      "step": 9814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.416437406707203e-06,
+      "loss": 0.9204,
+      "step": 9815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.414582286311352e-06,
+      "loss": 0.8444,
+      "step": 9816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.41272744525118e-06,
+      "loss": 0.8716,
+      "step": 9817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.410872883619446e-06,
+      "loss": 0.986,
+      "step": 9818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4090186015089e-06,
+      "loss": 0.9224,
+      "step": 9819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.407164599012278e-06,
+      "loss": 0.8924,
+      "step": 9820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.405310876222297e-06,
+      "loss": 0.9075,
+      "step": 9821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.403457433231674e-06,
+      "loss": 0.8229,
+      "step": 9822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.401604270133096e-06,
+      "loss": 1.0048,
+      "step": 9823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.399751387019245e-06,
+      "loss": 0.9614,
+      "step": 9824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.397898783982783e-06,
+      "loss": 1.0384,
+      "step": 9825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.396046461116368e-06,
+      "loss": 0.8982,
+      "step": 9826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.394194418512633e-06,
+      "loss": 0.6141,
+      "step": 9827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.392342656264203e-06,
+      "loss": 0.7928,
+      "step": 9828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.390491174463689e-06,
+      "loss": 0.9107,
+      "step": 9829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.38863997320368e-06,
+      "loss": 0.8818,
+      "step": 9830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3867890525767665e-06,
+      "loss": 0.8621,
+      "step": 9831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3849384126755115e-06,
+      "loss": 0.7758,
+      "step": 9832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.383088053592469e-06,
+      "loss": 0.9037,
+      "step": 9833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.381237975420178e-06,
+      "loss": 1.0128,
+      "step": 9834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.379388178251161e-06,
+      "loss": 0.8719,
+      "step": 9835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.377538662177937e-06,
+      "loss": 0.879,
+      "step": 9836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.375689427292998e-06,
+      "loss": 0.8804,
+      "step": 9837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.373840473688829e-06,
+      "loss": 0.8863,
+      "step": 9838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.371991801457898e-06,
+      "loss": 0.892,
+      "step": 9839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.370143410692656e-06,
+      "loss": 0.9137,
+      "step": 9840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3682953014855514e-06,
+      "loss": 0.9176,
+      "step": 9841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.366447473929007e-06,
+      "loss": 0.8616,
+      "step": 9842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.364599928115438e-06,
+      "loss": 0.8699,
+      "step": 9843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3627526641372416e-06,
+      "loss": 0.8883,
+      "step": 9844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.360905682086799e-06,
+      "loss": 0.8689,
+      "step": 9845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.359058982056488e-06,
+      "loss": 0.8403,
+      "step": 9846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.35721256413866e-06,
+      "loss": 0.859,
+      "step": 9847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.355366428425655e-06,
+      "loss": 0.798,
+      "step": 9848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.353520575009808e-06,
+      "loss": 0.8531,
+      "step": 9849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3516750039834275e-06,
+      "loss": 0.9975,
+      "step": 9850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3498297154388196e-06,
+      "loss": 0.895,
+      "step": 9851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.347984709468264e-06,
+      "loss": 0.9025,
+      "step": 9852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.346139986164036e-06,
+      "loss": 0.8703,
+      "step": 9853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.344295545618392e-06,
+      "loss": 0.9234,
+      "step": 9854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.342451387923572e-06,
+      "loss": 0.9217,
+      "step": 9855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.340607513171811e-06,
+      "loss": 0.9555,
+      "step": 9856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.338763921455322e-06,
+      "loss": 0.9063,
+      "step": 9857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.336920612866307e-06,
+      "loss": 0.8343,
+      "step": 9858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3350775874969495e-06,
+      "loss": 0.9979,
+      "step": 9859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.333234845439422e-06,
+      "loss": 0.8385,
+      "step": 9860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.331392386785887e-06,
+      "loss": 1.0039,
+      "step": 9861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.329550211628488e-06,
+      "loss": 0.9404,
+      "step": 9862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327708320059354e-06,
+      "loss": 0.8032,
+      "step": 9863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325866712170601e-06,
+      "loss": 0.9169,
+      "step": 9864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324025388054327e-06,
+      "loss": 0.8366,
+      "step": 9865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322184347802627e-06,
+      "loss": 0.9439,
+      "step": 9866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3203435915075695e-06,
+      "loss": 0.9055,
+      "step": 9867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318503119261216e-06,
+      "loss": 0.8549,
+      "step": 9868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316662931155609e-06,
+      "loss": 1.0413,
+      "step": 9869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314823027282778e-06,
+      "loss": 0.8606,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3129834077347465e-06,
+      "loss": 0.9216,
+      "step": 9871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311144072603512e-06,
+      "loss": 0.8438,
+      "step": 9872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309305021981061e-06,
+      "loss": 0.8097,
+      "step": 9873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307466255959367e-06,
+      "loss": 0.8926,
+      "step": 9874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305627774630395e-06,
+      "loss": 0.8945,
+      "step": 9875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3037895780860885e-06,
+      "loss": 0.831,
+      "step": 9876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301951666418377e-06,
+      "loss": 0.8211,
+      "step": 9877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300114039719178e-06,
+      "loss": 0.8471,
+      "step": 9878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29827669808039e-06,
+      "loss": 1.0064,
+      "step": 9879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296439641593906e-06,
+      "loss": 0.8937,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.294602870351602e-06,
+      "loss": 0.9537,
+      "step": 9881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292766384445336e-06,
+      "loss": 0.9437,
+      "step": 9882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290930183966954e-06,
+      "loss": 0.8224,
+      "step": 9883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289094269008285e-06,
+      "loss": 0.9315,
+      "step": 9884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287258639661145e-06,
+      "loss": 0.8164,
+      "step": 9885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285423296017342e-06,
+      "loss": 0.8654,
+      "step": 9886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283588238168662e-06,
+      "loss": 0.8355,
+      "step": 9887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281753466206878e-06,
+      "loss": 0.8457,
+      "step": 9888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2799189802237515e-06,
+      "loss": 0.887,
+      "step": 9889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278084780311024e-06,
+      "loss": 0.9236,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276250866560433e-06,
+      "loss": 0.8913,
+      "step": 9891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2744172390636914e-06,
+      "loss": 0.8826,
+      "step": 9892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272583897912504e-06,
+      "loss": 0.5864,
+      "step": 9893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270750843198556e-06,
+      "loss": 0.8854,
+      "step": 9894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268918075013525e-06,
+      "loss": 0.861,
+      "step": 9895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2670855934490705e-06,
+      "loss": 0.8559,
+      "step": 9896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265253398596836e-06,
+      "loss": 0.9709,
+      "step": 9897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2634214905484525e-06,
+      "loss": 0.8803,
+      "step": 9898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2615898693955335e-06,
+      "loss": 0.7569,
+      "step": 9899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2597585352296885e-06,
+      "loss": 0.8889,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257927488142502e-06,
+      "loss": 0.8882,
+      "step": 9901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2560967282255486e-06,
+      "loss": 0.9621,
+      "step": 9902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254266255570385e-06,
+      "loss": 0.802,
+      "step": 9903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252436070268554e-06,
+      "loss": 0.9053,
+      "step": 9904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250606172411594e-06,
+      "loss": 0.9287,
+      "step": 9905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248776562091016e-06,
+      "loss": 0.8924,
+      "step": 9906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246947239398324e-06,
+      "loss": 0.9249,
+      "step": 9907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245118204425002e-06,
+      "loss": 0.8634,
+      "step": 9908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243289457262521e-06,
+      "loss": 0.8889,
+      "step": 9909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241460998002344e-06,
+      "loss": 0.9987,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239632826735918e-06,
+      "loss": 0.8511,
+      "step": 9911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237804943554669e-06,
+      "loss": 0.8344,
+      "step": 9912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2359773485500126e-06,
+      "loss": 0.6259,
+      "step": 9913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234150041813349e-06,
+      "loss": 0.9654,
+      "step": 9914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232323023436064e-06,
+      "loss": 0.8836,
+      "step": 9915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230496293509534e-06,
+      "loss": 0.8215,
+      "step": 9916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228669852125114e-06,
+      "loss": 0.8888,
+      "step": 9917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226843699374146e-06,
+      "loss": 0.9266,
+      "step": 9918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.225017835347957e-06,
+      "loss": 0.8766,
+      "step": 9919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.223192260137868e-06,
+      "loss": 0.9164,
+      "step": 9920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.221366973835176e-06,
+      "loss": 0.7604,
+      "step": 9921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.219541976531164e-06,
+      "loss": 0.871,
+      "step": 9922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.217717268317106e-06,
+      "loss": 0.9632,
+      "step": 9923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2158928492842545e-06,
+      "loss": 0.8713,
+      "step": 9924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2140687195238574e-06,
+      "loss": 0.9798,
+      "step": 9925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.212244879127139e-06,
+      "loss": 0.982,
+      "step": 9926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.210421328185314e-06,
+      "loss": 0.774,
+      "step": 9927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.208598066789579e-06,
+      "loss": 0.9209,
+      "step": 9928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.206775095031116e-06,
+      "loss": 0.8253,
+      "step": 9929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.204952413001101e-06,
+      "loss": 0.7475,
+      "step": 9930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.203130020790686e-06,
+      "loss": 1.0791,
+      "step": 9931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.201307918491013e-06,
+      "loss": 0.7668,
+      "step": 9932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.199486106193208e-06,
+      "loss": 0.9252,
+      "step": 9933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.197664583988377e-06,
+      "loss": 0.8365,
+      "step": 9934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.195843351967626e-06,
+      "loss": 0.8365,
+      "step": 9935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.194022410222034e-06,
+      "loss": 0.8295,
+      "step": 9936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.192201758842669e-06,
+      "loss": 0.8576,
+      "step": 9937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.190381397920586e-06,
+      "loss": 0.8853,
+      "step": 9938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.188561327546818e-06,
+      "loss": 1.0188,
+      "step": 9939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.186741547812399e-06,
+      "loss": 0.8729,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.184922058808334e-06,
+      "loss": 0.8662,
+      "step": 9941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1831028606256164e-06,
+      "loss": 0.8748,
+      "step": 9942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.181283953355233e-06,
+      "loss": 0.6418,
+      "step": 9943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.179465337088144e-06,
+      "loss": 0.9716,
+      "step": 9944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.177647011915309e-06,
+      "loss": 0.8829,
+      "step": 9945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.17582897792766e-06,
+      "loss": 0.8369,
+      "step": 9946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174011235216121e-06,
+      "loss": 0.9124,
+      "step": 9947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.172193783871601e-06,
+      "loss": 0.8719,
+      "step": 9948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.170376623984988e-06,
+      "loss": 0.8483,
+      "step": 9949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.168559755647169e-06,
+      "loss": 0.8697,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.166743178949006e-06,
+      "loss": 0.9309,
+      "step": 9951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.164926893981348e-06,
+      "loss": 0.9192,
+      "step": 9952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1631109008350315e-06,
+      "loss": 0.8652,
+      "step": 9953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.161295199600871e-06,
+      "loss": 0.8861,
+      "step": 9954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.159479790369682e-06,
+      "loss": 0.9546,
+      "step": 9955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1576646732322514e-06,
+      "loss": 0.7653,
+      "step": 9956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.155849848279357e-06,
+      "loss": 0.8318,
+      "step": 9957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.154035315601761e-06,
+      "loss": 0.8476,
+      "step": 9958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1522210752902055e-06,
+      "loss": 0.8947,
+      "step": 9959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1504071274354345e-06,
+      "loss": 0.8638,
+      "step": 9960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.14859347212816e-06,
+      "loss": 0.9148,
+      "step": 9961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.146780109459086e-06,
+      "loss": 0.9017,
+      "step": 9962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.144967039518902e-06,
+      "loss": 0.7816,
+      "step": 9963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.14315426239828e-06,
+      "loss": 0.8611,
+      "step": 9964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1413417781878865e-06,
+      "loss": 0.8768,
+      "step": 9965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1395295869783615e-06,
+      "loss": 0.9211,
+      "step": 9966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.137717688860337e-06,
+      "loss": 0.8681,
+      "step": 9967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.135906083924426e-06,
+      "loss": 0.899,
+      "step": 9968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.134094772261236e-06,
+      "loss": 0.9576,
+      "step": 9969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.13228375396135e-06,
+      "loss": 0.9145,
+      "step": 9970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1304730291153405e-06,
+      "loss": 0.9771,
+      "step": 9971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.128662597813764e-06,
+      "loss": 0.5911,
+      "step": 9972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.12685246014716e-06,
+      "loss": 0.9614,
+      "step": 9973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.12504261620606e-06,
+      "loss": 0.9986,
+      "step": 9974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.123233066080981e-06,
+      "loss": 0.9631,
+      "step": 9975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.121423809862416e-06,
+      "loss": 0.953,
+      "step": 9976
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.11961484764085e-06,
+      "loss": 0.9492,
+      "step": 9977
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.117806179506755e-06,
+      "loss": 0.8753,
+      "step": 9978
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.115997805550577e-06,
+      "loss": 0.7864,
+      "step": 9979
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.114189725862766e-06,
+      "loss": 0.858,
+      "step": 9980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.112381940533742e-06,
+      "loss": 0.8328,
+      "step": 9981
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.110574449653915e-06,
+      "loss": 0.8342,
+      "step": 9982
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.108767253313683e-06,
+      "loss": 0.8504,
+      "step": 9983
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.106960351603419e-06,
+      "loss": 0.9533,
+      "step": 9984
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1051537446135e-06,
+      "loss": 0.9502,
+      "step": 9985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.103347432434272e-06,
+      "loss": 0.8682,
+      "step": 9986
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.101541415156071e-06,
+      "loss": 0.9882,
+      "step": 9987
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.099735692869218e-06,
+      "loss": 0.8286,
+      "step": 9988
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.097930265664024e-06,
+      "loss": 0.9045,
+      "step": 9989
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0961251336307796e-06,
+      "loss": 0.9211,
+      "step": 9990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.094320296859762e-06,
+      "loss": 0.959,
+      "step": 9991
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.092515755441233e-06,
+      "loss": 1.1235,
+      "step": 9992
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.090711509465437e-06,
+      "loss": 0.8857,
+      "step": 9993
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0889075590226166e-06,
+      "loss": 0.814,
+      "step": 9994
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.087103904202985e-06,
+      "loss": 0.9378,
+      "step": 9995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.085300545096745e-06,
+      "loss": 0.8526,
+      "step": 9996
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.083497481794088e-06,
+      "loss": 0.8396,
+      "step": 9997
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.081694714385182e-06,
+      "loss": 0.9025,
+      "step": 9998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.079892242960195e-06,
+      "loss": 0.7863,
+      "step": 9999
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.078090067609267e-06,
+      "loss": 0.868,
+      "step": 10000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.076288188422528e-06,
+      "loss": 0.9296,
+      "step": 10001
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.074486605490093e-06,
+      "loss": 0.9469,
+      "step": 10002
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.072685318902059e-06,
+      "loss": 0.9373,
+      "step": 10003
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.070884328748513e-06,
+      "loss": 0.8336,
+      "step": 10004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069083635119531e-06,
+      "loss": 0.6669,
+      "step": 10005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067283238105164e-06,
+      "loss": 0.7568,
+      "step": 10006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065483137795454e-06,
+      "loss": 0.9539,
+      "step": 10007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0636833342804234e-06,
+      "loss": 0.8823,
+      "step": 10008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061883827650084e-06,
+      "loss": 0.9085,
+      "step": 10009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060084617994436e-06,
+      "loss": 0.6192,
+      "step": 10010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0582857054034595e-06,
+      "loss": 0.6437,
+      "step": 10011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0564870899671195e-06,
+      "loss": 0.8843,
+      "step": 10012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054688771775365e-06,
+      "loss": 0.8517,
+      "step": 10013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.052890750918138e-06,
+      "loss": 0.8716,
+      "step": 10014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05109302748536e-06,
+      "loss": 0.8247,
+      "step": 10015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049295601566935e-06,
+      "loss": 0.9211,
+      "step": 10016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047498473252756e-06,
+      "loss": 1.031,
+      "step": 10017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045701642632698e-06,
+      "loss": 0.8281,
+      "step": 10018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043905109796629e-06,
+      "loss": 0.8568,
+      "step": 10019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042108874834394e-06,
+      "loss": 0.9141,
+      "step": 10020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040312937835824e-06,
+      "loss": 0.8772,
+      "step": 10021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038517298890738e-06,
+      "loss": 0.9315,
+      "step": 10022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036721958088936e-06,
+      "loss": 0.8407,
+      "step": 10023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034926915520212e-06,
+      "loss": 0.9186,
+      "step": 10024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033132171274334e-06,
+      "loss": 0.9419,
+      "step": 10025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031337725441062e-06,
+      "loss": 0.8581,
+      "step": 10026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0295435781101384e-06,
+      "loss": 0.614,
+      "step": 10027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027749729371288e-06,
+      "loss": 0.8308,
+      "step": 10028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025956179314232e-06,
+      "loss": 0.8786,
+      "step": 10029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024162928028663e-06,
+      "loss": 0.931,
+      "step": 10030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022369975604266e-06,
+      "loss": 0.9735,
+      "step": 10031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020577322130709e-06,
+      "loss": 0.6423,
+      "step": 10032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018784967697642e-06,
+      "loss": 0.9406,
+      "step": 10033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01699291239471e-06,
+      "loss": 0.8343,
+      "step": 10034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015201156311534e-06,
+      "loss": 0.8661,
+      "step": 10035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0134096995377195e-06,
+      "loss": 0.93,
+      "step": 10036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011618542162865e-06,
+      "loss": 0.8135,
+      "step": 10037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009827684276543e-06,
+      "loss": 0.8501,
+      "step": 10038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.008037125968325e-06,
+      "loss": 0.9208,
+      "step": 10039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006246867327757e-06,
+      "loss": 0.8511,
+      "step": 10040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00445690844437e-06,
+      "loss": 0.8491,
+      "step": 10041
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0026672494076845e-06,
+      "loss": 0.6004,
+      "step": 10042
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0008778903071995e-06,
+      "loss": 0.9935,
+      "step": 10043
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9990888312324125e-06,
+      "loss": 0.9622,
+      "step": 10044
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997300072272793e-06,
+      "loss": 0.9475,
+      "step": 10045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9955116135177995e-06,
+      "loss": 0.8282,
+      "step": 10046
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9937234550568745e-06,
+      "loss": 0.8846,
+      "step": 10047
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991935596979446e-06,
+      "loss": 0.9142,
+      "step": 10048
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990148039374931e-06,
+      "loss": 0.6193,
+      "step": 10049
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988360782332729e-06,
+      "loss": 0.8827,
+      "step": 10050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9865738259422194e-06,
+      "loss": 0.978,
+      "step": 10051
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984787170292773e-06,
+      "loss": 0.9372,
+      "step": 10052
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983000815473739e-06,
+      "loss": 0.9199,
+      "step": 10053
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9812147615744634e-06,
+      "loss": 0.949,
+      "step": 10054
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979429008684266e-06,
+      "loss": 0.8679,
+      "step": 10055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977643556892454e-06,
+      "loss": 0.8834,
+      "step": 10056
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975858406288323e-06,
+      "loss": 0.952,
+      "step": 10057
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974073556961147e-06,
+      "loss": 0.8367,
+      "step": 10058
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972289009000194e-06,
+      "loss": 0.8672,
+      "step": 10059
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970504762494711e-06,
+      "loss": 0.8733,
+      "step": 10060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968720817533931e-06,
+      "loss": 0.8562,
+      "step": 10061
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966937174207066e-06,
+      "loss": 0.8663,
+      "step": 10062
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965153832603329e-06,
+      "loss": 0.8687,
+      "step": 10063
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.963370792811902e-06,
+      "loss": 0.75,
+      "step": 10064
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96158805492196e-06,
+      "loss": 0.8403,
+      "step": 10065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.959805619022657e-06,
+      "loss": 0.8855,
+      "step": 10066
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.958023485203136e-06,
+      "loss": 0.89,
+      "step": 10067
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.956241653552525e-06,
+      "loss": 0.9949,
+      "step": 10068
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.95446012415994e-06,
+      "loss": 0.9402,
+      "step": 10069
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.952678897114477e-06,
+      "loss": 0.9216,
+      "step": 10070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.950897972505214e-06,
+      "loss": 0.9864,
+      "step": 10071
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.949117350421222e-06,
+      "loss": 0.7974,
+      "step": 10072
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.947337030951547e-06,
+      "loss": 0.8911,
+      "step": 10073
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.945557014185233e-06,
+      "loss": 0.8921,
+      "step": 10074
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.943777300211299e-06,
+      "loss": 0.95,
+      "step": 10075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.94199788911875e-06,
+      "loss": 0.948,
+      "step": 10076
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9402187809965765e-06,
+      "loss": 0.8194,
+      "step": 10077
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.938439975933752e-06,
+      "loss": 0.9156,
+      "step": 10078
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.936661474019245e-06,
+      "loss": 0.854,
+      "step": 10079
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.934883275341999e-06,
+      "loss": 0.9212,
+      "step": 10080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.933105379990941e-06,
+      "loss": 0.8882,
+      "step": 10081
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.931327788054985e-06,
+      "loss": 0.9193,
+      "step": 10082
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.929550499623038e-06,
+      "loss": 0.6331,
+      "step": 10083
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9277735147839825e-06,
+      "loss": 0.9346,
+      "step": 10084
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.925996833626687e-06,
+      "loss": 0.9141,
+      "step": 10085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.924220456240004e-06,
+      "loss": 0.897,
+      "step": 10086
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.922444382712774e-06,
+      "loss": 0.9653,
+      "step": 10087
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.920668613133827e-06,
+      "loss": 0.8324,
+      "step": 10088
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9188931475919664e-06,
+      "loss": 0.9836,
+      "step": 10089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.917117986175989e-06,
+      "loss": 0.9411,
+      "step": 10090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.91534312897467e-06,
+      "loss": 0.8887,
+      "step": 10091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.913568576076773e-06,
+      "loss": 0.914,
+      "step": 10092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.91179432757105e-06,
+      "loss": 0.9379,
+      "step": 10093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9100203835462335e-06,
+      "loss": 0.8624,
+      "step": 10094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.90824674409104e-06,
+      "loss": 0.9013,
+      "step": 10095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9064734092941705e-06,
+      "loss": 0.8537,
+      "step": 10096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.90470037924431e-06,
+      "loss": 0.8882,
+      "step": 10097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.902927654030135e-06,
+      "loss": 0.8724,
+      "step": 10098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.901155233740306e-06,
+      "loss": 0.8414,
+      "step": 10099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.899383118463459e-06,
+      "loss": 0.8463,
+      "step": 10100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.897611308288221e-06,
+      "loss": 0.8521,
+      "step": 10101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.895839803303205e-06,
+      "loss": 0.6242,
+      "step": 10102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.894068603597001e-06,
+      "loss": 0.8474,
+      "step": 10103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.892297709258197e-06,
+      "loss": 0.8324,
+      "step": 10104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.890527120375356e-06,
+      "loss": 0.8229,
+      "step": 10105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.888756837037026e-06,
+      "loss": 0.9395,
+      "step": 10106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8869868593317405e-06,
+      "loss": 0.8838,
+      "step": 10107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.885217187348025e-06,
+      "loss": 1.0428,
+      "step": 10108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.883447821174378e-06,
+      "loss": 0.7023,
+      "step": 10109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.881678760899291e-06,
+      "loss": 0.9614,
+      "step": 10110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.879910006611238e-06,
+      "loss": 0.8031,
+      "step": 10111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.87814155839867e-06,
+      "loss": 0.7919,
+      "step": 10112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8763734163500414e-06,
+      "loss": 0.8821,
+      "step": 10113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.874605580553773e-06,
+      "loss": 0.7461,
+      "step": 10114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.87283805109828e-06,
+      "loss": 0.9021,
+      "step": 10115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.871070828071957e-06,
+      "loss": 0.8689,
+      "step": 10116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.869303911563181e-06,
+      "loss": 0.8812,
+      "step": 10117
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8675373016603294e-06,
+      "loss": 0.9259,
+      "step": 10118
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.865770998451747e-06,
+      "loss": 0.8705,
+      "step": 10119
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.86400500202577e-06,
+      "loss": 0.8877,
+      "step": 10120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.862239312470718e-06,
+      "loss": 0.8477,
+      "step": 10121
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.860473929874894e-06,
+      "loss": 0.8822,
+      "step": 10122
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.858708854326594e-06,
+      "loss": 0.9208,
+      "step": 10123
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8569440859140875e-06,
+      "loss": 0.889,
+      "step": 10124
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.855179624725634e-06,
+      "loss": 0.886,
+      "step": 10125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.853415470849479e-06,
+      "loss": 0.9712,
+      "step": 10126
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8516516243738445e-06,
+      "loss": 0.8533,
+      "step": 10127
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.849888085386951e-06,
+      "loss": 0.8099,
+      "step": 10128
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.848124853976993e-06,
+      "loss": 0.9308,
+      "step": 10129
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.84636193023215e-06,
+      "loss": 0.8908,
+      "step": 10130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.844599314240593e-06,
+      "loss": 0.9305,
+      "step": 10131
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.842837006090469e-06,
+      "loss": 0.8995,
+      "step": 10132
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8410750058699175e-06,
+      "loss": 0.8846,
+      "step": 10133
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.839313313667059e-06,
+      "loss": 0.9016,
+      "step": 10134
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.837551929569997e-06,
+      "loss": 1.0838,
+      "step": 10135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8357908536668194e-06,
+      "loss": 0.9452,
+      "step": 10136
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8340300860456e-06,
+      "loss": 0.936,
+      "step": 10137
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832269626794404e-06,
+      "loss": 0.8347,
+      "step": 10138
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83050947600127e-06,
+      "loss": 0.9504,
+      "step": 10139
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.828749633754226e-06,
+      "loss": 0.8745,
+      "step": 10140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826990100141286e-06,
+      "loss": 0.9345,
+      "step": 10141
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825230875250441e-06,
+      "loss": 0.9942,
+      "step": 10142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823471959169682e-06,
+      "loss": 0.8058,
+      "step": 10143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821713351986972e-06,
+      "loss": 0.9689,
+      "step": 10144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81995505379026e-06,
+      "loss": 0.9141,
+      "step": 10145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818197064667482e-06,
+      "loss": 0.932,
+      "step": 10146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816439384706554e-06,
+      "loss": 0.8239,
+      "step": 10147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8146820139953878e-06,
+      "loss": 0.8243,
+      "step": 10148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8129249526218694e-06,
+      "loss": 0.8723,
+      "step": 10149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8111682006738713e-06,
+      "loss": 0.8571,
+      "step": 10150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8094117582392474e-06,
+      "loss": 0.9421,
+      "step": 10151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8076556254058493e-06,
+      "loss": 0.8297,
+      "step": 10152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8058998022615e-06,
+      "loss": 0.9607,
+      "step": 10153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8041442888940092e-06,
+      "loss": 0.8321,
+      "step": 10154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8023890853911737e-06,
+      "loss": 0.8901,
+      "step": 10155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800634191840772e-06,
+      "loss": 0.8854,
+      "step": 10156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7988796083305744e-06,
+      "loss": 0.8887,
+      "step": 10157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797125334948328e-06,
+      "loss": 0.8194,
+      "step": 10158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7953713717817662e-06,
+      "loss": 0.9109,
+      "step": 10159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7936177189186073e-06,
+      "loss": 0.9148,
+      "step": 10160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7918643764465524e-06,
+      "loss": 0.7856,
+      "step": 10161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79011134445329e-06,
+      "loss": 0.9052,
+      "step": 10162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.788358623026498e-06,
+      "loss": 0.9457,
+      "step": 10163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7866062122538274e-06,
+      "loss": 0.9703,
+      "step": 10164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7848541122229187e-06,
+      "loss": 0.9195,
+      "step": 10165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7831023230214e-06,
+      "loss": 0.9481,
+      "step": 10166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7813508447368754e-06,
+      "loss": 0.8951,
+      "step": 10167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7795996774569452e-06,
+      "loss": 0.8956,
+      "step": 10168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7778488212691875e-06,
+      "loss": 0.9407,
+      "step": 10169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7760982762611642e-06,
+      "loss": 0.8589,
+      "step": 10170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7743480425204217e-06,
+      "loss": 0.907,
+      "step": 10171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.77259812013449e-06,
+      "loss": 0.9732,
+      "step": 10172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770848509190891e-06,
+      "loss": 0.8021,
+      "step": 10173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7690992097771227e-06,
+      "loss": 0.9283,
+      "step": 10174
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7673502219806713e-06,
+      "loss": 0.858,
+      "step": 10175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7656015458890017e-06,
+      "loss": 0.6206,
+      "step": 10176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7638531815895764e-06,
+      "loss": 0.8854,
+      "step": 10177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7621051291698286e-06,
+      "loss": 0.71,
+      "step": 10178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760357388717182e-06,
+      "loss": 0.7873,
+      "step": 10179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.758609960319045e-06,
+      "loss": 0.8784,
+      "step": 10180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756862844062804e-06,
+      "loss": 0.8265,
+      "step": 10181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7551160400358435e-06,
+      "loss": 0.9117,
+      "step": 10182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7533695483255194e-06,
+      "loss": 0.8621,
+      "step": 10183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7516233690191774e-06,
+      "loss": 0.8754,
+      "step": 10184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7498775022041457e-06,
+      "loss": 1.023,
+      "step": 10185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748131947967735e-06,
+      "loss": 0.925,
+      "step": 10186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7463867063972505e-06,
+      "loss": 0.8653,
+      "step": 10187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7446417775799706e-06,
+      "loss": 0.8602,
+      "step": 10188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742897161603163e-06,
+      "loss": 0.8595,
+      "step": 10189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.741152858554077e-06,
+      "loss": 0.7533,
+      "step": 10190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739408868519946e-06,
+      "loss": 0.8255,
+      "step": 10191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7376651915879923e-06,
+      "loss": 0.9558,
+      "step": 10192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7359218278454233e-06,
+      "loss": 0.8745,
+      "step": 10193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7341787773794248e-06,
+      "loss": 0.8834,
+      "step": 10194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.73243604027717e-06,
+      "loss": 0.8944,
+      "step": 10195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7306936166258136e-06,
+      "loss": 0.8622,
+      "step": 10196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728951506512496e-06,
+      "loss": 0.9532,
+      "step": 10197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7272097100243487e-06,
+      "loss": 0.9954,
+      "step": 10198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7254682272484786e-06,
+      "loss": 0.8938,
+      "step": 10199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72372705827198e-06,
+      "loss": 0.9095,
+      "step": 10200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7219862031819276e-06,
+      "loss": 0.9096,
+      "step": 10201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7202456620653916e-06,
+      "loss": 0.84,
+      "step": 10202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718505435009415e-06,
+      "loss": 0.9607,
+      "step": 10203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7167655221010313e-06,
+      "loss": 0.8043,
+      "step": 10204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715025923427256e-06,
+      "loss": 0.8871,
+      "step": 10205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713286639075083e-06,
+      "loss": 0.8154,
+      "step": 10206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7115476691315065e-06,
+      "loss": 0.9477,
+      "step": 10207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.709809013683491e-06,
+      "loss": 0.5996,
+      "step": 10208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7080706728179895e-06,
+      "loss": 0.8544,
+      "step": 10209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7063326466219386e-06,
+      "loss": 0.8306,
+      "step": 10210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7045949351822565e-06,
+      "loss": 0.8698,
+      "step": 10211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7028575385858555e-06,
+      "loss": 0.7244,
+      "step": 10212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7011204569196225e-06,
+      "loss": 0.7895,
+      "step": 10213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6993836902704327e-06,
+      "loss": 0.8851,
+      "step": 10214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.697647238725143e-06,
+      "loss": 0.9382,
+      "step": 10215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.695911102370594e-06,
+      "loss": 0.7613,
+      "step": 10216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.694175281293618e-06,
+      "loss": 0.9017,
+      "step": 10217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6924397755810248e-06,
+      "loss": 0.8345,
+      "step": 10218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.690704585319609e-06,
+      "loss": 0.9677,
+      "step": 10219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6889697105961488e-06,
+      "loss": 0.9458,
+      "step": 10220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.687235151497406e-06,
+      "loss": 0.8784,
+      "step": 10221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6855009081101355e-06,
+      "loss": 0.9225,
+      "step": 10222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6837669805210653e-06,
+      "loss": 0.9781,
+      "step": 10223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6820333688169095e-06,
+      "loss": 0.9303,
+      "step": 10224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6803000730843762e-06,
+      "loss": 0.9472,
+      "step": 10225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6785670934101416e-06,
+      "loss": 0.9286,
+      "step": 10226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6768344298808825e-06,
+      "loss": 0.8875,
+      "step": 10227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6751020825832494e-06,
+      "loss": 0.7833,
+      "step": 10228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6733700516038783e-06,
+      "loss": 0.9717,
+      "step": 10229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.671638337029393e-06,
+      "loss": 0.6401,
+      "step": 10230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6699069389463926e-06,
+      "loss": 0.8858,
+      "step": 10231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6681758574414772e-06,
+      "loss": 1.0454,
+      "step": 10232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.666445092601216e-06,
+      "loss": 0.9794,
+      "step": 10233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6647146445121682e-06,
+      "loss": 0.8107,
+      "step": 10234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6629845132608755e-06,
+      "loss": 0.7307,
+      "step": 10235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6612546989338604e-06,
+      "loss": 0.9645,
+      "step": 10236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6595252016176417e-06,
+      "loss": 0.9238,
+      "step": 10237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6577960213987105e-06,
+      "loss": 0.8896,
+      "step": 10238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6560671583635467e-06,
+      "loss": 0.6397,
+      "step": 10239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6543386125986124e-06,
+      "loss": 0.9247,
+      "step": 10240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6526103841903515e-06,
+      "loss": 0.9112,
+      "step": 10241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6508824732252035e-06,
+      "loss": 0.9007,
+      "step": 10242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.649154879789579e-06,
+      "loss": 0.8704,
+      "step": 10243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6474276039698784e-06,
+      "loss": 0.9361,
+      "step": 10244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6457006458524825e-06,
+      "loss": 0.9544,
+      "step": 10245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.643974005523766e-06,
+      "loss": 0.8791,
+      "step": 10246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.642247683070077e-06,
+      "loss": 0.8481,
+      "step": 10247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.640521678577752e-06,
+      "loss": 0.8074,
+      "step": 10248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6387959921331117e-06,
+      "loss": 0.8239,
+      "step": 10249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.637070623822455e-06,
+      "loss": 0.8769,
+      "step": 10250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6353455737320798e-06,
+      "loss": 0.8378,
+      "step": 10251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6336208419482546e-06,
+      "loss": 0.8983,
+      "step": 10252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6318964285572343e-06,
+      "loss": 1.0176,
+      "step": 10253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.630172333645261e-06,
+      "loss": 0.8499,
+      "step": 10254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6284485572985564e-06,
+      "loss": 0.8065,
+      "step": 10255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.626725099603331e-06,
+      "loss": 0.6227,
+      "step": 10256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.625001960645783e-06,
+      "loss": 0.8786,
+      "step": 10257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.623279140512085e-06,
+      "loss": 0.5838,
+      "step": 10258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6215566392883983e-06,
+      "loss": 0.896,
+      "step": 10259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.619834457060867e-06,
+      "loss": 0.9002,
+      "step": 10260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.618112593915618e-06,
+      "loss": 0.8775,
+      "step": 10261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.61639104993877e-06,
+      "loss": 0.8435,
+      "step": 10262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.614669825216419e-06,
+      "loss": 0.9644,
+      "step": 10263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6129489198346437e-06,
+      "loss": 0.7909,
+      "step": 10264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6112283338795107e-06,
+      "loss": 0.8097,
+      "step": 10265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6095080674370653e-06,
+      "loss": 0.8759,
+      "step": 10266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6077881205933475e-06,
+      "loss": 0.8342,
+      "step": 10267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6060684934343717e-06,
+      "loss": 0.7908,
+      "step": 10268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.604349186046139e-06,
+      "loss": 0.8059,
+      "step": 10269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.602630198514632e-06,
+      "loss": 0.9097,
+      "step": 10270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6009115309258246e-06,
+      "loss": 0.872,
+      "step": 10271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.59919318336567e-06,
+      "loss": 0.9054,
+      "step": 10272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5974751559201027e-06,
+      "loss": 0.958,
+      "step": 10273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5957574486750448e-06,
+      "loss": 0.9527,
+      "step": 10274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5940400617163995e-06,
+      "loss": 0.8176,
+      "step": 10275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.592322995130061e-06,
+      "loss": 0.789,
+      "step": 10276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5906062490019e-06,
+      "loss": 0.8717,
+      "step": 10277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5888898234177737e-06,
+      "loss": 0.8631,
+      "step": 10278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5871737184635234e-06,
+      "loss": 0.9654,
+      "step": 10279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5854579342249706e-06,
+      "loss": 0.7811,
+      "step": 10280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.583742470787931e-06,
+      "loss": 0.8976,
+      "step": 10281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5820273282381943e-06,
+      "loss": 0.6387,
+      "step": 10282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5803125066615383e-06,
+      "loss": 0.828,
+      "step": 10283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5785980061437242e-06,
+      "loss": 0.859,
+      "step": 10284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.576883826770492e-06,
+      "loss": 0.8692,
+      "step": 10285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5751699686275786e-06,
+      "loss": 0.8673,
+      "step": 10286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5734564318006894e-06,
+      "loss": 0.8313,
+      "step": 10287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.571743216375527e-06,
+      "loss": 0.8722,
+      "step": 10288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570030322437771e-06,
+      "loss": 0.9662,
+      "step": 10289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.568317750073085e-06,
+      "loss": 0.9741,
+      "step": 10290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.566605499367113e-06,
+      "loss": 0.8191,
+      "step": 10291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5648935704054943e-06,
+      "loss": 0.6233,
+      "step": 10292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5631819632738427e-06,
+      "loss": 0.8713,
+      "step": 10293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.561470678057759e-06,
+      "loss": 0.879,
+      "step": 10294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559759714842823e-06,
+      "loss": 0.892,
+      "step": 10295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5580490737146088e-06,
+      "loss": 0.9133,
+      "step": 10296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5563387547586657e-06,
+      "loss": 0.961,
+      "step": 10297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55462875806053e-06,
+      "loss": 0.9228,
+      "step": 10298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5529190837057214e-06,
+      "loss": 0.9304,
+      "step": 10299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551209731779739e-06,
+      "loss": 1.0095,
+      "step": 10300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.549500702368077e-06,
+      "loss": 1.013,
+      "step": 10301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5477919955562034e-06,
+      "loss": 0.888,
+      "step": 10302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546083611429575e-06,
+      "loss": 0.8905,
+      "step": 10303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5443755500736287e-06,
+      "loss": 0.9249,
+      "step": 10304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5426678115737854e-06,
+      "loss": 0.8898,
+      "step": 10305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5409603960154583e-06,
+      "loss": 0.855,
+      "step": 10306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.539253303484034e-06,
+      "loss": 0.8857,
+      "step": 10307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5375465340648874e-06,
+      "loss": 0.8828,
+      "step": 10308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5358400878433762e-06,
+      "loss": 0.8923,
+      "step": 10309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53413396490484e-06,
+      "loss": 0.8925,
+      "step": 10310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5324281653346117e-06,
+      "loss": 0.914,
+      "step": 10311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5307226892179957e-06,
+      "loss": 0.9059,
+      "step": 10312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529017536640288e-06,
+      "loss": 0.7847,
+      "step": 10313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527312707686765e-06,
+      "loss": 0.8161,
+      "step": 10314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525608202442684e-06,
+      "loss": 0.9506,
+      "step": 10315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5239040209932986e-06,
+      "loss": 0.9482,
+      "step": 10316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5222001634238313e-06,
+      "loss": 0.8503,
+      "step": 10317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520496629819494e-06,
+      "loss": 0.8689,
+      "step": 10318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5187934202654884e-06,
+      "loss": 0.9489,
+      "step": 10319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5170905348469897e-06,
+      "loss": 0.8047,
+      "step": 10320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5153879736491658e-06,
+      "loss": 0.8764,
+      "step": 10321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5136857367571643e-06,
+      "loss": 0.8849,
+      "step": 10322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5119838242561146e-06,
+      "loss": 0.8823,
+      "step": 10323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510282236231133e-06,
+      "loss": 0.8192,
+      "step": 10324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5085809727673146e-06,
+      "loss": 0.8975,
+      "step": 10325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50688003394975e-06,
+      "loss": 0.9418,
+      "step": 10326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5051794198635026e-06,
+      "loss": 0.9306,
+      "step": 10327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503479130593621e-06,
+      "loss": 0.8928,
+      "step": 10328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501779166225141e-06,
+      "loss": 0.8595,
+      "step": 10329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5000795268430775e-06,
+      "loss": 0.8626,
+      "step": 10330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4983802125324373e-06,
+      "loss": 0.9212,
+      "step": 10331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4966812233782045e-06,
+      "loss": 0.864,
+      "step": 10332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4949825594653463e-06,
+      "loss": 0.8371,
+      "step": 10333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4932842208788166e-06,
+      "loss": 0.785,
+      "step": 10334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491586207703548e-06,
+      "loss": 0.7864,
+      "step": 10335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489888520024469e-06,
+      "loss": 0.8904,
+      "step": 10336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488191157926478e-06,
+      "loss": 0.9119,
+      "step": 10337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486494121494466e-06,
+      "loss": 0.8654,
+      "step": 10338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484797410813299e-06,
+      "loss": 0.7837,
+      "step": 10339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483101025967839e-06,
+      "loss": 0.9641,
+      "step": 10340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4814049670429206e-06,
+      "loss": 0.8344,
+      "step": 10341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4797092341233697e-06,
+      "loss": 0.8737,
+      "step": 10342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478013827293989e-06,
+      "loss": 0.9904,
+      "step": 10343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4763187466395676e-06,
+      "loss": 0.8518,
+      "step": 10344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4746239922448844e-06,
+      "loss": 0.9404,
+      "step": 10345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4729295641946947e-06,
+      "loss": 0.8174,
+      "step": 10346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4712354625737386e-06,
+      "loss": 0.9038,
+      "step": 10347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469541687466741e-06,
+      "loss": 0.8634,
+      "step": 10348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467848238958408e-06,
+      "loss": 0.9635,
+      "step": 10349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4661551171334328e-06,
+      "loss": 0.937,
+      "step": 10350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4644623220764963e-06,
+      "loss": 0.8855,
+      "step": 10351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462769853872254e-06,
+      "loss": 0.848,
+      "step": 10352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4610777126053497e-06,
+      "loss": 0.8933,
+      "step": 10353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.459385898360409e-06,
+      "loss": 0.6534,
+      "step": 10354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4576944112220388e-06,
+      "loss": 1.01,
+      "step": 10355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.456003251274841e-06,
+      "loss": 0.8341,
+      "step": 10356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.454312418603388e-06,
+      "loss": 0.7889,
+      "step": 10357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4526219132922424e-06,
+      "loss": 0.9727,
+      "step": 10358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.45093173542595e-06,
+      "loss": 0.9427,
+      "step": 10359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.449241885089033e-06,
+      "loss": 0.9065,
+      "step": 10360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4475523623660135e-06,
+      "loss": 0.8191,
+      "step": 10361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4458631673413825e-06,
+      "loss": 0.618,
+      "step": 10362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4441743000996174e-06,
+      "loss": 0.8383,
+      "step": 10363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4424857607251814e-06,
+      "loss": 1.002,
+      "step": 10364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4407975493025246e-06,
+      "loss": 0.881,
+      "step": 10365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.439109665916076e-06,
+      "loss": 0.9792,
+      "step": 10366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4374221106502493e-06,
+      "loss": 0.9596,
+      "step": 10367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.43573488358944e-06,
+      "loss": 0.8827,
+      "step": 10368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4340479848180276e-06,
+      "loss": 0.9509,
+      "step": 10369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.432361414420381e-06,
+      "loss": 0.8636,
+      "step": 10370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4306751724808472e-06,
+      "loss": 0.8851,
+      "step": 10371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.428989259083757e-06,
+      "loss": 0.7861,
+      "step": 10372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4273036743134257e-06,
+      "loss": 0.9543,
+      "step": 10373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4256184182541484e-06,
+      "loss": 0.8247,
+      "step": 10374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4239334909902155e-06,
+      "loss": 0.7871,
+      "step": 10375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4222488926058885e-06,
+      "loss": 0.9053,
+      "step": 10376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.420564623185415e-06,
+      "loss": 0.8939,
+      "step": 10377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4188806828130318e-06,
+      "loss": 0.8096,
+      "step": 10378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4171970715729498e-06,
+      "loss": 0.8723,
+      "step": 10379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.415513789549375e-06,
+      "loss": 0.9406,
+      "step": 10380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.413830836826486e-06,
+      "loss": 0.8736,
+      "step": 10381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4121482134884574e-06,
+      "loss": 0.8803,
+      "step": 10382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4104659196194347e-06,
+      "loss": 0.896,
+      "step": 10383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.408783955303552e-06,
+      "loss": 0.9149,
+      "step": 10384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4071023206249243e-06,
+      "loss": 0.8405,
+      "step": 10385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4054210156676603e-06,
+      "loss": 0.9707,
+      "step": 10386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4037400405158406e-06,
+      "loss": 0.901,
+      "step": 10387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.402059395253534e-06,
+      "loss": 0.7701,
+      "step": 10388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.400379079964787e-06,
+      "loss": 0.8825,
+      "step": 10389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.398699094733645e-06,
+      "loss": 0.8388,
+      "step": 10390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3970194396441204e-06,
+      "loss": 0.8842,
+      "step": 10391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.395340114780217e-06,
+      "loss": 0.9175,
+      "step": 10392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3936611202259183e-06,
+      "loss": 0.8524,
+      "step": 10393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3919824560651938e-06,
+      "loss": 0.856,
+      "step": 10394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.390304122382e-06,
+      "loss": 0.8723,
+      "step": 10395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.388626119260271e-06,
+      "loss": 0.8796,
+      "step": 10396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3869484467839264e-06,
+      "loss": 0.8549,
+      "step": 10397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.385271105036868e-06,
+      "loss": 0.8046,
+      "step": 10398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.383594094102981e-06,
+      "loss": 0.7628,
+      "step": 10399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.38191741406614e-06,
+      "loss": 0.931,
+      "step": 10400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.380241065010197e-06,
+      "loss": 0.809,
+      "step": 10401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3785650470189868e-06,
+      "loss": 0.9779,
+      "step": 10402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.376889360176332e-06,
+      "loss": 0.9167,
+      "step": 10403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.375214004566032e-06,
+      "loss": 0.8916,
+      "step": 10404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.373538980271881e-06,
+      "loss": 0.9314,
+      "step": 10405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3718642873776454e-06,
+      "loss": 0.8236,
+      "step": 10406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.37018992596708e-06,
+      "loss": 0.8415,
+      "step": 10407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.368515896123922e-06,
+      "loss": 0.9313,
+      "step": 10408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3668421979318897e-06,
+      "loss": 0.9093,
+      "step": 10409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3651688314746935e-06,
+      "loss": 0.8741,
+      "step": 10410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.363495796836017e-06,
+      "loss": 0.7993,
+      "step": 10411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.36182309409953e-06,
+      "loss": 0.9651,
+      "step": 10412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3601507233488917e-06,
+      "loss": 0.8462,
+      "step": 10413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.358478684667734e-06,
+      "loss": 0.8828,
+      "step": 10414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.356806978139686e-06,
+      "loss": 0.9072,
+      "step": 10415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3551356038483475e-06,
+      "loss": 0.8886,
+      "step": 10416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3534645618773074e-06,
+      "loss": 0.8147,
+      "step": 10417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3517938523101377e-06,
+      "loss": 0.9304,
+      "step": 10418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.350123475230389e-06,
+      "loss": 0.972,
+      "step": 10419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3484534307216065e-06,
+      "loss": 0.8879,
+      "step": 10420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.346783718867308e-06,
+      "loss": 0.9196,
+      "step": 10421
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.345114339751e-06,
+      "loss": 0.8891,
+      "step": 10422
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.343445293456169e-06,
+      "loss": 0.8258,
+      "step": 10423
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.341776580066284e-06,
+      "loss": 0.8179,
+      "step": 10424
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3401081996648076e-06,
+      "loss": 0.9217,
+      "step": 10425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3384401523351727e-06,
+      "loss": 1.0285,
+      "step": 10426
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3367724381608034e-06,
+      "loss": 0.9345,
+      "step": 10427
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3351050572251033e-06,
+      "loss": 0.9218,
+      "step": 10428
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.333438009611458e-06,
+      "loss": 0.8622,
+      "step": 10429
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3317712954032455e-06,
+      "loss": 0.8538,
+      "step": 10430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330104914683818e-06,
+      "loss": 0.9121,
+      "step": 10431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3284388675365143e-06,
+      "loss": 0.8916,
+      "step": 10432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3267731540446512e-06,
+      "loss": 0.7983,
+      "step": 10433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3251077742915417e-06,
+      "loss": 0.8886,
+      "step": 10434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.32344272836047e-06,
+      "loss": 0.8478,
+      "step": 10435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321778016334708e-06,
+      "loss": 0.8172,
+      "step": 10436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3201136382975096e-06,
+      "loss": 0.952,
+      "step": 10437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3184495943321127e-06,
+      "loss": 0.9104,
+      "step": 10438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3167858845217416e-06,
+      "loss": 0.9008,
+      "step": 10439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3151225089496007e-06,
+      "loss": 0.8609,
+      "step": 10440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3134594676988773e-06,
+      "loss": 0.8627,
+      "step": 10441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3117967608527414e-06,
+      "loss": 0.8422,
+      "step": 10442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310134388494346e-06,
+      "loss": 0.8619,
+      "step": 10443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3084723507068317e-06,
+      "loss": 0.8998,
+      "step": 10444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306810647573323e-06,
+      "loss": 0.7964,
+      "step": 10445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.305149279176921e-06,
+      "loss": 0.943,
+      "step": 10446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3034882456007135e-06,
+      "loss": 0.9261,
+      "step": 10447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3018275469277705e-06,
+      "loss": 0.8701,
+      "step": 10448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300167183241144e-06,
+      "loss": 0.8586,
+      "step": 10449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2985071546238776e-06,
+      "loss": 0.9817,
+      "step": 10450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2968474611589894e-06,
+      "loss": 0.9222,
+      "step": 10451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295188102929483e-06,
+      "loss": 0.8571,
+      "step": 10452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.293529080018345e-06,
+      "loss": 0.9129,
+      "step": 10453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.291870392508543e-06,
+      "loss": 0.8892,
+      "step": 10454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290212040483036e-06,
+      "loss": 1.0065,
+      "step": 10455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2885540240247594e-06,
+      "loss": 0.6045,
+      "step": 10456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2868963432166323e-06,
+      "loss": 0.8648,
+      "step": 10457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2852389981415554e-06,
+      "loss": 0.8621,
+      "step": 10458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2835819888824204e-06,
+      "loss": 0.9536,
+      "step": 10459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281925315522094e-06,
+      "loss": 0.6898,
+      "step": 10460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2802689781434303e-06,
+      "loss": 0.8989,
+      "step": 10461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278612976829265e-06,
+      "loss": 0.8565,
+      "step": 10462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276957311662413e-06,
+      "loss": 0.8594,
+      "step": 10463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275301982725685e-06,
+      "loss": 0.8902,
+      "step": 10464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273646990101862e-06,
+      "loss": 0.8964,
+      "step": 10465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2719923338737126e-06,
+      "loss": 0.908,
+      "step": 10466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2703380141239903e-06,
+      "loss": 0.7846,
+      "step": 10467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268684030935425e-06,
+      "loss": 0.8469,
+      "step": 10468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2670303843907434e-06,
+      "loss": 0.9249,
+      "step": 10469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.265377074572643e-06,
+      "loss": 0.8463,
+      "step": 10470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2637241015638087e-06,
+      "loss": 0.9452,
+      "step": 10471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2620714654469088e-06,
+      "loss": 0.8026,
+      "step": 10472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2604191663045893e-06,
+      "loss": 0.9139,
+      "step": 10473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258767204219493e-06,
+      "loss": 0.8269,
+      "step": 10474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257115579274228e-06,
+      "loss": 0.8402,
+      "step": 10475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2554642915514046e-06,
+      "loss": 0.9696,
+      "step": 10476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253813341133599e-06,
+      "loss": 0.8402,
+      "step": 10477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252162728103382e-06,
+      "loss": 0.9149,
+      "step": 10478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2505124525432987e-06,
+      "loss": 0.8682,
+      "step": 10479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2488625145358865e-06,
+      "loss": 0.902,
+      "step": 10480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2472129141636612e-06,
+      "loss": 0.8571,
+      "step": 10481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.245563651509119e-06,
+      "loss": 0.8344,
+      "step": 10482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2439147266547423e-06,
+      "loss": 0.7583,
+      "step": 10483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2422661396830003e-06,
+      "loss": 0.8289,
+      "step": 10484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2406178906763387e-06,
+      "loss": 0.8817,
+      "step": 10485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2389699797171903e-06,
+      "loss": 0.9564,
+      "step": 10486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2373224068879684e-06,
+      "loss": 0.8712,
+      "step": 10487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235675172271068e-06,
+      "loss": 1.0127,
+      "step": 10488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234028275948876e-06,
+      "loss": 0.8646,
+      "step": 10489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232381718003754e-06,
+      "loss": 0.9123,
+      "step": 10490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2307354985180484e-06,
+      "loss": 0.7661,
+      "step": 10491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2290896175740893e-06,
+      "loss": 0.8493,
+      "step": 10492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.227444075254187e-06,
+      "loss": 0.9382,
+      "step": 10493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225798871640644e-06,
+      "loss": 0.8886,
+      "step": 10494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2241540068157353e-06,
+      "loss": 0.8981,
+      "step": 10495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2225094808617253e-06,
+      "loss": 0.9116,
+      "step": 10496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2208652938608575e-06,
+      "loss": 0.7824,
+      "step": 10497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2192214458953577e-06,
+      "loss": 0.8336,
+      "step": 10498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2175779370474434e-06,
+      "loss": 1.052,
+      "step": 10499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.215934767399308e-06,
+      "loss": 0.8227,
+      "step": 10500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.214291937033126e-06,
+      "loss": 0.9957,
+      "step": 10501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2126494460310564e-06,
+      "loss": 0.8582,
+      "step": 10502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2110072944752493e-06,
+      "loss": 0.6235,
+      "step": 10503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.209365482447828e-06,
+      "loss": 0.9715,
+      "step": 10504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2077240100309013e-06,
+      "loss": 0.8506,
+      "step": 10505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.20608287730656e-06,
+      "loss": 0.8787,
+      "step": 10506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.204442084356886e-06,
+      "loss": 0.8888,
+      "step": 10507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2028016312639297e-06,
+      "loss": 0.952,
+      "step": 10508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.201161518109741e-06,
+      "loss": 0.5933,
+      "step": 10509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1995217449763416e-06,
+      "loss": 0.8402,
+      "step": 10510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.197882311945738e-06,
+      "loss": 0.613,
+      "step": 10511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1962432190999217e-06,
+      "loss": 0.9116,
+      "step": 10512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.194604466520862e-06,
+      "loss": 0.8394,
+      "step": 10513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.192966054290524e-06,
+      "loss": 0.9208,
+      "step": 10514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.191327982490843e-06,
+      "loss": 0.8432,
+      "step": 10515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.18969025120374e-06,
+      "loss": 0.7941,
+      "step": 10516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1880528605111227e-06,
+      "loss": 0.8758,
+      "step": 10517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.186415810494876e-06,
+      "loss": 0.904,
+      "step": 10518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1847791012368777e-06,
+      "loss": 0.817,
+      "step": 10519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.183142732818979e-06,
+      "loss": 0.7992,
+      "step": 10520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1815067053230177e-06,
+      "loss": 0.9562,
+      "step": 10521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.179871018830812e-06,
+      "loss": 0.8451,
+      "step": 10522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.178235673424165e-06,
+      "loss": 0.9264,
+      "step": 10523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.176600669184867e-06,
+      "loss": 0.9304,
+      "step": 10524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.174966006194686e-06,
+      "loss": 0.9083,
+      "step": 10525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1733316845353734e-06,
+      "loss": 0.9056,
+      "step": 10526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1716977042886597e-06,
+      "loss": 0.836,
+      "step": 10527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1700640655362713e-06,
+      "loss": 0.8234,
+      "step": 10528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1684307683599045e-06,
+      "loss": 0.8301,
+      "step": 10529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1667978128412435e-06,
+      "loss": 0.8606,
+      "step": 10530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.165165199061955e-06,
+      "loss": 0.7763,
+      "step": 10531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1635329271036853e-06,
+      "loss": 0.9898,
+      "step": 10532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.161900997048073e-06,
+      "loss": 0.7915,
+      "step": 10533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1602694089767315e-06,
+      "loss": 0.9927,
+      "step": 10534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1586381629712582e-06,
+      "loss": 0.9771,
+      "step": 10535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.157007259113235e-06,
+      "loss": 0.9489,
+      "step": 10536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1553766974842216e-06,
+      "loss": 0.8427,
+      "step": 10537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.153746478165769e-06,
+      "loss": 0.8774,
+      "step": 10538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.152116601239411e-06,
+      "loss": 0.8758,
+      "step": 10539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.150487066786656e-06,
+      "loss": 0.9335,
+      "step": 10540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.148857874889001e-06,
+      "loss": 1.0183,
+      "step": 10541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1472290256279227e-06,
+      "loss": 0.6259,
+      "step": 10542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1456005190848794e-06,
+      "loss": 0.8993,
+      "step": 10543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1439723553413226e-06,
+      "loss": 0.9162,
+      "step": 10544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1423445344786765e-06,
+      "loss": 0.845,
+      "step": 10545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1407170565783506e-06,
+      "loss": 0.6175,
+      "step": 10546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1390899217217376e-06,
+      "loss": 1.0536,
+      "step": 10547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.137463129990209e-06,
+      "loss": 0.9567,
+      "step": 10548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.135836681465131e-06,
+      "loss": 0.8963,
+      "step": 10549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.134210576227842e-06,
+      "loss": 0.8479,
+      "step": 10550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.132584814359665e-06,
+      "loss": 1.0409,
+      "step": 10551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1309593959419027e-06,
+      "loss": 0.8623,
+      "step": 10552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1293343210558548e-06,
+      "loss": 0.7757,
+      "step": 10553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.127709589782787e-06,
+      "loss": 0.8083,
+      "step": 10554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1260852022039566e-06,
+      "loss": 0.836,
+      "step": 10555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1244611584006024e-06,
+      "loss": 0.8787,
+      "step": 10556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.12283745845394e-06,
+      "loss": 0.9209,
+      "step": 10557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1212141024451827e-06,
+      "loss": 0.939,
+      "step": 10558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1195910904555104e-06,
+      "loss": 0.8397,
+      "step": 10559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1179684225660955e-06,
+      "loss": 0.8932,
+      "step": 10560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1163460988580883e-06,
+      "loss": 1.0546,
+      "step": 10561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1147241194126223e-06,
+      "loss": 0.8275,
+      "step": 10562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1131024843108204e-06,
+      "loss": 0.7731,
+      "step": 10563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1114811936337797e-06,
+      "loss": 0.9221,
+      "step": 10564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.109860247462585e-06,
+      "loss": 0.8859,
+      "step": 10565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1082396458783006e-06,
+      "loss": 0.9267,
+      "step": 10566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1066193889619733e-06,
+      "loss": 0.8604,
+      "step": 10567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.104999476794641e-06,
+      "loss": 0.8917,
+      "step": 10568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1033799094573123e-06,
+      "loss": 0.8253,
+      "step": 10569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1017606870309903e-06,
+      "loss": 0.8551,
+      "step": 10570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1001418095966506e-06,
+      "loss": 0.7536,
+      "step": 10571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0985232772352567e-06,
+      "loss": 0.8355,
+      "step": 10572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096905090027752e-06,
+      "loss": 0.8562,
+      "step": 10573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0952872480550687e-06,
+      "loss": 0.8413,
+      "step": 10574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0936697513981163e-06,
+      "loss": 0.871,
+      "step": 10575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.092052600137788e-06,
+      "loss": 0.8737,
+      "step": 10576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0904357943549556e-06,
+      "loss": 0.8523,
+      "step": 10577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.088819334130487e-06,
+      "loss": 0.9038,
+      "step": 10578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0872032195452184e-06,
+      "loss": 0.82,
+      "step": 10579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0855874506799753e-06,
+      "loss": 0.9742,
+      "step": 10580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083972027615566e-06,
+      "loss": 1.0328,
+      "step": 10581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0823569504327744e-06,
+      "loss": 0.8231,
+      "step": 10582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0807422192123824e-06,
+      "loss": 0.8692,
+      "step": 10583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0791278340351407e-06,
+      "loss": 0.9147,
+      "step": 10584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077513794981787e-06,
+      "loss": 0.8795,
+      "step": 10585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0759001021330424e-06,
+      "loss": 0.973,
+      "step": 10586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0742867555696066e-06,
+      "loss": 0.5914,
+      "step": 10587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.072673755372173e-06,
+      "loss": 0.5801,
+      "step": 10588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0710611016214066e-06,
+      "loss": 0.8901,
+      "step": 10589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069448794397959e-06,
+      "loss": 0.9315,
+      "step": 10590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.067836833782464e-06,
+      "loss": 0.9503,
+      "step": 10591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0662252198555355e-06,
+      "loss": 0.8317,
+      "step": 10592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064613952697778e-06,
+      "loss": 0.9297,
+      "step": 10593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063003032389772e-06,
+      "loss": 0.9428,
+      "step": 10594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0613924590120815e-06,
+      "loss": 0.9016,
+      "step": 10595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0597822326452496e-06,
+      "loss": 0.8675,
+      "step": 10596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0581723533698148e-06,
+      "loss": 0.8996,
+      "step": 10597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0565628212662858e-06,
+      "loss": 0.9049,
+      "step": 10598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.054953636415158e-06,
+      "loss": 0.8544,
+      "step": 10599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0533447988969047e-06,
+      "loss": 0.8796,
+      "step": 10600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.051736308791994e-06,
+      "loss": 0.9773,
+      "step": 10601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0501281661808624e-06,
+      "loss": 0.8527,
+      "step": 10602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0485203711439427e-06,
+      "loss": 0.871,
+      "step": 10603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046912923761639e-06,
+      "loss": 0.7774,
+      "step": 10604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045305824114344e-06,
+      "loss": 0.9328,
+      "step": 10605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.043699072282429e-06,
+      "loss": 0.8433,
+      "step": 10606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0420926683462494e-06,
+      "loss": 1.0277,
+      "step": 10607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0404866123861476e-06,
+      "loss": 0.8091,
+      "step": 10608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0388809044824454e-06,
+      "loss": 0.9492,
+      "step": 10609
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0372755447154433e-06,
+      "loss": 0.9306,
+      "step": 10610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0356705331654304e-06,
+      "loss": 0.9327,
+      "step": 10611
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0340658699126726e-06,
+      "loss": 0.6089,
+      "step": 10612
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032461555037426e-06,
+      "loss": 0.9761,
+      "step": 10613
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.030857588619923e-06,
+      "loss": 0.5628,
+      "step": 10614
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0292539707403822e-06,
+      "loss": 0.7968,
+      "step": 10615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027650701479e-06,
+      "loss": 0.8676,
+      "step": 10616
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0260477809159573e-06,
+      "loss": 0.8676,
+      "step": 10617
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0244452091314246e-06,
+      "loss": 0.8661,
+      "step": 10618
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022842986205545e-06,
+      "loss": 0.9502,
+      "step": 10619
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0212411122184503e-06,
+      "loss": 0.8562,
+      "step": 10620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0196395872502483e-06,
+      "loss": 0.8487,
+      "step": 10621
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01803841138104e-06,
+      "loss": 0.8027,
+      "step": 10622
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0164375846909e-06,
+      "loss": 0.8814,
+      "step": 10623
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014837107259887e-06,
+      "loss": 0.8924,
+      "step": 10624
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013236979168046e-06,
+      "loss": 0.8811,
+      "step": 10625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0116372004953965e-06,
+      "loss": 0.967,
+      "step": 10626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010037771321953e-06,
+      "loss": 0.8917,
+      "step": 10627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008438691727703e-06,
+      "loss": 0.8824,
+      "step": 10628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006839961792618e-06,
+      "loss": 0.9471,
+      "step": 10629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.005241581596654e-06,
+      "loss": 0.9225,
+      "step": 10630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003643551219745e-06,
+      "loss": 0.6052,
+      "step": 10631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002045870741813e-06,
+      "loss": 0.5789,
+      "step": 10632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0004485402427664e-06,
+      "loss": 0.9429,
+      "step": 10633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998851559802485e-06,
+      "loss": 0.8913,
+      "step": 10634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997254929500837e-06,
+      "loss": 0.9102,
+      "step": 10635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995658649417672e-06,
+      "loss": 0.8595,
+      "step": 10636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99406271963282e-06,
+      "loss": 0.863,
+      "step": 10637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992467140226102e-06,
+      "loss": 0.9553,
+      "step": 10638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990871911277312e-06,
+      "loss": 1.0512,
+      "step": 10639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.98927703286623e-06,
+      "loss": 0.7838,
+      "step": 10640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.98768250507262e-06,
+      "loss": 0.9026,
+      "step": 10641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.986088327976221e-06,
+      "loss": 0.8518,
+      "step": 10642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9844945016567684e-06,
+      "loss": 0.8805,
+      "step": 10643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9829010261939697e-06,
+      "loss": 0.9793,
+      "step": 10644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9813079016675152e-06,
+      "loss": 0.8925,
+      "step": 10645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.979715128157077e-06,
+      "loss": 0.845,
+      "step": 10646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.978122705742319e-06,
+      "loss": 0.9546,
+      "step": 10647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9765306345028777e-06,
+      "loss": 0.7838,
+      "step": 10648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.974938914518375e-06,
+      "loss": 0.9168,
+      "step": 10649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9733475458684146e-06,
+      "loss": 0.9198,
+      "step": 10650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.971756528632581e-06,
+      "loss": 0.7746,
+      "step": 10651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9701658628904493e-06,
+      "loss": 0.8993,
+      "step": 10652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9685755487215697e-06,
+      "loss": 0.8601,
+      "step": 10653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.966985586205475e-06,
+      "loss": 0.8658,
+      "step": 10654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.965395975421681e-06,
+      "loss": 0.8651,
+      "step": 10655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9638067164496853e-06,
+      "loss": 0.8124,
+      "step": 10656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9622178093689747e-06,
+      "loss": 0.952,
+      "step": 10657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.960629254259009e-06,
+      "loss": 0.9219,
+      "step": 10658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.959041051199236e-06,
+      "loss": 0.8246,
+      "step": 10659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.957453200269084e-06,
+      "loss": 0.93,
+      "step": 10660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9558657015479587e-06,
+      "loss": 0.9188,
+      "step": 10661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.954278555115263e-06,
+      "loss": 0.8399,
+      "step": 10662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9526917610503635e-06,
+      "loss": 0.8006,
+      "step": 10663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9511053194326267e-06,
+      "loss": 0.6522,
+      "step": 10664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.949519230341389e-06,
+      "loss": 0.8755,
+      "step": 10665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9479334938559734e-06,
+      "loss": 0.9792,
+      "step": 10666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9463481100556813e-06,
+      "loss": 1.0061,
+      "step": 10667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.944763079019808e-06,
+      "loss": 0.8604,
+      "step": 10668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9431784008276186e-06,
+      "loss": 0.9705,
+      "step": 10669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.941594075558366e-06,
+      "loss": 0.973,
+      "step": 10670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.940010103291283e-06,
+      "loss": 0.8387,
+      "step": 10671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9384264841055907e-06,
+      "loss": 0.8496,
+      "step": 10672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9368432180804875e-06,
+      "loss": 0.8884,
+      "step": 10673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9352603052951533e-06,
+      "loss": 0.9583,
+      "step": 10674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9336777458287523e-06,
+      "loss": 0.8528,
+      "step": 10675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9320955397604288e-06,
+      "loss": 0.888,
+      "step": 10676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9305136871693176e-06,
+      "loss": 0.8912,
+      "step": 10677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.865,
+      "step": 10678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.927351042735147e-06,
+      "loss": 0.8537,
+      "step": 10679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.925770251050256e-06,
+      "loss": 0.8105,
+      "step": 10680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9241898131589086e-06,
+      "loss": 0.8712,
+      "step": 10681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.922609729140151e-06,
+      "loss": 0.5657,
+      "step": 10682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.921029999073003e-06,
+      "loss": 0.8768,
+      "step": 10683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9194506230364685e-06,
+      "loss": 0.6174,
+      "step": 10684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.917871601109535e-06,
+      "loss": 0.9626,
+      "step": 10685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.91629293337117e-06,
+      "loss": 0.8515,
+      "step": 10686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.914714619900331e-06,
+      "loss": 0.7628,
+      "step": 10687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.913136660775946e-06,
+      "loss": 0.7672,
+      "step": 10688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9115590560769346e-06,
+      "loss": 0.921,
+      "step": 10689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9099818058821916e-06,
+      "loss": 0.8845,
+      "step": 10690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9084049102706035e-06,
+      "loss": 0.6469,
+      "step": 10691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.90682836932103e-06,
+      "loss": 0.8531,
+      "step": 10692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.905252183112316e-06,
+      "loss": 0.872,
+      "step": 10693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.903676351723288e-06,
+      "loss": 0.7966,
+      "step": 10694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9021008752327604e-06,
+      "loss": 0.9288,
+      "step": 10695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.90052575371952e-06,
+      "loss": 0.8856,
+      "step": 10696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8989509872623455e-06,
+      "loss": 0.9096,
+      "step": 10697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.897376575939993e-06,
+      "loss": 0.936,
+      "step": 10698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895802519831199e-06,
+      "loss": 0.8872,
+      "step": 10699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894228819014686e-06,
+      "loss": 0.8832,
+      "step": 10700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.892655473569155e-06,
+      "loss": 0.8096,
+      "step": 10701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8910824835732954e-06,
+      "loss": 0.8721,
+      "step": 10702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8895098491057737e-06,
+      "loss": 0.9482,
+      "step": 10703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8879375702452385e-06,
+      "loss": 0.8989,
+      "step": 10704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.886365647070323e-06,
+      "loss": 0.8436,
+      "step": 10705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8847940796596376e-06,
+      "loss": 0.9053,
+      "step": 10706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8832228680917863e-06,
+      "loss": 0.957,
+      "step": 10707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.881652012445344e-06,
+      "loss": 0.5981,
+      "step": 10708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.880081512798871e-06,
+      "loss": 0.9013,
+      "step": 10709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.878511369230913e-06,
+      "loss": 0.8343,
+      "step": 10710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8769415818199896e-06,
+      "loss": 0.9384,
+      "step": 10711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8753721506446162e-06,
+      "loss": 0.8931,
+      "step": 10712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8738030757832778e-06,
+      "loss": 0.8359,
+      "step": 10713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8722343573144485e-06,
+      "loss": 0.9533,
+      "step": 10714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8706659953165785e-06,
+      "loss": 0.7499,
+      "step": 10715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8690979898681092e-06,
+      "loss": 0.8551,
+      "step": 10716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8675303410474566e-06,
+      "loss": 0.8825,
+      "step": 10717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8659630489330216e-06,
+      "loss": 0.9685,
+      "step": 10718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8643961136031872e-06,
+      "loss": 0.7658,
+      "step": 10719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8628295351363156e-06,
+      "loss": 0.8068,
+      "step": 10720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8612633136107593e-06,
+      "loss": 0.9024,
+      "step": 10721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8596974491048435e-06,
+      "loss": 0.9615,
+      "step": 10722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.858131941696881e-06,
+      "loss": 0.9002,
+      "step": 10723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856566791465166e-06,
+      "loss": 0.9082,
+      "step": 10724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8550019984879694e-06,
+      "loss": 0.8659,
+      "step": 10725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853437562843553e-06,
+      "loss": 0.8338,
+      "step": 10726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851873484610159e-06,
+      "loss": 0.9147,
+      "step": 10727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850309763866008e-06,
+      "loss": 0.9353,
+      "step": 10728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848746400689303e-06,
+      "loss": 0.9114,
+      "step": 10729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8471833951582296e-06,
+      "loss": 0.944,
+      "step": 10730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8456207473509544e-06,
+      "loss": 0.7946,
+      "step": 10731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844058457345633e-06,
+      "loss": 0.7632,
+      "step": 10732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842496525220396e-06,
+      "loss": 0.8673,
+      "step": 10733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.840934951053358e-06,
+      "loss": 0.9091,
+      "step": 10734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.839373734922616e-06,
+      "loss": 0.6257,
+      "step": 10735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8378128769062442e-06,
+      "loss": 0.9747,
+      "step": 10736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8362523770823115e-06,
+      "loss": 0.8807,
+      "step": 10737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8346922355288573e-06,
+      "loss": 0.883,
+      "step": 10738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8331324523239067e-06,
+      "loss": 0.8307,
+      "step": 10739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8315730275454645e-06,
+      "loss": 0.7817,
+      "step": 10740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830013961271526e-06,
+      "loss": 0.8536,
+      "step": 10741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8284552535800582e-06,
+      "loss": 0.9421,
+      "step": 10742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8268969045490156e-06,
+      "loss": 0.8631,
+      "step": 10743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825338914256335e-06,
+      "loss": 0.9529,
+      "step": 10744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823781282779928e-06,
+      "loss": 1.0109,
+      "step": 10745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822224010197703e-06,
+      "loss": 0.99,
+      "step": 10746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8206670965875372e-06,
+      "loss": 0.8998,
+      "step": 10747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8191105420272945e-06,
+      "loss": 0.9255,
+      "step": 10748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8175543465948205e-06,
+      "loss": 0.8738,
+      "step": 10749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815998510367941e-06,
+      "loss": 0.7462,
+      "step": 10750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.814443033424472e-06,
+      "loss": 0.9887,
+      "step": 10751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8128879158422007e-06,
+      "loss": 0.8432,
+      "step": 10752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8113331576989034e-06,
+      "loss": 0.8964,
+      "step": 10753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8097787590723335e-06,
+      "loss": 0.9365,
+      "step": 10754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.808224720040227e-06,
+      "loss": 0.9931,
+      "step": 10755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80667104068031e-06,
+      "loss": 0.6382,
+      "step": 10756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8051177210702794e-06,
+      "loss": 0.7639,
+      "step": 10757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8035647612878236e-06,
+      "loss": 0.9245,
+      "step": 10758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8020121614106064e-06,
+      "loss": 0.8096,
+      "step": 10759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8004599215162766e-06,
+      "loss": 0.908,
+      "step": 10760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.798908041682459e-06,
+      "loss": 1.0698,
+      "step": 10761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7973565219867737e-06,
+      "loss": 0.7672,
+      "step": 10762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7958053625068105e-06,
+      "loss": 0.9419,
+      "step": 10763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7942545633201455e-06,
+      "loss": 0.9062,
+      "step": 10764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792704124504334e-06,
+      "loss": 0.9678,
+      "step": 10765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7911540461369225e-06,
+      "loss": 0.9036,
+      "step": 10766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.789604328295429e-06,
+      "loss": 0.7691,
+      "step": 10767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7880549710573567e-06,
+      "loss": 0.8783,
+      "step": 10768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786505974500193e-06,
+      "loss": 0.8872,
+      "step": 10769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7849573387014018e-06,
+      "loss": 0.9216,
+      "step": 10770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7834090637384394e-06,
+      "loss": 0.8063,
+      "step": 10771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.781861149688734e-06,
+      "loss": 0.7365,
+      "step": 10772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.780313596629699e-06,
+      "loss": 0.941,
+      "step": 10773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7787664046387307e-06,
+      "loss": 0.8768,
+      "step": 10774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777219573793204e-06,
+      "loss": 0.8363,
+      "step": 10775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7756731041704834e-06,
+      "loss": 0.9167,
+      "step": 10776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.774126995847908e-06,
+      "loss": 0.9062,
+      "step": 10777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7725812489028004e-06,
+      "loss": 0.8618,
+      "step": 10778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.771035863412467e-06,
+      "loss": 0.8302,
+      "step": 10779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76949083945419e-06,
+      "loss": 0.7886,
+      "step": 10780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7679461771052463e-06,
+      "loss": 0.9337,
+      "step": 10781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7664018764428834e-06,
+      "loss": 0.9026,
+      "step": 10782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7648579375443353e-06,
+      "loss": 0.8371,
+      "step": 10783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7633143604868117e-06,
+      "loss": 0.8905,
+      "step": 10784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761771145347517e-06,
+      "loss": 0.9914,
+      "step": 10785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760228292203627e-06,
+      "loss": 0.8387,
+      "step": 10786
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7586858011323015e-06,
+      "loss": 0.7602,
+      "step": 10787
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7571436722106794e-06,
+      "loss": 0.7722,
+      "step": 10788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.755601905515892e-06,
+      "loss": 0.9965,
+      "step": 10789
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.754060501125039e-06,
+      "loss": 0.8513,
+      "step": 10790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.752519459115215e-06,
+      "loss": 0.9307,
+      "step": 10791
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7509787795634868e-06,
+      "loss": 0.798,
+      "step": 10792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.749438462546906e-06,
+      "loss": 0.9569,
+      "step": 10793
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.747898508142507e-06,
+      "loss": 0.832,
+      "step": 10794
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7463589164273007e-06,
+      "loss": 0.9729,
+      "step": 10795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7448196874782927e-06,
+      "loss": 0.8184,
+      "step": 10796
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.743280821372458e-06,
+      "loss": 0.8454,
+      "step": 10797
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.741742318186759e-06,
+      "loss": 0.8749,
+      "step": 10798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.740204177998136e-06,
+      "loss": 0.8685,
+      "step": 10799
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.738666400883514e-06,
+      "loss": 0.9097,
+      "step": 10800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7371289869198036e-06,
+      "loss": 0.9058,
+      "step": 10801
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7355919361838914e-06,
+      "loss": 0.839,
+      "step": 10802
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7340552487526473e-06,
+      "loss": 0.8606,
+      "step": 10803
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.732518924702924e-06,
+      "loss": 0.8786,
+      "step": 10804
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7309829641115515e-06,
+      "loss": 0.89,
+      "step": 10805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7294473670553524e-06,
+      "loss": 0.8716,
+      "step": 10806
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7279121336111216e-06,
+      "loss": 0.7969,
+      "step": 10807
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7263772638556376e-06,
+      "loss": 0.8024,
+      "step": 10808
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.72484275786566e-06,
+      "loss": 0.8694,
+      "step": 10809
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7233086157179365e-06,
+      "loss": 0.9754,
+      "step": 10810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7217748374891908e-06,
+      "loss": 0.8873,
+      "step": 10811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.720241423256128e-06,
+      "loss": 0.833,
+      "step": 10812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7187083730954366e-06,
+      "loss": 0.9709,
+      "step": 10813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.717175687083785e-06,
+      "loss": 0.8764,
+      "step": 10814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7156433652978307e-06,
+      "loss": 0.8854,
+      "step": 10815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.714111407814205e-06,
+      "loss": 0.9852,
+      "step": 10816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7125798147095217e-06,
+      "loss": 0.9154,
+      "step": 10817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.711048586060381e-06,
+      "loss": 0.9158,
+      "step": 10818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.709517721943358e-06,
+      "loss": 0.8753,
+      "step": 10819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7079872224350147e-06,
+      "loss": 0.9767,
+      "step": 10820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7064570876119e-06,
+      "loss": 0.8556,
+      "step": 10821
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.704927317550533e-06,
+      "loss": 0.7597,
+      "step": 10822
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7033979123274213e-06,
+      "loss": 0.8275,
+      "step": 10823
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7018688720190524e-06,
+      "loss": 0.99,
+      "step": 10824
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.700340196701893e-06,
+      "loss": 0.936,
+      "step": 10825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6988118864524003e-06,
+      "loss": 0.5923,
+      "step": 10826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6972839413470053e-06,
+      "loss": 0.9227,
+      "step": 10827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.695756361462123e-06,
+      "loss": 0.811,
+      "step": 10828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.69422914687415e-06,
+      "loss": 0.8257,
+      "step": 10829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6927022976594607e-06,
+      "loss": 0.8262,
+      "step": 10830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6911758138944223e-06,
+      "loss": 0.8861,
+      "step": 10831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.689649695655373e-06,
+      "loss": 0.8166,
+      "step": 10832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.688123943018638e-06,
+      "loss": 0.854,
+      "step": 10833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6865985560605168e-06,
+      "loss": 0.8891,
+      "step": 10834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.685073534857305e-06,
+      "loss": 0.739,
+      "step": 10835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6835488794852683e-06,
+      "loss": 0.8799,
+      "step": 10836
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6820245900206553e-06,
+      "loss": 0.9783,
+      "step": 10837
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6805006665396994e-06,
+      "loss": 0.8666,
+      "step": 10838
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6789771091186113e-06,
+      "loss": 0.9336,
+      "step": 10839
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6774539178335933e-06,
+      "loss": 0.9323,
+      "step": 10840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6759310927608175e-06,
+      "loss": 0.8917,
+      "step": 10841
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6744086339764452e-06,
+      "loss": 0.8855,
+      "step": 10842
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.672886541556615e-06,
+      "loss": 0.7513,
+      "step": 10843
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6713648155774473e-06,
+      "loss": 0.8738,
+      "step": 10844
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.669843456115052e-06,
+      "loss": 0.944,
+      "step": 10845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6683224632455127e-06,
+      "loss": 0.9611,
+      "step": 10846
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6668018370448947e-06,
+      "loss": 0.8517,
+      "step": 10847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.665281577589247e-06,
+      "loss": 0.8671,
+      "step": 10848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6637616849545987e-06,
+      "loss": 0.6259,
+      "step": 10849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6622421592169677e-06,
+      "loss": 0.5975,
+      "step": 10850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.660723000452341e-06,
+      "loss": 0.906,
+      "step": 10851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.659204208736702e-06,
+      "loss": 0.7968,
+      "step": 10852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6576857841460023e-06,
+      "loss": 0.8735,
+      "step": 10853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6561677267561824e-06,
+      "loss": 0.8264,
+      "step": 10854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6546500366431604e-06,
+      "loss": 0.8611,
+      "step": 10855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653132713882842e-06,
+      "loss": 0.8191,
+      "step": 10856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6516157585511105e-06,
+      "loss": 0.969,
+      "step": 10857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65009917072383e-06,
+      "loss": 0.8477,
+      "step": 10858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6485829504768445e-06,
+      "loss": 0.9425,
+      "step": 10859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6470670978859893e-06,
+      "loss": 0.8988,
+      "step": 10860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645551613027071e-06,
+      "loss": 0.7086,
+      "step": 10861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6440364959758814e-06,
+      "loss": 0.8876,
+      "step": 10862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642521746808193e-06,
+      "loss": 0.7471,
+      "step": 10863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6410073655997605e-06,
+      "loss": 0.9129,
+      "step": 10864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6394933524263235e-06,
+      "loss": 0.8282,
+      "step": 10865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6379797073635992e-06,
+      "loss": 0.9098,
+      "step": 10866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6364664304872867e-06,
+      "loss": 0.8595,
+      "step": 10867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6349535218730682e-06,
+      "loss": 0.9312,
+      "step": 10868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633440981596602e-06,
+      "loss": 0.7142,
+      "step": 10869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6319288097335406e-06,
+      "loss": 0.884,
+      "step": 10870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6304170063595057e-06,
+      "loss": 0.6932,
+      "step": 10871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6289055715501055e-06,
+      "loss": 0.8321,
+      "step": 10872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.62739450538093e-06,
+      "loss": 0.8423,
+      "step": 10873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625883807927545e-06,
+      "loss": 0.8122,
+      "step": 10874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6243734792655116e-06,
+      "loss": 0.8832,
+      "step": 10875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622863519470359e-06,
+      "loss": 0.8759,
+      "step": 10876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.621353928617603e-06,
+      "loss": 0.9075,
+      "step": 10877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619844706782737e-06,
+      "loss": 0.8661,
+      "step": 10878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618335854041246e-06,
+      "loss": 0.848,
+      "step": 10879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616827370468589e-06,
+      "loss": 0.9501,
+      "step": 10880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6153192561402053e-06,
+      "loss": 0.895,
+      "step": 10881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6138115111315165e-06,
+      "loss": 0.8509,
+      "step": 10882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6123041355179325e-06,
+      "loss": 0.9275,
+      "step": 10883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6107971293748334e-06,
+      "loss": 0.9881,
+      "step": 10884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.609290492777594e-06,
+      "loss": 0.8857,
+      "step": 10885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6077842258015597e-06,
+      "loss": 0.791,
+      "step": 10886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6062783285220616e-06,
+      "loss": 0.9891,
+      "step": 10887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6047728010144126e-06,
+      "loss": 0.8556,
+      "step": 10888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603267643353903e-06,
+      "loss": 0.5836,
+      "step": 10889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6017628556158136e-06,
+      "loss": 0.8146,
+      "step": 10890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600258437875399e-06,
+      "loss": 0.8416,
+      "step": 10891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598754390207897e-06,
+      "loss": 0.9226,
+      "step": 10892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5972507126885294e-06,
+      "loss": 0.8212,
+      "step": 10893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5957474053924912e-06,
+      "loss": 0.9159,
+      "step": 10894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594244468394974e-06,
+      "loss": 0.8336,
+      "step": 10895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5927419017711376e-06,
+      "loss": 0.9307,
+      "step": 10896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591239705596128e-06,
+      "loss": 0.8968,
+      "step": 10897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.589737879945072e-06,
+      "loss": 0.8769,
+      "step": 10898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588236424893077e-06,
+      "loss": 0.9746,
+      "step": 10899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5867353405152375e-06,
+      "loss": 1.019,
+      "step": 10900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5852346268866236e-06,
+      "loss": 0.8214,
+      "step": 10901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5837342840822865e-06,
+      "loss": 0.8747,
+      "step": 10902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5822343121772586e-06,
+      "loss": 0.9778,
+      "step": 10903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580734711246562e-06,
+      "loss": 0.9007,
+      "step": 10904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579235481365192e-06,
+      "loss": 0.881,
+      "step": 10905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5777366226081256e-06,
+      "loss": 0.8564,
+      "step": 10906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576238135050325e-06,
+      "loss": 0.9064,
+      "step": 10907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574740018766728e-06,
+      "loss": 0.8483,
+      "step": 10908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5732422738322638e-06,
+      "loss": 0.8076,
+      "step": 10909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.571744900321834e-06,
+      "loss": 0.962,
+      "step": 10910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5702478983103252e-06,
+      "loss": 0.8344,
+      "step": 10911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5687512678726047e-06,
+      "loss": 0.8388,
+      "step": 10912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567255009083518e-06,
+      "loss": 0.8276,
+      "step": 10913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565759122017898e-06,
+      "loss": 0.8506,
+      "step": 10914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5642636067505598e-06,
+      "loss": 0.9542,
+      "step": 10915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5627684633562946e-06,
+      "loss": 0.951,
+      "step": 10916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5612736919098746e-06,
+      "loss": 0.8063,
+      "step": 10917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5597792924860587e-06,
+      "loss": 0.7988,
+      "step": 10918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.558285265159578e-06,
+      "loss": 0.8816,
+      "step": 10919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5567916100051594e-06,
+      "loss": 1.0214,
+      "step": 10920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5552983270974997e-06,
+      "loss": 1.041,
+      "step": 10921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5538054165112803e-06,
+      "loss": 0.7808,
+      "step": 10922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5523128783211627e-06,
+      "loss": 0.8951,
+      "step": 10923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55082071260179e-06,
+      "loss": 0.7838,
+      "step": 10924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5493289194277926e-06,
+      "loss": 0.8799,
+      "step": 10925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5478374988737753e-06,
+      "loss": 0.9268,
+      "step": 10926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5463464510143256e-06,
+      "loss": 0.925,
+      "step": 10927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544855775924011e-06,
+      "loss": 0.9535,
+      "step": 10928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543365473677387e-06,
+      "loss": 0.8322,
+      "step": 10929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541875544348985e-06,
+      "loss": 0.7857,
+      "step": 10930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5403859880133173e-06,
+      "loss": 0.8458,
+      "step": 10931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.53889680474488e-06,
+      "loss": 0.9239,
+      "step": 10932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5374079946181473e-06,
+      "loss": 0.9346,
+      "step": 10933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5359195577075802e-06,
+      "loss": 0.931,
+      "step": 10934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.534431494087617e-06,
+      "loss": 0.6411,
+      "step": 10935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.532943803832677e-06,
+      "loss": 0.913,
+      "step": 10936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.531456487017163e-06,
+      "loss": 0.9134,
+      "step": 10937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.529969543715455e-06,
+      "loss": 0.8139,
+      "step": 10938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5284829740019214e-06,
+      "loss": 0.7308,
+      "step": 10939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5269967779509073e-06,
+      "loss": 0.901,
+      "step": 10940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5255109556367395e-06,
+      "loss": 0.7931,
+      "step": 10941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.524025507133726e-06,
+      "loss": 0.8387,
+      "step": 10942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5225404325161528e-06,
+      "loss": 0.8418,
+      "step": 10943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.521055731858297e-06,
+      "loss": 0.9489,
+      "step": 10944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5195714052344057e-06,
+      "loss": 0.5981,
+      "step": 10945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5180874527187183e-06,
+      "loss": 0.836,
+      "step": 10946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.516603874385445e-06,
+      "loss": 0.8718,
+      "step": 10947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5151206703087817e-06,
+      "loss": 0.7979,
+      "step": 10948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5136378405629093e-06,
+      "loss": 0.9403,
+      "step": 10949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.512155385221985e-06,
+      "loss": 0.8344,
+      "step": 10950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.510673304360147e-06,
+      "loss": 0.9353,
+      "step": 10951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.509191598051519e-06,
+      "loss": 0.7933,
+      "step": 10952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5077102663702003e-06,
+      "loss": 0.8342,
+      "step": 10953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5062293093902778e-06,
+      "loss": 0.9004,
+      "step": 10954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5047487271858163e-06,
+      "loss": 0.8654,
+      "step": 10955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5032685198308615e-06,
+      "loss": 0.8673,
+      "step": 10956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.50178868739944e-06,
+      "loss": 0.9513,
+      "step": 10957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5003092299655586e-06,
+      "loss": 0.9049,
+      "step": 10958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.498830147603212e-06,
+      "loss": 0.906,
+      "step": 10959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4973514403863707e-06,
+      "loss": 0.954,
+      "step": 10960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4958731083889854e-06,
+      "loss": 0.857,
+      "step": 10961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4943951516849897e-06,
+      "loss": 0.7962,
+      "step": 10962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4929175703482964e-06,
+      "loss": 0.9653,
+      "step": 10963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4914403644528084e-06,
+      "loss": 0.7959,
+      "step": 10964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4899635340723992e-06,
+      "loss": 0.9836,
+      "step": 10965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.488487079280928e-06,
+      "loss": 0.9748,
+      "step": 10966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.487011000152233e-06,
+      "loss": 0.8512,
+      "step": 10967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4855352967601345e-06,
+      "loss": 0.8854,
+      "step": 10968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4840599691784385e-06,
+      "loss": 0.8635,
+      "step": 10969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4825850174809274e-06,
+      "loss": 0.6301,
+      "step": 10970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4811104417413655e-06,
+      "loss": 0.8346,
+      "step": 10971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.479636242033495e-06,
+      "loss": 0.8489,
+      "step": 10972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4781624184310505e-06,
+      "loss": 0.9817,
+      "step": 10973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.476688971007736e-06,
+      "loss": 0.9067,
+      "step": 10974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4752158998372398e-06,
+      "loss": 0.9283,
+      "step": 10975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.473743204993232e-06,
+      "loss": 0.8983,
+      "step": 10976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4722708865493696e-06,
+      "loss": 0.7198,
+      "step": 10977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.470798944579279e-06,
+      "loss": 0.8396,
+      "step": 10978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4693273791565797e-06,
+      "loss": 0.6189,
+      "step": 10979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.467856190354867e-06,
+      "loss": 0.773,
+      "step": 10980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4663853782477143e-06,
+      "loss": 0.8403,
+      "step": 10981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4649149429086805e-06,
+      "loss": 0.8316,
+      "step": 10982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.463444884411301e-06,
+      "loss": 0.926,
+      "step": 10983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4619752028291033e-06,
+      "loss": 0.8033,
+      "step": 10984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4605058982355836e-06,
+      "loss": 1.0595,
+      "step": 10985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.459036970704225e-06,
+      "loss": 1.0156,
+      "step": 10986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4575684203084904e-06,
+      "loss": 0.8871,
+      "step": 10987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.456100247121822e-06,
+      "loss": 0.8626,
+      "step": 10988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.454632451217652e-06,
+      "loss": 0.6174,
+      "step": 10989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4531650326693825e-06,
+      "loss": 0.9091,
+      "step": 10990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4516979915504035e-06,
+      "loss": 0.8902,
+      "step": 10991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4502313279340826e-06,
+      "loss": 0.8066,
+      "step": 10992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.448765041893767e-06,
+      "loss": 0.7804,
+      "step": 10993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.447299133502795e-06,
+      "loss": 0.8375,
+      "step": 10994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4458336028344753e-06,
+      "loss": 0.8867,
+      "step": 10995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.444368449962101e-06,
+      "loss": 0.8182,
+      "step": 10996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4429036749589454e-06,
+      "loss": 0.8197,
+      "step": 10997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441439277898269e-06,
+      "loss": 0.9445,
+      "step": 10998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4399752588533054e-06,
+      "loss": 0.9597,
+      "step": 10999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4385116178972736e-06,
+      "loss": 0.8057,
+      "step": 11000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437048355103371e-06,
+      "loss": 0.9743,
+      "step": 11001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4355854705447756e-06,
+      "loss": 0.926,
+      "step": 11002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434122964294656e-06,
+      "loss": 0.9295,
+      "step": 11003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4326608364261493e-06,
+      "loss": 0.8061,
+      "step": 11004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4311990870123793e-06,
+      "loss": 0.8907,
+      "step": 11005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4297377161264512e-06,
+      "loss": 0.6192,
+      "step": 11006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4282767238414485e-06,
+      "loss": 0.8463,
+      "step": 11007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426816110230438e-06,
+      "loss": 0.8584,
+      "step": 11008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425355875366472e-06,
+      "loss": 0.7885,
+      "step": 11009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423896019322576e-06,
+      "loss": 0.8317,
+      "step": 11010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422436542171759e-06,
+      "loss": 0.9993,
+      "step": 11011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4209774439870126e-06,
+      "loss": 0.8657,
+      "step": 11012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4195187248413066e-06,
+      "loss": 0.868,
+      "step": 11013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4180603848075977e-06,
+      "loss": 0.8874,
+      "step": 11014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4166024239588183e-06,
+      "loss": 0.8736,
+      "step": 11015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4151448423678823e-06,
+      "loss": 0.6163,
+      "step": 11016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413687640107687e-06,
+      "loss": 0.7994,
+      "step": 11017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4122308172511057e-06,
+      "loss": 0.8339,
+      "step": 11018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4107743738710022e-06,
+      "loss": 0.8745,
+      "step": 11019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409318310040213e-06,
+      "loss": 0.8579,
+      "step": 11020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4078626258315573e-06,
+      "loss": 0.8151,
+      "step": 11021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4064073213178353e-06,
+      "loss": 0.8649,
+      "step": 11022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4049523965718323e-06,
+      "loss": 1.0617,
+      "step": 11023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40349785166631e-06,
+      "loss": 0.8641,
+      "step": 11024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.402043686674013e-06,
+      "loss": 0.8615,
+      "step": 11025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4005899016676657e-06,
+      "loss": 0.7382,
+      "step": 11026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3991364967199715e-06,
+      "loss": 0.6422,
+      "step": 11027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3976834719036226e-06,
+      "loss": 0.8256,
+      "step": 11028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396230827291285e-06,
+      "loss": 0.8838,
+      "step": 11029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3947785629556076e-06,
+      "loss": 0.7854,
+      "step": 11030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39332667896922e-06,
+      "loss": 0.8969,
+      "step": 11031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3918751754047307e-06,
+      "loss": 0.8866,
+      "step": 11032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390424052334739e-06,
+      "loss": 0.6138,
+      "step": 11033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3889733098318126e-06,
+      "loss": 0.8142,
+      "step": 11034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387522947968506e-06,
+      "loss": 0.6032,
+      "step": 11035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3860729668173545e-06,
+      "loss": 0.7528,
+      "step": 11036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384623366450872e-06,
+      "loss": 0.8347,
+      "step": 11037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383174146941559e-06,
+      "loss": 0.813,
+      "step": 11038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3817253083618896e-06,
+      "loss": 0.8191,
+      "step": 11039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3802768507843275e-06,
+      "loss": 0.9461,
+      "step": 11040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3788287742813098e-06,
+      "loss": 0.8088,
+      "step": 11041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3773810789252527e-06,
+      "loss": 0.6064,
+      "step": 11042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375933764788566e-06,
+      "loss": 0.8124,
+      "step": 11043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374486831943628e-06,
+      "loss": 0.942,
+      "step": 11044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3730402804628006e-06,
+      "loss": 0.9022,
+      "step": 11045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3715941104184304e-06,
+      "loss": 0.9683,
+      "step": 11046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3701483218828402e-06,
+      "loss": 0.8174,
+      "step": 11047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3687029149283393e-06,
+      "loss": 0.863,
+      "step": 11048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3672578896272146e-06,
+      "loss": 0.7142,
+      "step": 11049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3658132460517324e-06,
+      "loss": 0.572,
+      "step": 11050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3643689842741425e-06,
+      "loss": 0.9226,
+      "step": 11051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.362925104366671e-06,
+      "loss": 0.8277,
+      "step": 11052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3614816064015357e-06,
+      "loss": 0.8159,
+      "step": 11053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3600384904509255e-06,
+      "loss": 0.8542,
+      "step": 11054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3585957565870112e-06,
+      "loss": 0.7804,
+      "step": 11055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357153404881947e-06,
+      "loss": 0.8747,
+      "step": 11056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3557114354078635e-06,
+      "loss": 0.9092,
+      "step": 11057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354269848236884e-06,
+      "loss": 0.6108,
+      "step": 11058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352828643441101e-06,
+      "loss": 0.9023,
+      "step": 11059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351387821092589e-06,
+      "loss": 0.8452,
+      "step": 11060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349947381263409e-06,
+      "loss": 1.0341,
+      "step": 11061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348507324025595e-06,
+      "loss": 0.8478,
+      "step": 11062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3470676494511723e-06,
+      "loss": 0.769,
+      "step": 11063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3456283576121396e-06,
+      "loss": 0.8134,
+      "step": 11064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3441894485804774e-06,
+      "loss": 0.9238,
+      "step": 11065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3427509224281454e-06,
+      "loss": 0.8625,
+      "step": 11066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3413127792270906e-06,
+      "loss": 0.9001,
+      "step": 11067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339875019049237e-06,
+      "loss": 0.8732,
+      "step": 11068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3384376419664866e-06,
+      "loss": 0.8389,
+      "step": 11069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3370006480507246e-06,
+      "loss": 0.9395,
+      "step": 11070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33556403737382e-06,
+      "loss": 0.8233,
+      "step": 11071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334127810007617e-06,
+      "loss": 0.8705,
+      "step": 11072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3326919660239477e-06,
+      "loss": 1.0318,
+      "step": 11073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331256505494619e-06,
+      "loss": 0.8344,
+      "step": 11074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329821428491419e-06,
+      "loss": 0.8427,
+      "step": 11075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328386735086121e-06,
+      "loss": 0.8071,
+      "step": 11076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3269524253504718e-06,
+      "loss": 0.8533,
+      "step": 11077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3255184993562086e-06,
+      "loss": 0.5978,
+      "step": 11078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.324084957175042e-06,
+      "loss": 0.7939,
+      "step": 11079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3226517988786667e-06,
+      "loss": 0.9069,
+      "step": 11080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3212190245387556e-06,
+      "loss": 0.848,
+      "step": 11081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3197866342269627e-06,
+      "loss": 0.8469,
+      "step": 11082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3183546280149292e-06,
+      "loss": 0.8416,
+      "step": 11083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.316923005974269e-06,
+      "loss": 0.888,
+      "step": 11084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.31549176817658e-06,
+      "loss": 0.9325,
+      "step": 11085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.31406091469344e-06,
+      "loss": 0.8504,
+      "step": 11086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3126304455964064e-06,
+      "loss": 0.8378,
+      "step": 11087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3112003609570253e-06,
+      "loss": 0.8126,
+      "step": 11088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.309770660846814e-06,
+      "loss": 0.8139,
+      "step": 11089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3083413453372736e-06,
+      "loss": 0.8159,
+      "step": 11090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3069124144998844e-06,
+      "loss": 0.8917,
+      "step": 11091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3054838684061143e-06,
+      "loss": 0.8467,
+      "step": 11092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.304055707127406e-06,
+      "loss": 0.6289,
+      "step": 11093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.302627930735182e-06,
+      "loss": 0.9983,
+      "step": 11094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3012005393008496e-06,
+      "loss": 0.8076,
+      "step": 11095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2997735328957904e-06,
+      "loss": 0.9216,
+      "step": 11096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.298346911591378e-06,
+      "loss": 0.8969,
+      "step": 11097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2969206754589577e-06,
+      "loss": 0.8817,
+      "step": 11098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2954948245698572e-06,
+      "loss": 0.9368,
+      "step": 11099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.294069358995384e-06,
+      "loss": 0.935,
+      "step": 11100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2926442788068283e-06,
+      "loss": 0.747,
+      "step": 11101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.29121958407546e-06,
+      "loss": 0.9776,
+      "step": 11102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.289795274872536e-06,
+      "loss": 0.9361,
+      "step": 11103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2883713512692852e-06,
+      "loss": 0.9464,
+      "step": 11104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.286947813336918e-06,
+      "loss": 0.944,
+      "step": 11105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.28552466114663e-06,
+      "loss": 0.7708,
+      "step": 11106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2841018947695914e-06,
+      "loss": 0.8649,
+      "step": 11107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.282679514276963e-06,
+      "loss": 0.8589,
+      "step": 11108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.281257519739879e-06,
+      "loss": 0.8183,
+      "step": 11109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.279835911229452e-06,
+      "loss": 0.9467,
+      "step": 11110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2784146888167824e-06,
+      "loss": 0.5871,
+      "step": 11111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.276993852572944e-06,
+      "loss": 0.9627,
+      "step": 11112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.275573402569e-06,
+      "loss": 0.9168,
+      "step": 11113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2741533388759863e-06,
+      "loss": 0.8769,
+      "step": 11114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2727336615649243e-06,
+      "loss": 0.5921,
+      "step": 11115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2713143707068096e-06,
+      "loss": 0.976,
+      "step": 11116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.26989546637263e-06,
+      "loss": 0.8748,
+      "step": 11117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2684769486333445e-06,
+      "loss": 0.9343,
+      "step": 11118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.267058817559895e-06,
+      "loss": 0.8891,
+      "step": 11119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2656410732232035e-06,
+      "loss": 0.8571,
+      "step": 11120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2642237156941727e-06,
+      "loss": 0.9503,
+      "step": 11121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.262806745043692e-06,
+      "loss": 0.8214,
+      "step": 11122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2613901613426226e-06,
+      "loss": 0.9025,
+      "step": 11123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2599739646618114e-06,
+      "loss": 0.9641,
+      "step": 11124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2585581550720827e-06,
+      "loss": 0.6449,
+      "step": 11125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.257142732644243e-06,
+      "loss": 0.7821,
+      "step": 11126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.255727697449084e-06,
+      "loss": 0.8448,
+      "step": 11127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2543130495573718e-06,
+      "loss": 0.9047,
+      "step": 11128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2528987890398544e-06,
+      "loss": 0.9644,
+      "step": 11129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2514849159672625e-06,
+      "loss": 0.8927,
+      "step": 11130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2500714304103013e-06,
+      "loss": 0.7793,
+      "step": 11131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.248658332439669e-06,
+      "loss": 0.9233,
+      "step": 11132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.247245622126031e-06,
+      "loss": 0.8167,
+      "step": 11133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.245833299540043e-06,
+      "loss": 0.6046,
+      "step": 11134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2444213647523373e-06,
+      "loss": 0.8434,
+      "step": 11135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2430098178335227e-06,
+      "loss": 0.839,
+      "step": 11136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2415986588541995e-06,
+      "loss": 0.7951,
+      "step": 11137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.240187887884938e-06,
+      "loss": 0.9571,
+      "step": 11138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2387775049962935e-06,
+      "loss": 0.9645,
+      "step": 11139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237367510258802e-06,
+      "loss": 0.8757,
+      "step": 11140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2359579037429755e-06,
+      "loss": 0.9152,
+      "step": 11141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2345486855193187e-06,
+      "loss": 0.8446,
+      "step": 11142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233139855658304e-06,
+      "loss": 0.9643,
+      "step": 11143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.231731414230389e-06,
+      "loss": 0.7939,
+      "step": 11144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2303233613060148e-06,
+      "loss": 0.8565,
+      "step": 11145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2289156969555948e-06,
+      "loss": 0.8717,
+      "step": 11146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2275084212495344e-06,
+      "loss": 0.8798,
+      "step": 11147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2261015342582125e-06,
+      "loss": 0.8174,
+      "step": 11148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224695036051988e-06,
+      "loss": 0.9377,
+      "step": 11149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2232889267012038e-06,
+      "loss": 0.8922,
+      "step": 11150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2218832062761773e-06,
+      "loss": 0.8799,
+      "step": 11151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2204778748472178e-06,
+      "loss": 0.9355,
+      "step": 11152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219072932484605e-06,
+      "loss": 0.8605,
+      "step": 11153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217668379258602e-06,
+      "loss": 0.771,
+      "step": 11154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216264215239452e-06,
+      "loss": 0.8209,
+      "step": 11155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2148604404973784e-06,
+      "loss": 0.7793,
+      "step": 11156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2134570551025903e-06,
+      "loss": 0.7885,
+      "step": 11157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2120540591252714e-06,
+      "loss": 0.6128,
+      "step": 11158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210651452635587e-06,
+      "loss": 0.9248,
+      "step": 11159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2092492357036823e-06,
+      "loss": 0.8453,
+      "step": 11160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207847408399688e-06,
+      "loss": 0.8291,
+      "step": 11161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2064459707937112e-06,
+      "loss": 0.8958,
+      "step": 11162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205044922955838e-06,
+      "loss": 0.8212,
+      "step": 11163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2036442649561384e-06,
+      "loss": 0.9526,
+      "step": 11164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2022439968646582e-06,
+      "loss": 0.9351,
+      "step": 11165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2008441187514295e-06,
+      "loss": 0.7722,
+      "step": 11166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1994446306864647e-06,
+      "loss": 0.9631,
+      "step": 11167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1980455327397545e-06,
+      "loss": 1.0409,
+      "step": 11168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196646824981267e-06,
+      "loss": 0.8948,
+      "step": 11169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1952485074809547e-06,
+      "loss": 0.847,
+      "step": 11170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1938505803087475e-06,
+      "loss": 0.9326,
+      "step": 11171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1924530435345627e-06,
+      "loss": 0.9321,
+      "step": 11172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1910558972282916e-06,
+      "loss": 0.9333,
+      "step": 11173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1896591414598066e-06,
+      "loss": 0.79,
+      "step": 11174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1882627762989628e-06,
+      "loss": 0.786,
+      "step": 11175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18686680181559e-06,
+      "loss": 0.9331,
+      "step": 11176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185471218079511e-06,
+      "loss": 0.7893,
+      "step": 11177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1840760251605175e-06,
+      "loss": 0.8638,
+      "step": 11178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1826812231283857e-06,
+      "loss": 0.8735,
+      "step": 11179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1812868120528695e-06,
+      "loss": 0.8798,
+      "step": 11180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179892792003706e-06,
+      "loss": 0.9375,
+      "step": 11181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178499163050617e-06,
+      "loss": 0.9015,
+      "step": 11182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1771059252632965e-06,
+      "loss": 0.9881,
+      "step": 11183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1757130787114234e-06,
+      "loss": 1.0061,
+      "step": 11184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.174320623464652e-06,
+      "loss": 0.8295,
+      "step": 11185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1729285595926286e-06,
+      "loss": 0.6152,
+      "step": 11186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171536887164969e-06,
+      "loss": 0.882,
+      "step": 11187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1701456062512726e-06,
+      "loss": 0.8849,
+      "step": 11188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1687547169211207e-06,
+      "loss": 0.7453,
+      "step": 11189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16736421924407e-06,
+      "loss": 0.8647,
+      "step": 11190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165974113289667e-06,
+      "loss": 0.7988,
+      "step": 11191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164584399127431e-06,
+      "loss": 0.9918,
+      "step": 11192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163195076826864e-06,
+      "loss": 0.8586,
+      "step": 11193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1618061464574468e-06,
+      "loss": 0.8869,
+      "step": 11194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1604176080886408e-06,
+      "loss": 0.8409,
+      "step": 11195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159029461789892e-06,
+      "loss": 0.8416,
+      "step": 11196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1576417076306254e-06,
+      "loss": 0.8556,
+      "step": 11197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156254345680242e-06,
+      "loss": 0.8589,
+      "step": 11198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1548673760081274e-06,
+      "loss": 0.8895,
+      "step": 11199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1534807986836448e-06,
+      "loss": 0.6105,
+      "step": 11200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1520946137761367e-06,
+      "loss": 0.7701,
+      "step": 11201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1507088213549345e-06,
+      "loss": 0.9086,
+      "step": 11202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14932342148934e-06,
+      "loss": 0.9777,
+      "step": 11203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14793841424864e-06,
+      "loss": 0.9144,
+      "step": 11204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1465537997021015e-06,
+      "loss": 0.8585,
+      "step": 11205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451695779189675e-06,
+      "loss": 0.979,
+      "step": 11206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143785748968472e-06,
+      "loss": 0.8714,
+      "step": 11207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142402312919818e-06,
+      "loss": 0.6141,
+      "step": 11208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141019269842195e-06,
+      "loss": 0.9701,
+      "step": 11209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.139636619804767e-06,
+      "loss": 0.8217,
+      "step": 11210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138254362876688e-06,
+      "loss": 0.8512,
+      "step": 11211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136872499127086e-06,
+      "loss": 0.9088,
+      "step": 11212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1354910286250694e-06,
+      "loss": 0.8836,
+      "step": 11213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134109951439727e-06,
+      "loss": 0.6041,
+      "step": 11214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1327292676401256e-06,
+      "loss": 0.8334,
+      "step": 11215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131348977295322e-06,
+      "loss": 0.819,
+      "step": 11216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1299690804743435e-06,
+      "loss": 0.9156,
+      "step": 11217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1285895772462016e-06,
+      "loss": 0.8427,
+      "step": 11218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1272104676798868e-06,
+      "loss": 0.8602,
+      "step": 11219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125831751844367e-06,
+      "loss": 0.866,
+      "step": 11220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.124453429808602e-06,
+      "loss": 0.8892,
+      "step": 11221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.123075501641518e-06,
+      "loss": 0.8386,
+      "step": 11222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1216979674120297e-06,
+      "loss": 0.8628,
+      "step": 11223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1203208271890286e-06,
+      "loss": 0.9329,
+      "step": 11224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1189440810413863e-06,
+      "loss": 0.8623,
+      "step": 11225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.11756772903796e-06,
+      "loss": 0.8457,
+      "step": 11226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.116191771247579e-06,
+      "loss": 0.8836,
+      "step": 11227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.114816207739061e-06,
+      "loss": 0.9174,
+      "step": 11228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1134410385811997e-06,
+      "loss": 0.8931,
+      "step": 11229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1120662638427645e-06,
+      "loss": 0.9935,
+      "step": 11230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.110691883592517e-06,
+      "loss": 0.601,
+      "step": 11231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.109317897899189e-06,
+      "loss": 0.8554,
+      "step": 11232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1079443068314964e-06,
+      "loss": 0.8308,
+      "step": 11233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1065711104581333e-06,
+      "loss": 0.8785,
+      "step": 11234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1051983088477732e-06,
+      "loss": 0.8598,
+      "step": 11235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1038259020690787e-06,
+      "loss": 0.7632,
+      "step": 11236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.102453890190682e-06,
+      "loss": 0.8581,
+      "step": 11237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1010822732811997e-06,
+      "loss": 0.9249,
+      "step": 11238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.099711051409229e-06,
+      "loss": 1.0141,
+      "step": 11239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0983402246433437e-06,
+      "loss": 0.9156,
+      "step": 11240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.096969793052108e-06,
+      "loss": 0.7592,
+      "step": 11241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.095599756704054e-06,
+      "loss": 0.9445,
+      "step": 11242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0942301156677024e-06,
+      "loss": 0.8721,
+      "step": 11243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0928608700115493e-06,
+      "loss": 0.7557,
+      "step": 11244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.09149201980407e-06,
+      "loss": 0.8426,
+      "step": 11245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0901235651137285e-06,
+      "loss": 0.8295,
+      "step": 11246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0887555060089616e-06,
+      "loss": 0.9207,
+      "step": 11247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0873878425581882e-06,
+      "loss": 0.8608,
+      "step": 11248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.086020574829806e-06,
+      "loss": 0.9225,
+      "step": 11249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.084653702892192e-06,
+      "loss": 0.9431,
+      "step": 11250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0832872268137116e-06,
+      "loss": 0.8658,
+      "step": 11251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.081921146662701e-06,
+      "loss": 0.9632,
+      "step": 11252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0805554625074808e-06,
+      "loss": 0.9008,
+      "step": 11253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0791901744163477e-06,
+      "loss": 0.8266,
+      "step": 11254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0778252824575882e-06,
+      "loss": 0.5932,
+      "step": 11255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0764607866994593e-06,
+      "loss": 0.8895,
+      "step": 11256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.075096687210202e-06,
+      "loss": 0.8325,
+      "step": 11257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0737329840580365e-06,
+      "loss": 0.6164,
+      "step": 11258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0723696773111625e-06,
+      "loss": 0.8652,
+      "step": 11259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0710067670377622e-06,
+      "loss": 0.817,
+      "step": 11260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0696442533060004e-06,
+      "loss": 0.8989,
+      "step": 11261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0682821361840155e-06,
+      "loss": 0.8887,
+      "step": 11262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.066920415739929e-06,
+      "loss": 0.887,
+      "step": 11263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0655590920418434e-06,
+      "loss": 0.8777,
+      "step": 11264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.064198165157838e-06,
+      "loss": 0.9327,
+      "step": 11265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0628376351559796e-06,
+      "loss": 0.9092,
+      "step": 11266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0614775021043087e-06,
+      "loss": 0.9404,
+      "step": 11267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.060117766070847e-06,
+      "loss": 0.9044,
+      "step": 11268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0587584271235984e-06,
+      "loss": 0.8884,
+      "step": 11269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.057399485330541e-06,
+      "loss": 0.8212,
+      "step": 11270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.056040940759644e-06,
+      "loss": 0.7947,
+      "step": 11271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.054682793478847e-06,
+      "loss": 0.8987,
+      "step": 11272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0533250435560735e-06,
+      "loss": 0.936,
+      "step": 11273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0519676910592277e-06,
+      "loss": 0.9146,
+      "step": 11274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.05061073605619e-06,
+      "loss": 0.8992,
+      "step": 11275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0492541786148278e-06,
+      "loss": 0.9012,
+      "step": 11276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.047898018802984e-06,
+      "loss": 0.7522,
+      "step": 11277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.046542256688481e-06,
+      "loss": 0.9169,
+      "step": 11278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.045186892339119e-06,
+      "loss": 0.8163,
+      "step": 11279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0438319258226904e-06,
+      "loss": 0.7674,
+      "step": 11280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0424773572069533e-06,
+      "loss": 0.9098,
+      "step": 11281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041123186559655e-06,
+      "loss": 0.6299,
+      "step": 11282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039769413948517e-06,
+      "loss": 0.8446,
+      "step": 11283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038416039441241e-06,
+      "loss": 0.894,
+      "step": 11284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037063063105519e-06,
+      "loss": 0.7935,
+      "step": 11285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0357104850090117e-06,
+      "loss": 0.8101,
+      "step": 11286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034358305219364e-06,
+      "loss": 0.7851,
+      "step": 11287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0330065238041984e-06,
+      "loss": 0.8238,
+      "step": 11288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0316551408311204e-06,
+      "loss": 0.8646,
+      "step": 11289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030304156367715e-06,
+      "loss": 0.9803,
+      "step": 11290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0289535704815513e-06,
+      "loss": 0.8632,
+      "step": 11291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.02760338324017e-06,
+      "loss": 0.8508,
+      "step": 11292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0262535947110973e-06,
+      "loss": 0.8988,
+      "step": 11293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0249042049618384e-06,
+      "loss": 0.9505,
+      "step": 11294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023555214059876e-06,
+      "loss": 0.8091,
+      "step": 11295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022206622072679e-06,
+      "loss": 0.9022,
+      "step": 11296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0208584290676913e-06,
+      "loss": 0.8386,
+      "step": 11297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0195106351123374e-06,
+      "loss": 0.8551,
+      "step": 11298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0181632402740216e-06,
+      "loss": 0.8426,
+      "step": 11299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0168162446201323e-06,
+      "loss": 0.981,
+      "step": 11300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0154696482180335e-06,
+      "loss": 0.8523,
+      "step": 11301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0141234511350704e-06,
+      "loss": 0.9416,
+      "step": 11302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012777653438569e-06,
+      "loss": 0.8218,
+      "step": 11303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011432255195832e-06,
+      "loss": 0.8534,
+      "step": 11304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0100872564741504e-06,
+      "loss": 0.8556,
+      "step": 11305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0087426573407855e-06,
+      "loss": 0.8793,
+      "step": 11306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0073984578629845e-06,
+      "loss": 0.8209,
+      "step": 11307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006054658107972e-06,
+      "loss": 0.8681,
+      "step": 11308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0047112581429517e-06,
+      "loss": 0.8031,
+      "step": 11309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003368258035114e-06,
+      "loss": 0.8602,
+      "step": 11310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002025657851623e-06,
+      "loss": 0.9888,
+      "step": 11311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0006834576596224e-06,
+      "loss": 0.8289,
+      "step": 11312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999341657526239e-06,
+      "loss": 0.8438,
+      "step": 11313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9980002575185753e-06,
+      "loss": 0.9458,
+      "step": 11314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9966592577037224e-06,
+      "loss": 0.8684,
+      "step": 11315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9953186581487427e-06,
+      "loss": 0.9895,
+      "step": 11316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9939784589206823e-06,
+      "loss": 0.8188,
+      "step": 11317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9926386600865666e-06,
+      "loss": 0.9493,
+      "step": 11318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9912992617133987e-06,
+      "loss": 0.7007,
+      "step": 11319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989960263868168e-06,
+      "loss": 0.9544,
+      "step": 11320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988621666617836e-06,
+      "loss": 0.8193,
+      "step": 11321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987283470029353e-06,
+      "loss": 0.8517,
+      "step": 11322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9859456741696414e-06,
+      "loss": 0.5696,
+      "step": 11323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9846082791056043e-06,
+      "loss": 1.0128,
+      "step": 11324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9832712849041324e-06,
+      "loss": 0.869,
+      "step": 11325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981934691632088e-06,
+      "loss": 0.9238,
+      "step": 11326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9805984993563166e-06,
+      "loss": 0.931,
+      "step": 11327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979262708143642e-06,
+      "loss": 0.8712,
+      "step": 11328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977927318060868e-06,
+      "loss": 0.5951,
+      "step": 11329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9765923291747847e-06,
+      "loss": 0.9082,
+      "step": 11330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9752577415521544e-06,
+      "loss": 0.8528,
+      "step": 11331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9739235552597213e-06,
+      "loss": 0.922,
+      "step": 11332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9725897703642105e-06,
+      "loss": 0.9716,
+      "step": 11333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971256386932324e-06,
+      "loss": 0.8687,
+      "step": 11334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9699234050307524e-06,
+      "loss": 0.8962,
+      "step": 11335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9685908247261565e-06,
+      "loss": 0.9297,
+      "step": 11336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9672586460851818e-06,
+      "loss": 0.87,
+      "step": 11337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9659268691744516e-06,
+      "loss": 0.8663,
+      "step": 11338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9645954940605694e-06,
+      "loss": 0.861,
+      "step": 11339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963264520810123e-06,
+      "loss": 0.9271,
+      "step": 11340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961933949489674e-06,
+      "loss": 0.907,
+      "step": 11341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9606037801657674e-06,
+      "loss": 0.8445,
+      "step": 11342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9592740129049257e-06,
+      "loss": 0.8758,
+      "step": 11343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9579446477736517e-06,
+      "loss": 0.7826,
+      "step": 11344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956615684838432e-06,
+      "loss": 0.8836,
+      "step": 11345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95528712416573e-06,
+      "loss": 0.7827,
+      "step": 11346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9539589658219883e-06,
+      "loss": 0.9082,
+      "step": 11347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952631209873628e-06,
+      "loss": 0.9347,
+      "step": 11348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951303856387056e-06,
+      "loss": 0.9286,
+      "step": 11349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9499769054286532e-06,
+      "loss": 0.8504,
+      "step": 11350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948650357064784e-06,
+      "loss": 0.9017,
+      "step": 11351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.947324211361791e-06,
+      "loss": 0.7941,
+      "step": 11352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945998468385992e-06,
+      "loss": 0.8392,
+      "step": 11353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9446731282036946e-06,
+      "loss": 0.8269,
+      "step": 11354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9433481908811826e-06,
+      "loss": 0.9261,
+      "step": 11355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.942023656484715e-06,
+      "loss": 0.8374,
+      "step": 11356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940699525080536e-06,
+      "loss": 0.8842,
+      "step": 11357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9393757967348658e-06,
+      "loss": 0.8347,
+      "step": 11358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9380524715139036e-06,
+      "loss": 0.8643,
+      "step": 11359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936729549483838e-06,
+      "loss": 0.8298,
+      "step": 11360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9354070307108254e-06,
+      "loss": 0.7752,
+      "step": 11361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934084915261009e-06,
+      "loss": 0.7967,
+      "step": 11362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9327632032005083e-06,
+      "loss": 0.8252,
+      "step": 11363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9314418945954226e-06,
+      "loss": 0.8433,
+      "step": 11364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9301209895118376e-06,
+      "loss": 0.9674,
+      "step": 11365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9288004880158117e-06,
+      "loss": 0.8527,
+      "step": 11366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.927480390173385e-06,
+      "loss": 0.9666,
+      "step": 11367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9261606960505776e-06,
+      "loss": 0.8446,
+      "step": 11368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9248414057133867e-06,
+      "loss": 1.007,
+      "step": 11369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9235225192277984e-06,
+      "loss": 0.8891,
+      "step": 11370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.922204036659768e-06,
+      "loss": 0.9079,
+      "step": 11371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.920885958075237e-06,
+      "loss": 0.881,
+      "step": 11372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9195682835401195e-06,
+      "loss": 0.8082,
+      "step": 11373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9182510131203223e-06,
+      "loss": 0.8245,
+      "step": 11374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9169341468817204e-06,
+      "loss": 0.85,
+      "step": 11375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.915617684890172e-06,
+      "loss": 0.9274,
+      "step": 11376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9143016272115157e-06,
+      "loss": 0.8956,
+      "step": 11377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9129859739115663e-06,
+      "loss": 0.889,
+      "step": 11378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116707250561293e-06,
+      "loss": 0.8013,
+      "step": 11379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.910355880710978e-06,
+      "loss": 0.9006,
+      "step": 11380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.90904144094187e-06,
+      "loss": 0.8781,
+      "step": 11381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.907727405814542e-06,
+      "loss": 0.8323,
+      "step": 11382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.90641377539471e-06,
+      "loss": 0.8822,
+      "step": 11383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9051005497480724e-06,
+      "loss": 0.9091,
+      "step": 11384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9037877289403085e-06,
+      "loss": 0.8862,
+      "step": 11385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9024753130370722e-06,
+      "loss": 0.7586,
+      "step": 11386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9011633021039976e-06,
+      "loss": 0.8794,
+      "step": 11387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8998516962067037e-06,
+      "loss": 0.857,
+      "step": 11388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.898540495410781e-06,
+      "loss": 0.8598,
+      "step": 11389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8972296997818106e-06,
+      "loss": 1.0509,
+      "step": 11390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8959193093853455e-06,
+      "loss": 0.8723,
+      "step": 11391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8946093242869201e-06,
+      "loss": 0.8778,
+      "step": 11392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8932997445520451e-06,
+      "loss": 0.7951,
+      "step": 11393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8919905702462216e-06,
+      "loss": 0.9467,
+      "step": 11394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8906818014349204e-06,
+      "loss": 0.8837,
+      "step": 11395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8893734381835938e-06,
+      "loss": 0.8444,
+      "step": 11396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.888065480557677e-06,
+      "loss": 0.7691,
+      "step": 11397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8867579286225802e-06,
+      "loss": 0.8186,
+      "step": 11398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8854507824436995e-06,
+      "loss": 0.9692,
+      "step": 11399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8841440420864067e-06,
+      "loss": 0.7914,
+      "step": 11400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8828377076160541e-06,
+      "loss": 0.8954,
+      "step": 11401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8815317790979715e-06,
+      "loss": 0.9358,
+      "step": 11402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8802262565974705e-06,
+      "loss": 0.8056,
+      "step": 11403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8789211401798458e-06,
+      "loss": 1.0046,
+      "step": 11404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.877616429910366e-06,
+      "loss": 0.8636,
+      "step": 11405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8763121258542816e-06,
+      "loss": 0.8631,
+      "step": 11406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8750082280768244e-06,
+      "loss": 0.8766,
+      "step": 11407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8737047366432004e-06,
+      "loss": 0.8016,
+      "step": 11408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8724016516186038e-06,
+      "loss": 0.8445,
+      "step": 11409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8710989730682039e-06,
+      "loss": 0.86,
+      "step": 11410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8697967010571473e-06,
+      "loss": 0.9024,
+      "step": 11411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8684948356505639e-06,
+      "loss": 0.8218,
+      "step": 11412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8671933769135598e-06,
+      "loss": 0.9366,
+      "step": 11413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8658923249112271e-06,
+      "loss": 0.9486,
+      "step": 11414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8645916797086283e-06,
+      "loss": 0.9149,
+      "step": 11415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.863291441370817e-06,
+      "loss": 1.0844,
+      "step": 11416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8619916099628178e-06,
+      "loss": 0.8899,
+      "step": 11417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8606921855496329e-06,
+      "loss": 0.9024,
+      "step": 11418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8593931681962562e-06,
+      "loss": 0.7878,
+      "step": 11419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.858094557967649e-06,
+      "loss": 0.7906,
+      "step": 11420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.856796354928757e-06,
+      "loss": 0.9915,
+      "step": 11421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8554985591445074e-06,
+      "loss": 0.9206,
+      "step": 11422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8542011706798002e-06,
+      "loss": 0.8094,
+      "step": 11423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8529041895995258e-06,
+      "loss": 0.9952,
+      "step": 11424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.851607615968547e-06,
+      "loss": 0.8244,
+      "step": 11425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.850311449851705e-06,
+      "loss": 0.8256,
+      "step": 11426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8490156913138258e-06,
+      "loss": 0.8934,
+      "step": 11427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8477203404197076e-06,
+      "loss": 0.9576,
+      "step": 11428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464253972341395e-06,
+      "loss": 0.8732,
+      "step": 11429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84513086182188e-06,
+      "loss": 0.849,
+      "step": 11430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8438367342476725e-06,
+      "loss": 0.7533,
+      "step": 11431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8425430145762368e-06,
+      "loss": 0.8736,
+      "step": 11432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.841249702872272e-06,
+      "loss": 0.94,
+      "step": 11433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8399567992004642e-06,
+      "loss": 0.8365,
+      "step": 11434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8386643036254704e-06,
+      "loss": 0.8399,
+      "step": 11435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8373722162119302e-06,
+      "loss": 0.8833,
+      "step": 11436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8360805370244628e-06,
+      "loss": 0.8046,
+      "step": 11437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8347892661276657e-06,
+      "loss": 0.8546,
+      "step": 11438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8334984035861226e-06,
+      "loss": 0.8626,
+      "step": 11439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8322079494643874e-06,
+      "loss": 0.9156,
+      "step": 11440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8309179038269987e-06,
+      "loss": 0.8582,
+      "step": 11441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8296282667384724e-06,
+      "loss": 0.8456,
+      "step": 11442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8283390382633081e-06,
+      "loss": 0.852,
+      "step": 11443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827050218465981e-06,
+      "loss": 1.0644,
+      "step": 11444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8257618074109474e-06,
+      "loss": 0.8559,
+      "step": 11445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824473805162642e-06,
+      "loss": 1.0035,
+      "step": 11446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823186211785477e-06,
+      "loss": 0.8961,
+      "step": 11447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8218990273438507e-06,
+      "loss": 0.812,
+      "step": 11448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8206122519021384e-06,
+      "loss": 0.864,
+      "step": 11449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8193258855246932e-06,
+      "loss": 0.8587,
+      "step": 11450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8180399282758464e-06,
+      "loss": 0.7318,
+      "step": 11451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8167543802199116e-06,
+      "loss": 0.7613,
+      "step": 11452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8154692414211784e-06,
+      "loss": 0.9096,
+      "step": 11453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814184511943925e-06,
+      "loss": 0.9041,
+      "step": 11454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.812900191852398e-06,
+      "loss": 0.8866,
+      "step": 11455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8116162812108295e-06,
+      "loss": 0.9221,
+      "step": 11456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810332780083429e-06,
+      "loss": 0.7932,
+      "step": 11457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8090496885343867e-06,
+      "loss": 0.7875,
+      "step": 11458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807767006627874e-06,
+      "loss": 0.7898,
+      "step": 11459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8064847344280378e-06,
+      "loss": 0.8732,
+      "step": 11460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805202871999009e-06,
+      "loss": 0.8466,
+      "step": 11461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8039214194048926e-06,
+      "loss": 0.8972,
+      "step": 11462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8026403767097745e-06,
+      "loss": 0.9012,
+      "step": 11463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8013597439777286e-06,
+      "loss": 0.8169,
+      "step": 11464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8000795212727961e-06,
+      "loss": 0.8792,
+      "step": 11465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7987997086590048e-06,
+      "loss": 0.7709,
+      "step": 11466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7975203062003567e-06,
+      "loss": 1.0604,
+      "step": 11467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7962413139608425e-06,
+      "loss": 0.8229,
+      "step": 11468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794962732004424e-06,
+      "loss": 0.9783,
+      "step": 11469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7936845603950447e-06,
+      "loss": 0.9606,
+      "step": 11470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7924067991966277e-06,
+      "loss": 0.7846,
+      "step": 11471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7911294484730747e-06,
+      "loss": 0.8179,
+      "step": 11472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7898525082882723e-06,
+      "loss": 0.8439,
+      "step": 11473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7885759787060797e-06,
+      "loss": 0.7879,
+      "step": 11474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787299859790338e-06,
+      "loss": 0.8699,
+      "step": 11475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.786024151604868e-06,
+      "loss": 0.9362,
+      "step": 11476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7847488542134684e-06,
+      "loss": 0.9265,
+      "step": 11477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7834739676799206e-06,
+      "loss": 0.8122,
+      "step": 11478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7821994920679864e-06,
+      "loss": 0.9731,
+      "step": 11479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7809254274414011e-06,
+      "loss": 0.9655,
+      "step": 11480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.779651773863883e-06,
+      "loss": 0.8163,
+      "step": 11481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.778378531399131e-06,
+      "loss": 0.7911,
+      "step": 11482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7771057001108171e-06,
+      "loss": 0.9284,
+      "step": 11483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7758332800626055e-06,
+      "loss": 0.8134,
+      "step": 11484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774561271318126e-06,
+      "loss": 0.9288,
+      "step": 11485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7732896739409966e-06,
+      "loss": 0.9348,
+      "step": 11486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7720184879948088e-06,
+      "loss": 0.7934,
+      "step": 11487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7707477135431406e-06,
+      "loss": 0.9669,
+      "step": 11488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7694773506495444e-06,
+      "loss": 0.8506,
+      "step": 11489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7682073993775517e-06,
+      "loss": 0.943,
+      "step": 11490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.766937859790675e-06,
+      "loss": 0.8351,
+      "step": 11491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7656687319524045e-06,
+      "loss": 0.9377,
+      "step": 11492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7644000159262165e-06,
+      "loss": 0.8484,
+      "step": 11493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7631317117755576e-06,
+      "loss": 0.9074,
+      "step": 11494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7618638195638581e-06,
+      "loss": 0.829,
+      "step": 11495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7605963393545278e-06,
+      "loss": 0.9253,
+      "step": 11496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7593292712109534e-06,
+      "loss": 0.8601,
+      "step": 11497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7580626151965063e-06,
+      "loss": 0.9149,
+      "step": 11498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.756796371374533e-06,
+      "loss": 0.8095,
+      "step": 11499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.75553053980836e-06,
+      "loss": 0.8741,
+      "step": 11500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7542651205612938e-06,
+      "loss": 0.5955,
+      "step": 11501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.753000113696617e-06,
+      "loss": 0.9136,
+      "step": 11502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7517355192775997e-06,
+      "loss": 0.8843,
+      "step": 11503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.750471337367483e-06,
+      "loss": 0.8479,
+      "step": 11504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492075680294929e-06,
+      "loss": 0.8444,
+      "step": 11505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7479442113268308e-06,
+      "loss": 0.8702,
+      "step": 11506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7466812673226775e-06,
+      "loss": 0.891,
+      "step": 11507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7454187360801988e-06,
+      "loss": 0.643,
+      "step": 11508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7441566176625312e-06,
+      "loss": 0.9432,
+      "step": 11509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7428949121328008e-06,
+      "loss": 0.9182,
+      "step": 11510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.741633619554105e-06,
+      "loss": 0.8852,
+      "step": 11511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.740372739989521e-06,
+      "loss": 0.6366,
+      "step": 11512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7391122735021115e-06,
+      "loss": 0.8181,
+      "step": 11513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.737852220154912e-06,
+      "loss": 0.7831,
+      "step": 11514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7365925800109396e-06,
+      "loss": 0.9024,
+      "step": 11515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7353333531331918e-06,
+      "loss": 0.9745,
+      "step": 11516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7340745395846415e-06,
+      "loss": 0.9015,
+      "step": 11517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7328161394282494e-06,
+      "loss": 0.8983,
+      "step": 11518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7315581527269464e-06,
+      "loss": 0.7847,
+      "step": 11519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7303005795436478e-06,
+      "loss": 0.7926,
+      "step": 11520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7290434199412465e-06,
+      "loss": 0.891,
+      "step": 11521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.727786673982612e-06,
+      "loss": 0.9228,
+      "step": 11522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7265303417306023e-06,
+      "loss": 0.9407,
+      "step": 11523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7252744232480457e-06,
+      "loss": 0.9899,
+      "step": 11524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7240189185977507e-06,
+      "loss": 0.9857,
+      "step": 11525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7227638278425096e-06,
+      "loss": 1.01,
+      "step": 11526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7215091510450888e-06,
+      "loss": 0.6073,
+      "step": 11527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7202548882682402e-06,
+      "loss": 0.9062,
+      "step": 11528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7190010395746903e-06,
+      "loss": 0.8449,
+      "step": 11529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7177476050271458e-06,
+      "loss": 0.9355,
+      "step": 11530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164945846882918e-06,
+      "loss": 0.9192,
+      "step": 11531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715241978620793e-06,
+      "loss": 0.7277,
+      "step": 11532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713989786887298e-06,
+      "loss": 0.8196,
+      "step": 11533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127380095504298e-06,
+      "loss": 0.7944,
+      "step": 11534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711486646672791e-06,
+      "loss": 0.8158,
+      "step": 11535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710235698316961e-06,
+      "loss": 0.8321,
+      "step": 11536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.708985164545508e-06,
+      "loss": 0.8929,
+      "step": 11537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7077350454209696e-06,
+      "loss": 0.8682,
+      "step": 11538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7064853410058667e-06,
+      "loss": 0.9664,
+      "step": 11539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7052360513626997e-06,
+      "loss": 0.9033,
+      "step": 11540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7039871765539451e-06,
+      "loss": 0.8559,
+      "step": 11541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7027387166420616e-06,
+      "loss": 0.8272,
+      "step": 11542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7014906716894919e-06,
+      "loss": 0.9568,
+      "step": 11543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.700243041758648e-06,
+      "loss": 0.9715,
+      "step": 11544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6989958269119267e-06,
+      "loss": 0.8862,
+      "step": 11545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6977490272117037e-06,
+      "loss": 0.8659,
+      "step": 11546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6965026427203302e-06,
+      "loss": 0.9627,
+      "step": 11547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6952566735001452e-06,
+      "loss": 0.978,
+      "step": 11548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.69401111961346e-06,
+      "loss": 0.8065,
+      "step": 11549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6927659811225649e-06,
+      "loss": 0.8915,
+      "step": 11550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6915212580897332e-06,
+      "loss": 0.8396,
+      "step": 11551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.690276950577211e-06,
+      "loss": 0.8799,
+      "step": 11552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6890330586472347e-06,
+      "loss": 0.945,
+      "step": 11553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6877895823620105e-06,
+      "loss": 0.8918,
+      "step": 11554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6865465217837263e-06,
+      "loss": 0.789,
+      "step": 11555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6853038769745466e-06,
+      "loss": 0.8441,
+      "step": 11556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6840616479966243e-06,
+      "loss": 0.9354,
+      "step": 11557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6828198349120816e-06,
+      "loss": 0.8438,
+      "step": 11558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6815784377830247e-06,
+      "loss": 0.7892,
+      "step": 11559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.680337456671537e-06,
+      "loss": 0.9549,
+      "step": 11560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6790968916396789e-06,
+      "loss": 0.7952,
+      "step": 11561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6778567427494997e-06,
+      "loss": 0.9063,
+      "step": 11562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6766170100630174e-06,
+      "loss": 0.8879,
+      "step": 11563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6753776936422328e-06,
+      "loss": 0.8567,
+      "step": 11564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6741387935491283e-06,
+      "loss": 0.8849,
+      "step": 11565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6729003098456576e-06,
+      "loss": 0.8772,
+      "step": 11566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6716622425937668e-06,
+      "loss": 0.889,
+      "step": 11567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6704245918553697e-06,
+      "loss": 0.9988,
+      "step": 11568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6691873576923635e-06,
+      "loss": 0.782,
+      "step": 11569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6679505401666253e-06,
+      "loss": 1.0445,
+      "step": 11570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6667141393400067e-06,
+      "loss": 0.902,
+      "step": 11571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6654781552743438e-06,
+      "loss": 0.8943,
+      "step": 11572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.664242588031455e-06,
+      "loss": 0.7997,
+      "step": 11573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6630074376731286e-06,
+      "loss": 0.8915,
+      "step": 11574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6617727042611364e-06,
+      "loss": 0.9141,
+      "step": 11575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6605383878572313e-06,
+      "loss": 0.9306,
+      "step": 11576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.659304488523139e-06,
+      "loss": 0.95,
+      "step": 11577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6580710063205741e-06,
+      "loss": 0.9668,
+      "step": 11578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6568379413112235e-06,
+      "loss": 0.9981,
+      "step": 11579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6556052935567534e-06,
+      "loss": 0.8552,
+      "step": 11580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6543730631188093e-06,
+      "loss": 0.8289,
+      "step": 11581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6531412500590204e-06,
+      "loss": 0.9752,
+      "step": 11582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651909854438991e-06,
+      "loss": 0.9489,
+      "step": 11583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6506788763203042e-06,
+      "loss": 0.6236,
+      "step": 11584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6494483157645235e-06,
+      "loss": 0.8485,
+      "step": 11585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.648218172833188e-06,
+      "loss": 0.8628,
+      "step": 11586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.646988447587825e-06,
+      "loss": 0.846,
+      "step": 11587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645759140089932e-06,
+      "loss": 0.8271,
+      "step": 11588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6445302504009885e-06,
+      "loss": 0.9165,
+      "step": 11589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6433017785824545e-06,
+      "loss": 0.8575,
+      "step": 11590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.642073724695763e-06,
+      "loss": 0.8472,
+      "step": 11591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6408460888023381e-06,
+      "loss": 0.8508,
+      "step": 11592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.639618870963572e-06,
+      "loss": 0.9577,
+      "step": 11593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6383920712408407e-06,
+      "loss": 0.8652,
+      "step": 11594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637165689695498e-06,
+      "loss": 0.9828,
+      "step": 11595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6359397263888753e-06,
+      "loss": 0.903,
+      "step": 11596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6347141813822887e-06,
+      "loss": 0.9393,
+      "step": 11597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6334890547370286e-06,
+      "loss": 0.836,
+      "step": 11598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.632264346514364e-06,
+      "loss": 0.9565,
+      "step": 11599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6310400567755459e-06,
+      "loss": 0.8981,
+      "step": 11600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6298161855817996e-06,
+      "loss": 0.8771,
+      "step": 11601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6285927329943386e-06,
+      "loss": 0.7966,
+      "step": 11602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.627369699074347e-06,
+      "loss": 0.7593,
+      "step": 11603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6261470838829884e-06,
+      "loss": 0.8934,
+      "step": 11604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6249248874814117e-06,
+      "loss": 0.5963,
+      "step": 11605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6237031099307365e-06,
+      "loss": 0.8557,
+      "step": 11606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.622481751292071e-06,
+      "loss": 0.9275,
+      "step": 11607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6212608116264949e-06,
+      "loss": 0.8122,
+      "step": 11608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6200402909950685e-06,
+      "loss": 0.9422,
+      "step": 11609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.618820189458834e-06,
+      "loss": 0.8289,
+      "step": 11610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6176005070788058e-06,
+      "loss": 0.952,
+      "step": 11611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6163812439159888e-06,
+      "loss": 0.8897,
+      "step": 11612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6151624000313559e-06,
+      "loss": 0.8595,
+      "step": 11613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6139439754858654e-06,
+      "loss": 0.8654,
+      "step": 11614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.612725970340453e-06,
+      "loss": 0.8833,
+      "step": 11615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6115083846560286e-06,
+      "loss": 0.8221,
+      "step": 11616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6102912184934915e-06,
+      "loss": 0.9211,
+      "step": 11617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6090744719137108e-06,
+      "loss": 0.8366,
+      "step": 11618
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6078581449775399e-06,
+      "loss": 0.8877,
+      "step": 11619
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6066422377458079e-06,
+      "loss": 0.8393,
+      "step": 11620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.605426750279322e-06,
+      "loss": 0.5949,
+      "step": 11621
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6042116826388766e-06,
+      "loss": 0.8195,
+      "step": 11622
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6029970348852341e-06,
+      "loss": 0.8792,
+      "step": 11623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.601782807079144e-06,
+      "loss": 0.8315,
+      "step": 11624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60056899928133e-06,
+      "loss": 0.7656,
+      "step": 11625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5993556115524945e-06,
+      "loss": 0.8678,
+      "step": 11626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5981426439533254e-06,
+      "loss": 0.7828,
+      "step": 11627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5969300965444844e-06,
+      "loss": 0.851,
+      "step": 11628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.595717969386611e-06,
+      "loss": 0.8149,
+      "step": 11629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.594506262540324e-06,
+      "loss": 0.9496,
+      "step": 11630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5932949760662274e-06,
+      "loss": 0.9644,
+      "step": 11631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5920841100248974e-06,
+      "loss": 0.9375,
+      "step": 11632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5908736644768919e-06,
+      "loss": 0.9438,
+      "step": 11633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5896636394827458e-06,
+      "loss": 0.9864,
+      "step": 11634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588454035102973e-06,
+      "loss": 0.8409,
+      "step": 11635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587244851398071e-06,
+      "loss": 0.8805,
+      "step": 11636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586036088428513e-06,
+      "loss": 0.81,
+      "step": 11637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5848277462547501e-06,
+      "loss": 0.8849,
+      "step": 11638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5836198249372147e-06,
+      "loss": 0.7816,
+      "step": 11639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5824123245363143e-06,
+      "loss": 0.9106,
+      "step": 11640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5812052451124371e-06,
+      "loss": 0.8804,
+      "step": 11641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579998586725956e-06,
+      "loss": 0.9026,
+      "step": 11642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5787923494372159e-06,
+      "loss": 0.797,
+      "step": 11643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577586533306541e-06,
+      "loss": 0.9237,
+      "step": 11644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5763811383942363e-06,
+      "loss": 0.7903,
+      "step": 11645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5751761647605845e-06,
+      "loss": 0.7926,
+      "step": 11646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5739716124658533e-06,
+      "loss": 0.878,
+      "step": 11647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5727674815702798e-06,
+      "loss": 0.8473,
+      "step": 11648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571563772134086e-06,
+      "loss": 0.9538,
+      "step": 11649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570360484217468e-06,
+      "loss": 0.9251,
+      "step": 11650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.56915761788061e-06,
+      "loss": 0.9066,
+      "step": 11651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5679551731836672e-06,
+      "loss": 0.9032,
+      "step": 11652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5667531501867738e-06,
+      "loss": 0.9299,
+      "step": 11653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5655515489500473e-06,
+      "loss": 0.9824,
+      "step": 11654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564350369533577e-06,
+      "loss": 0.8638,
+      "step": 11655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5631496119974431e-06,
+      "loss": 0.8756,
+      "step": 11656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5619492764016919e-06,
+      "loss": 0.9914,
+      "step": 11657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5607493628063574e-06,
+      "loss": 0.8762,
+      "step": 11658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5595498712714463e-06,
+      "loss": 0.8452,
+      "step": 11659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5583508018569472e-06,
+      "loss": 0.8733,
+      "step": 11660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5571521546228307e-06,
+      "loss": 0.8935,
+      "step": 11661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5559539296290404e-06,
+      "loss": 0.7295,
+      "step": 11662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5547561269355028e-06,
+      "loss": 0.8784,
+      "step": 11663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5535587466021207e-06,
+      "loss": 0.8466,
+      "step": 11664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5523617886887755e-06,
+      "loss": 0.6292,
+      "step": 11665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5511652532553312e-06,
+      "loss": 0.9053,
+      "step": 11666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5499691403616302e-06,
+      "loss": 0.9563,
+      "step": 11667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.54877345006749e-06,
+      "loss": 0.8057,
+      "step": 11668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5475781824327096e-06,
+      "loss": 0.8784,
+      "step": 11669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5463833375170657e-06,
+      "loss": 0.8581,
+      "step": 11670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5451889153803123e-06,
+      "loss": 0.9293,
+      "step": 11671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5439949160821878e-06,
+      "loss": 0.9175,
+      "step": 11672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.542801339682406e-06,
+      "loss": 0.8893,
+      "step": 11673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5416081862406573e-06,
+      "loss": 0.8166,
+      "step": 11674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5404154558166117e-06,
+      "loss": 0.8607,
+      "step": 11675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.539223148469925e-06,
+      "loss": 0.8811,
+      "step": 11676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5380312642602236e-06,
+      "loss": 0.8167,
+      "step": 11677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5368398032471154e-06,
+      "loss": 0.9523,
+      "step": 11678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5356487654901864e-06,
+      "loss": 0.9624,
+      "step": 11679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.534458151049002e-06,
+      "loss": 0.8055,
+      "step": 11680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5332679599831091e-06,
+      "loss": 0.8021,
+      "step": 11681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.53207819235203e-06,
+      "loss": 0.9376,
+      "step": 11682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5308888482152673e-06,
+      "loss": 0.6167,
+      "step": 11683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.529699927632301e-06,
+      "loss": 0.9151,
+      "step": 11684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5285114306625892e-06,
+      "loss": 0.9254,
+      "step": 11685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5273233573655744e-06,
+      "loss": 0.8018,
+      "step": 11686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5261357078006733e-06,
+      "loss": 0.9447,
+      "step": 11687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.524948482027281e-06,
+      "loss": 0.8641,
+      "step": 11688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5237616801047727e-06,
+      "loss": 0.8226,
+      "step": 11689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5225753020924994e-06,
+      "loss": 0.8932,
+      "step": 11690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5213893480497988e-06,
+      "loss": 0.6036,
+      "step": 11691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5202038180359813e-06,
+      "loss": 0.8639,
+      "step": 11692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.519018712110335e-06,
+      "loss": 0.9209,
+      "step": 11693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5178340303321315e-06,
+      "loss": 0.7092,
+      "step": 11694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5166497727606133e-06,
+      "loss": 0.961,
+      "step": 11695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5154659394550142e-06,
+      "loss": 0.8784,
+      "step": 11696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5142825304745356e-06,
+      "loss": 0.8758,
+      "step": 11697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5130995458783604e-06,
+      "loss": 0.9656,
+      "step": 11698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5119169857256554e-06,
+      "loss": 0.898,
+      "step": 11699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.510734850075558e-06,
+      "loss": 0.8136,
+      "step": 11700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5095531389871942e-06,
+      "loss": 0.8081,
+      "step": 11701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.50837185251966e-06,
+      "loss": 0.899,
+      "step": 11702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5071909907320337e-06,
+      "loss": 0.7359,
+      "step": 11703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5060105536833713e-06,
+      "loss": 0.7648,
+      "step": 11704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.504830541432707e-06,
+      "loss": 0.9269,
+      "step": 11705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5036509540390597e-06,
+      "loss": 0.967,
+      "step": 11706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5024717915614194e-06,
+      "loss": 0.8467,
+      "step": 11707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5012930540587588e-06,
+      "loss": 0.9159,
+      "step": 11708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5001147415900286e-06,
+      "loss": 0.9986,
+      "step": 11709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4989368542141547e-06,
+      "loss": 0.8169,
+      "step": 11710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49775939199005e-06,
+      "loss": 0.9196,
+      "step": 11711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.496582354976599e-06,
+      "loss": 0.8815,
+      "step": 11712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4954057432326673e-06,
+      "loss": 0.8077,
+      "step": 11713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4942295568171006e-06,
+      "loss": 0.9727,
+      "step": 11714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493053795788717e-06,
+      "loss": 0.9262,
+      "step": 11715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4918784602063242e-06,
+      "loss": 0.8601,
+      "step": 11716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4907035501287005e-06,
+      "loss": 1.0138,
+      "step": 11717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4895290656146043e-06,
+      "loss": 0.8376,
+      "step": 11718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4883550067227748e-06,
+      "loss": 0.8438,
+      "step": 11719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4871813735119245e-06,
+      "loss": 0.9271,
+      "step": 11720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4860081660407533e-06,
+      "loss": 0.8295,
+      "step": 11721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4848353843679352e-06,
+      "loss": 0.9147,
+      "step": 11722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4836630285521202e-06,
+      "loss": 0.8595,
+      "step": 11723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4824910986519391e-06,
+      "loss": 0.8894,
+      "step": 11724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4813195947260051e-06,
+      "loss": 0.9345,
+      "step": 11725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4801485168329066e-06,
+      "loss": 0.9729,
+      "step": 11726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4789778650312104e-06,
+      "loss": 0.8929,
+      "step": 11727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4778076393794605e-06,
+      "loss": 0.7944,
+      "step": 11728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4766378399361824e-06,
+      "loss": 0.9178,
+      "step": 11729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4754684667598818e-06,
+      "loss": 0.9298,
+      "step": 11730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4742995199090405e-06,
+      "loss": 0.8316,
+      "step": 11731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4731309994421205e-06,
+      "loss": 0.8548,
+      "step": 11732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4719629054175587e-06,
+      "loss": 0.6186,
+      "step": 11733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470795237893774e-06,
+      "loss": 0.9245,
+      "step": 11734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4696279969291627e-06,
+      "loss": 0.9263,
+      "step": 11735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4684611825821026e-06,
+      "loss": 0.9226,
+      "step": 11736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4672947949109474e-06,
+      "loss": 0.94,
+      "step": 11737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4661288339740287e-06,
+      "loss": 0.8112,
+      "step": 11738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4649632998296591e-06,
+      "loss": 0.8836,
+      "step": 11739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4637981925361266e-06,
+      "loss": 0.7396,
+      "step": 11740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4626335121517044e-06,
+      "loss": 0.8704,
+      "step": 11741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461469258734637e-06,
+      "loss": 0.909,
+      "step": 11742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4603054323431509e-06,
+      "loss": 0.8598,
+      "step": 11743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4591420330354488e-06,
+      "loss": 0.9166,
+      "step": 11744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579790608697198e-06,
+      "loss": 0.8476,
+      "step": 11745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4568165159041213e-06,
+      "loss": 0.9371,
+      "step": 11746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4556543981967974e-06,
+      "loss": 0.9247,
+      "step": 11747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4544927078058635e-06,
+      "loss": 0.8547,
+      "step": 11748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4533314447894187e-06,
+      "loss": 0.875,
+      "step": 11749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4521706092055421e-06,
+      "loss": 0.6052,
+      "step": 11750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510102011122874e-06,
+      "loss": 0.8172,
+      "step": 11751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449850220567689e-06,
+      "loss": 0.8058,
+      "step": 11752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4486906676297585e-06,
+      "loss": 0.5953,
+      "step": 11753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447531542356485e-06,
+      "loss": 0.8629,
+      "step": 11754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4463728448058434e-06,
+      "loss": 0.8896,
+      "step": 11755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4452145750357782e-06,
+      "loss": 0.7319,
+      "step": 11756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4440567331042176e-06,
+      "loss": 0.9016,
+      "step": 11757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4428993190690676e-06,
+      "loss": 0.9723,
+      "step": 11758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4417423329882085e-06,
+      "loss": 0.9349,
+      "step": 11759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4405857749195062e-06,
+      "loss": 0.8618,
+      "step": 11760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4394296449208035e-06,
+      "loss": 0.9208,
+      "step": 11761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4382739430499193e-06,
+      "loss": 0.6065,
+      "step": 11762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4371186693646522e-06,
+      "loss": 0.9068,
+      "step": 11763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4359638239227768e-06,
+      "loss": 0.8742,
+      "step": 11764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4348094067820495e-06,
+      "loss": 0.876,
+      "step": 11765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4336554180002081e-06,
+      "loss": 0.7792,
+      "step": 11766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4325018576349625e-06,
+      "loss": 0.7502,
+      "step": 11767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4313487257440039e-06,
+      "loss": 0.8206,
+      "step": 11768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430196022385002e-06,
+      "loss": 0.9074,
+      "step": 11769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4290437476156072e-06,
+      "loss": 0.8187,
+      "step": 11770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4278919014934467e-06,
+      "loss": 0.8441,
+      "step": 11771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4267404840761245e-06,
+      "loss": 0.9914,
+      "step": 11772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4255894954212257e-06,
+      "loss": 0.8032,
+      "step": 11773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4244389355863098e-06,
+      "loss": 0.8862,
+      "step": 11774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.423288804628924e-06,
+      "loss": 0.9329,
+      "step": 11775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4221391026065856e-06,
+      "loss": 0.8724,
+      "step": 11776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.420989829576792e-06,
+      "loss": 0.8591,
+      "step": 11777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419840985597022e-06,
+      "loss": 0.9235,
+      "step": 11778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4186925707247267e-06,
+      "loss": 0.821,
+      "step": 11779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4175445850173464e-06,
+      "loss": 0.8924,
+      "step": 11780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4163970285322902e-06,
+      "loss": 0.8692,
+      "step": 11781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4152499013269506e-06,
+      "loss": 0.8874,
+      "step": 11782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4141032034586965e-06,
+      "loss": 0.9856,
+      "step": 11783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4129569349848737e-06,
+      "loss": 0.8372,
+      "step": 11784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.411811095962814e-06,
+      "loss": 0.8914,
+      "step": 11785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4106656864498191e-06,
+      "loss": 0.6259,
+      "step": 11786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4095207065031746e-06,
+      "loss": 0.9511,
+      "step": 11787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4083761561801413e-06,
+      "loss": 0.8768,
+      "step": 11788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4072320355379588e-06,
+      "loss": 0.8549,
+      "step": 11789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4060883446338503e-06,
+      "loss": 0.8023,
+      "step": 11790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4049450835250112e-06,
+      "loss": 0.8961,
+      "step": 11791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4038022522686157e-06,
+      "loss": 0.843,
+      "step": 11792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4026598509218236e-06,
+      "loss": 0.9426,
+      "step": 11793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4015178795417627e-06,
+      "loss": 0.9711,
+      "step": 11794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4003763381855507e-06,
+      "loss": 0.5983,
+      "step": 11795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.399235226910275e-06,
+      "loss": 0.956,
+      "step": 11796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3980945457730033e-06,
+      "loss": 0.797,
+      "step": 11797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3969542948307847e-06,
+      "loss": 0.8753,
+      "step": 11798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395814474140642e-06,
+      "loss": 0.7725,
+      "step": 11799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3946750837595845e-06,
+      "loss": 0.9111,
+      "step": 11800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.393536123744591e-06,
+      "loss": 0.8082,
+      "step": 11801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3923975941526247e-06,
+      "loss": 0.8721,
+      "step": 11802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3912594950406233e-06,
+      "loss": 0.713,
+      "step": 11803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3901218264655048e-06,
+      "loss": 0.8879,
+      "step": 11804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.388984588484168e-06,
+      "loss": 0.793,
+      "step": 11805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3878477811534886e-06,
+      "loss": 0.7486,
+      "step": 11806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3867114045303177e-06,
+      "loss": 0.7327,
+      "step": 11807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3855754586714886e-06,
+      "loss": 0.9651,
+      "step": 11808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3844399436338085e-06,
+      "loss": 0.7997,
+      "step": 11809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3833048594740728e-06,
+      "loss": 0.9615,
+      "step": 11810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3821702062490438e-06,
+      "loss": 0.907,
+      "step": 11811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3810359840154697e-06,
+      "loss": 0.9653,
+      "step": 11812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3799021928300737e-06,
+      "loss": 0.8321,
+      "step": 11813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3787688327495562e-06,
+      "loss": 0.6247,
+      "step": 11814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3776359038306031e-06,
+      "loss": 0.7651,
+      "step": 11815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3765034061298721e-06,
+      "loss": 0.9125,
+      "step": 11816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3753713397040014e-06,
+      "loss": 0.771,
+      "step": 11817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3742397046096034e-06,
+      "loss": 0.8927,
+      "step": 11818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3731085009032797e-06,
+      "loss": 0.8349,
+      "step": 11819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3719777286415991e-06,
+      "loss": 0.8401,
+      "step": 11820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3708473878811158e-06,
+      "loss": 0.8069,
+      "step": 11821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3697174786783585e-06,
+      "loss": 0.9492,
+      "step": 11822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3685880010898345e-06,
+      "loss": 0.9057,
+      "step": 11823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3674589551720308e-06,
+      "loss": 0.7197,
+      "step": 11824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3663303409814178e-06,
+      "loss": 0.8713,
+      "step": 11825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3652021585744358e-06,
+      "loss": 0.9492,
+      "step": 11826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3640744080075063e-06,
+      "loss": 0.861,
+      "step": 11827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3629470893370312e-06,
+      "loss": 0.9423,
+      "step": 11828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3618202026193861e-06,
+      "loss": 0.7893,
+      "step": 11829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.360693747910934e-06,
+      "loss": 0.9277,
+      "step": 11830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3595677252680074e-06,
+      "loss": 0.7618,
+      "step": 11831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3584421347469223e-06,
+      "loss": 0.8884,
+      "step": 11832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3573169764039685e-06,
+      "loss": 0.8916,
+      "step": 11833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3561922502954172e-06,
+      "loss": 0.9566,
+      "step": 11834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3550679564775216e-06,
+      "loss": 0.8474,
+      "step": 11835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3539440950065053e-06,
+      "loss": 0.8552,
+      "step": 11836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3528206659385778e-06,
+      "loss": 0.8542,
+      "step": 11837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.351697669329919e-06,
+      "loss": 0.9093,
+      "step": 11838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3505751052366956e-06,
+      "loss": 0.8508,
+      "step": 11839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3494529737150486e-06,
+      "loss": 0.919,
+      "step": 11840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3483312748210975e-06,
+      "loss": 0.5556,
+      "step": 11841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3472100086109385e-06,
+      "loss": 0.8348,
+      "step": 11842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3460891751406468e-06,
+      "loss": 0.8758,
+      "step": 11843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.344968774466282e-06,
+      "loss": 0.9755,
+      "step": 11844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3438488066438738e-06,
+      "loss": 0.8887,
+      "step": 11845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.342729271729435e-06,
+      "loss": 0.9194,
+      "step": 11846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.341610169778953e-06,
+      "loss": 0.8762,
+      "step": 11847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3404915008483965e-06,
+      "loss": 0.8282,
+      "step": 11848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3393732649937141e-06,
+      "loss": 0.8311,
+      "step": 11849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3382554622708299e-06,
+      "loss": 0.8981,
+      "step": 11850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.337138092735646e-06,
+      "loss": 0.8777,
+      "step": 11851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336021156444044e-06,
+      "loss": 0.9625,
+      "step": 11852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3349046534518806e-06,
+      "loss": 0.895,
+      "step": 11853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3337885838149988e-06,
+      "loss": 0.59,
+      "step": 11854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3326729475892154e-06,
+      "loss": 0.8969,
+      "step": 11855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.331557744830322e-06,
+      "loss": 0.6192,
+      "step": 11856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330442975594093e-06,
+      "loss": 0.9079,
+      "step": 11857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3293286399362803e-06,
+      "loss": 0.9339,
+      "step": 11858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.328214737912611e-06,
+      "loss": 0.8827,
+      "step": 11859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3271012695787954e-06,
+      "loss": 0.8637,
+      "step": 11860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3259882349905207e-06,
+      "loss": 0.8739,
+      "step": 11861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3248756342034507e-06,
+      "loss": 0.8327,
+      "step": 11862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3237634672732247e-06,
+      "loss": 0.8506,
+      "step": 11863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32265173425547e-06,
+      "loss": 0.8412,
+      "step": 11864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3215404352057837e-06,
+      "loss": 0.5765,
+      "step": 11865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.320429570179742e-06,
+      "loss": 0.8498,
+      "step": 11866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.319319139232903e-06,
+      "loss": 0.8543,
+      "step": 11867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3182091424207988e-06,
+      "loss": 0.8063,
+      "step": 11868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3170995797989449e-06,
+      "loss": 0.913,
+      "step": 11869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3159904514228316e-06,
+      "loss": 0.8512,
+      "step": 11870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3148817573479278e-06,
+      "loss": 0.7758,
+      "step": 11871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3137734976296812e-06,
+      "loss": 0.6258,
+      "step": 11872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3126656723235155e-06,
+      "loss": 0.9489,
+      "step": 11873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3115582814848383e-06,
+      "loss": 0.6102,
+      "step": 11874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3104513251690298e-06,
+      "loss": 0.9715,
+      "step": 11875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3093448034314526e-06,
+      "loss": 0.9005,
+      "step": 11876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3082387163274435e-06,
+      "loss": 0.8763,
+      "step": 11877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071330639123181e-06,
+      "loss": 0.8575,
+      "step": 11878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3060278462413767e-06,
+      "loss": 0.8568,
+      "step": 11879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3049230633698905e-06,
+      "loss": 0.8688,
+      "step": 11880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3038187153531123e-06,
+      "loss": 0.917,
+      "step": 11881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.302714802246271e-06,
+      "loss": 0.826,
+      "step": 11882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3016113241045736e-06,
+      "loss": 0.5707,
+      "step": 11883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3005082809832103e-06,
+      "loss": 0.8255,
+      "step": 11884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299405672937345e-06,
+      "loss": 0.9653,
+      "step": 11885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2983035000221178e-06,
+      "loss": 0.7699,
+      "step": 11886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297201762292656e-06,
+      "loss": 0.9465,
+      "step": 11887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2961004598040528e-06,
+      "loss": 0.75,
+      "step": 11888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2949995926113923e-06,
+      "loss": 0.7929,
+      "step": 11889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2938991607697271e-06,
+      "loss": 0.8741,
+      "step": 11890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292799164334092e-06,
+      "loss": 0.8457,
+      "step": 11891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2916996033594998e-06,
+      "loss": 0.9036,
+      "step": 11892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2906004779009395e-06,
+      "loss": 0.9351,
+      "step": 11893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2895017880133831e-06,
+      "loss": 0.8714,
+      "step": 11894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2884035337517776e-06,
+      "loss": 0.9924,
+      "step": 11895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2873057151710467e-06,
+      "loss": 0.8448,
+      "step": 11896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2862083323260944e-06,
+      "loss": 0.7554,
+      "step": 11897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2851113852718011e-06,
+      "loss": 0.8259,
+      "step": 11898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.284014874063032e-06,
+      "loss": 0.8721,
+      "step": 11899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.282918798754621e-06,
+      "loss": 0.8866,
+      "step": 11900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2818231594013852e-06,
+      "loss": 0.6206,
+      "step": 11901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.280727956058121e-06,
+      "loss": 0.8859,
+      "step": 11902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2796331887795976e-06,
+      "loss": 0.828,
+      "step": 11903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785388576205704e-06,
+      "loss": 0.7312,
+      "step": 11904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2774449626357665e-06,
+      "loss": 0.8237,
+      "step": 11905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2763515038798935e-06,
+      "loss": 0.9024,
+      "step": 11906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2752584814076352e-06,
+      "loss": 0.8333,
+      "step": 11907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.274165895273659e-06,
+      "loss": 0.9065,
+      "step": 11908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273073745532606e-06,
+      "loss": 0.9502,
+      "step": 11909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2719820322390953e-06,
+      "loss": 0.7964,
+      "step": 11910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2708907554477257e-06,
+      "loss": 0.9194,
+      "step": 11911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2697999152130702e-06,
+      "loss": 0.8927,
+      "step": 11912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2687095115896896e-06,
+      "loss": 0.8555,
+      "step": 11913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.267619544632115e-06,
+      "loss": 0.8894,
+      "step": 11914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2665300143948545e-06,
+      "loss": 0.9325,
+      "step": 11915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2654409209324003e-06,
+      "loss": 0.9172,
+      "step": 11916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2643522642992156e-06,
+      "loss": 0.8685,
+      "step": 11917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263264044549748e-06,
+      "loss": 0.7749,
+      "step": 11918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2621762617384248e-06,
+      "loss": 0.8435,
+      "step": 11919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2610889159196449e-06,
+      "loss": 0.7768,
+      "step": 11920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600020071477882e-06,
+      "loss": 0.9439,
+      "step": 11921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.258915535477212e-06,
+      "loss": 0.8278,
+      "step": 11922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.257829500962251e-06,
+      "loss": 0.9008,
+      "step": 11923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2567439036572238e-06,
+      "loss": 0.6182,
+      "step": 11924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2556587436164202e-06,
+      "loss": 0.9053,
+      "step": 11925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545740208941115e-06,
+      "loss": 0.854,
+      "step": 11926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2534897355445463e-06,
+      "loss": 0.9031,
+      "step": 11927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2524058876219491e-06,
+      "loss": 0.6134,
+      "step": 11928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25132247718053e-06,
+      "loss": 0.7759,
+      "step": 11929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2502395042744686e-06,
+      "loss": 0.9239,
+      "step": 11930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2491569689579263e-06,
+      "loss": 0.6286,
+      "step": 11931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248074871285041e-06,
+      "loss": 0.8207,
+      "step": 11932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2469932113099337e-06,
+      "loss": 0.7786,
+      "step": 11933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2459119890866989e-06,
+      "loss": 0.9739,
+      "step": 11934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244831204669409e-06,
+      "loss": 0.895,
+      "step": 11935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2437508581121161e-06,
+      "loss": 0.8842,
+      "step": 11936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2426709494688484e-06,
+      "loss": 0.8848,
+      "step": 11937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2415914787936179e-06,
+      "loss": 0.8115,
+      "step": 11938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2405124461404083e-06,
+      "loss": 0.8281,
+      "step": 11939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2394338515631844e-06,
+      "loss": 0.894,
+      "step": 11940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238355695115887e-06,
+      "loss": 0.6412,
+      "step": 11941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2372779768524357e-06,
+      "loss": 0.8674,
+      "step": 11942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2362006968267326e-06,
+      "loss": 0.814,
+      "step": 11943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2351238550926526e-06,
+      "loss": 0.8613,
+      "step": 11944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2340474517040502e-06,
+      "loss": 0.9466,
+      "step": 11945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2329714867147569e-06,
+      "loss": 0.7954,
+      "step": 11946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2318959601785829e-06,
+      "loss": 0.9765,
+      "step": 11947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2308208721493176e-06,
+      "loss": 0.7991,
+      "step": 11948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2297462226807312e-06,
+      "loss": 0.7471,
+      "step": 11949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.228672011826566e-06,
+      "loss": 0.8818,
+      "step": 11950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2275982396405451e-06,
+      "loss": 0.8915,
+      "step": 11951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2265249061763696e-06,
+      "loss": 0.8616,
+      "step": 11952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2254520114877155e-06,
+      "loss": 0.9995,
+      "step": 11953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2243795556282455e-06,
+      "loss": 0.8859,
+      "step": 11954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2233075386515937e-06,
+      "loss": 0.8784,
+      "step": 11955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.222235960611371e-06,
+      "loss": 0.9857,
+      "step": 11956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2211648215611683e-06,
+      "loss": 0.8028,
+      "step": 11957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2200941215545581e-06,
+      "loss": 0.7962,
+      "step": 11958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2190238606450855e-06,
+      "loss": 0.8653,
+      "step": 11959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2179540388862776e-06,
+      "loss": 0.75,
+      "step": 11960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2168846563316373e-06,
+      "loss": 0.8938,
+      "step": 11961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2158157130346426e-06,
+      "loss": 1.0572,
+      "step": 11962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2147472090487589e-06,
+      "loss": 0.7799,
+      "step": 11963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.213679144427421e-06,
+      "loss": 0.8469,
+      "step": 11964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.212611519224045e-06,
+      "loss": 0.9015,
+      "step": 11965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.211544333492023e-06,
+      "loss": 0.8683,
+      "step": 11966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2104775872847274e-06,
+      "loss": 1.0275,
+      "step": 11967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2094112806555091e-06,
+      "loss": 0.9903,
+      "step": 11968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2083454136576955e-06,
+      "loss": 0.8828,
+      "step": 11969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2072799863445928e-06,
+      "loss": 0.8732,
+      "step": 11970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2062149987694826e-06,
+      "loss": 0.7927,
+      "step": 11971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.205150450985626e-06,
+      "loss": 0.9894,
+      "step": 11972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2040863430462668e-06,
+      "loss": 0.8622,
+      "step": 11973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2030226750046203e-06,
+      "loss": 0.8331,
+      "step": 11974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2019594469138829e-06,
+      "loss": 0.6176,
+      "step": 11975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2008966588272276e-06,
+      "loss": 0.9088,
+      "step": 11976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.199834310797805e-06,
+      "loss": 0.8978,
+      "step": 11977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1987724028787472e-06,
+      "loss": 0.8646,
+      "step": 11978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.197710935123162e-06,
+      "loss": 0.7912,
+      "step": 11979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1966499075841309e-06,
+      "loss": 0.8686,
+      "step": 11980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1955893203147228e-06,
+      "loss": 0.8227,
+      "step": 11981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1945291733679765e-06,
+      "loss": 0.7938,
+      "step": 11982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1934694667969139e-06,
+      "loss": 0.9294,
+      "step": 11983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1924102006545303e-06,
+      "loss": 0.9094,
+      "step": 11984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1913513749938034e-06,
+      "loss": 0.8344,
+      "step": 11985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.190292989867684e-06,
+      "loss": 0.91,
+      "step": 11986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1892350453291046e-06,
+      "loss": 0.9138,
+      "step": 11987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.188177541430977e-06,
+      "loss": 0.8552,
+      "step": 11988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.187120478226187e-06,
+      "loss": 0.9118,
+      "step": 11989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1860638557675984e-06,
+      "loss": 0.6067,
+      "step": 11990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1850076741080585e-06,
+      "loss": 0.8484,
+      "step": 11991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1839519333003825e-06,
+      "loss": 0.7418,
+      "step": 11992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1828966333973757e-06,
+      "loss": 0.6328,
+      "step": 11993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1818417744518141e-06,
+      "loss": 0.8199,
+      "step": 11994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1807873565164507e-06,
+      "loss": 0.9384,
+      "step": 11995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1797333796440203e-06,
+      "loss": 0.888,
+      "step": 11996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1786798438872315e-06,
+      "loss": 0.7707,
+      "step": 11997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177626749298777e-06,
+      "loss": 0.9315,
+      "step": 11998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176574095931322e-06,
+      "loss": 0.9044,
+      "step": 11999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1755218838375115e-06,
+      "loss": 0.6266,
+      "step": 12000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1744701130699665e-06,
+      "loss": 0.8556,
+      "step": 12001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1734187836812916e-06,
+      "loss": 0.8769,
+      "step": 12002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1723678957240625e-06,
+      "loss": 0.8914,
+      "step": 12003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1713174492508372e-06,
+      "loss": 0.8179,
+      "step": 12004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1702674443141494e-06,
+      "loss": 0.6219,
+      "step": 12005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1692178809665089e-06,
+      "loss": 0.8074,
+      "step": 12006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1681687592604118e-06,
+      "loss": 0.5764,
+      "step": 12007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.167120079248324e-06,
+      "loss": 0.8961,
+      "step": 12008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1660718409826898e-06,
+      "loss": 0.8185,
+      "step": 12009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1650240445159354e-06,
+      "loss": 0.9152,
+      "step": 12010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.16397668990046e-06,
+      "loss": 0.8629,
+      "step": 12011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.162929777188645e-06,
+      "loss": 0.8672,
+      "step": 12012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1618833064328506e-06,
+      "loss": 0.8974,
+      "step": 12013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1608372776854104e-06,
+      "loss": 0.8764,
+      "step": 12014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597916909986383e-06,
+      "loss": 0.859,
+      "step": 12015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1587465464248248e-06,
+      "loss": 0.9656,
+      "step": 12016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1577018440162392e-06,
+      "loss": 0.8985,
+      "step": 12017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1566575838251304e-06,
+      "loss": 0.9164,
+      "step": 12018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1556137659037237e-06,
+      "loss": 0.9558,
+      "step": 12019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1545703903042204e-06,
+      "loss": 0.8822,
+      "step": 12020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1535274570788014e-06,
+      "loss": 0.7536,
+      "step": 12021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1524849662796245e-06,
+      "loss": 0.9026,
+      "step": 12022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1514429179588305e-06,
+      "loss": 0.6372,
+      "step": 12023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1504013121685309e-06,
+      "loss": 0.9701,
+      "step": 12024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1493601489608185e-06,
+      "loss": 0.8785,
+      "step": 12025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1483194283877607e-06,
+      "loss": 0.7636,
+      "step": 12026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1472791505014113e-06,
+      "loss": 0.9202,
+      "step": 12027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146239315353793e-06,
+      "loss": 0.9501,
+      "step": 12028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1451999229969114e-06,
+      "loss": 0.8724,
+      "step": 12029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1441609734827452e-06,
+      "loss": 0.8152,
+      "step": 12030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1431224668632535e-06,
+      "loss": 0.914,
+      "step": 12031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1420844031903778e-06,
+      "loss": 0.8229,
+      "step": 12032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141046782516032e-06,
+      "loss": 0.906,
+      "step": 12033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1400096048921072e-06,
+      "loss": 0.7934,
+      "step": 12034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1389728703704762e-06,
+      "loss": 0.8938,
+      "step": 12035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1379365790029851e-06,
+      "loss": 0.9353,
+      "step": 12036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1369007308414647e-06,
+      "loss": 0.8171,
+      "step": 12037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1358653259377162e-06,
+      "loss": 0.8049,
+      "step": 12038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1348303643435233e-06,
+      "loss": 0.9366,
+      "step": 12039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1337958461106458e-06,
+      "loss": 0.7522,
+      "step": 12040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132761771290819e-06,
+      "loss": 0.6952,
+      "step": 12041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317281399357626e-06,
+      "loss": 0.8747,
+      "step": 12042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1306949520971677e-06,
+      "loss": 0.8414,
+      "step": 12043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1296622078267084e-06,
+      "loss": 0.8747,
+      "step": 12044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1286299071760309e-06,
+      "loss": 0.7902,
+      "step": 12045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1275980501967642e-06,
+      "loss": 0.7895,
+      "step": 12046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265666369405105e-06,
+      "loss": 1.0205,
+      "step": 12047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1255356674588557e-06,
+      "loss": 0.8835,
+      "step": 12048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1245051418033592e-06,
+      "loss": 0.6252,
+      "step": 12049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234750600255585e-06,
+      "loss": 0.8656,
+      "step": 12050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1224454221769688e-06,
+      "loss": 0.7959,
+      "step": 12051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214162283090857e-06,
+      "loss": 0.8487,
+      "step": 12052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1203874784733805e-06,
+      "loss": 0.9058,
+      "step": 12053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1193591727213027e-06,
+      "loss": 0.9158,
+      "step": 12054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1183313111042792e-06,
+      "loss": 0.8277,
+      "step": 12055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1173038936737123e-06,
+      "loss": 0.8364,
+      "step": 12056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1162769204809897e-06,
+      "loss": 0.9212,
+      "step": 12057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1152503915774703e-06,
+      "loss": 0.9203,
+      "step": 12058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1142243070144909e-06,
+      "loss": 0.7917,
+      "step": 12059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1131986668433691e-06,
+      "loss": 0.8038,
+      "step": 12060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1121734711153965e-06,
+      "loss": 0.8417,
+      "step": 12061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1111487198818482e-06,
+      "loss": 0.8269,
+      "step": 12062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1101244131939715e-06,
+      "loss": 0.8552,
+      "step": 12063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.109100551102995e-06,
+      "loss": 0.8554,
+      "step": 12064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1080771336601226e-06,
+      "loss": 0.8548,
+      "step": 12065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1070541609165353e-06,
+      "loss": 0.7941,
+      "step": 12066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1060316329233966e-06,
+      "loss": 0.8557,
+      "step": 12067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1050095497318447e-06,
+      "loss": 0.8454,
+      "step": 12068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103987911392994e-06,
+      "loss": 0.9289,
+      "step": 12069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1029667179579396e-06,
+      "loss": 0.8009,
+      "step": 12070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10194596947775e-06,
+      "loss": 0.9104,
+      "step": 12071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1009256660034795e-06,
+      "loss": 0.8905,
+      "step": 12072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0999058075861513e-06,
+      "loss": 0.6262,
+      "step": 12073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.098886394276769e-06,
+      "loss": 0.8218,
+      "step": 12074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.09786742612632e-06,
+      "loss": 0.8137,
+      "step": 12075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.09684890318576e-06,
+      "loss": 0.8621,
+      "step": 12076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0958308255060313e-06,
+      "loss": 0.8241,
+      "step": 12077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0948131931380456e-06,
+      "loss": 0.9538,
+      "step": 12078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093796006132699e-06,
+      "loss": 0.8873,
+      "step": 12079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0927792645408607e-06,
+      "loss": 0.883,
+      "step": 12080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.091762968413379e-06,
+      "loss": 0.9003,
+      "step": 12081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907471178010843e-06,
+      "loss": 0.8201,
+      "step": 12082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0897317127547768e-06,
+      "loss": 0.8191,
+      "step": 12083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0887167533252418e-06,
+      "loss": 0.8204,
+      "step": 12084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0877022395632364e-06,
+      "loss": 0.9717,
+      "step": 12085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0866881715194977e-06,
+      "loss": 0.8771,
+      "step": 12086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085674549244744e-06,
+      "loss": 0.82,
+      "step": 12087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.084661372789666e-06,
+      "loss": 0.859,
+      "step": 12088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0836486422049353e-06,
+      "loss": 0.9422,
+      "step": 12089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0826363575412001e-06,
+      "loss": 0.8484,
+      "step": 12090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0816245188490825e-06,
+      "loss": 0.8767,
+      "step": 12091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.080613126179193e-06,
+      "loss": 0.8883,
+      "step": 12092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0796021795821088e-06,
+      "loss": 0.8282,
+      "step": 12093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0785916791083894e-06,
+      "loss": 0.8893,
+      "step": 12094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0775816248085701e-06,
+      "loss": 0.8953,
+      "step": 12095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0765720167331683e-06,
+      "loss": 0.9032,
+      "step": 12096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0755628549326757e-06,
+      "loss": 0.7937,
+      "step": 12097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0745541394575609e-06,
+      "loss": 0.9561,
+      "step": 12098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0735458703582724e-06,
+      "loss": 0.9486,
+      "step": 12099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.072538047685232e-06,
+      "loss": 0.866,
+      "step": 12100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0715306714888473e-06,
+      "loss": 0.9959,
+      "step": 12101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.070523741819497e-06,
+      "loss": 0.9134,
+      "step": 12102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0695172587275382e-06,
+      "loss": 0.9285,
+      "step": 12103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0685112222633077e-06,
+      "loss": 0.9325,
+      "step": 12104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0675056324771172e-06,
+      "loss": 0.8329,
+      "step": 12105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0665004894192587e-06,
+      "loss": 0.9136,
+      "step": 12106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0654957931400045e-06,
+      "loss": 0.6428,
+      "step": 12107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0644915436895986e-06,
+      "loss": 0.9349,
+      "step": 12108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0634877411182642e-06,
+      "loss": 0.8578,
+      "step": 12109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0624843854762034e-06,
+      "loss": 0.8813,
+      "step": 12110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0614814768135938e-06,
+      "loss": 0.8551,
+      "step": 12111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0604790151805976e-06,
+      "loss": 0.9642,
+      "step": 12112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0594770006273458e-06,
+      "loss": 0.9417,
+      "step": 12113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0584754332039516e-06,
+      "loss": 0.6386,
+      "step": 12114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0574743129605047e-06,
+      "loss": 0.8644,
+      "step": 12115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0564736399470709e-06,
+      "loss": 0.8169,
+      "step": 12116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0554734142137003e-06,
+      "loss": 0.9418,
+      "step": 12117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0544736358104113e-06,
+      "loss": 0.874,
+      "step": 12118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0534743047872065e-06,
+      "loss": 0.856,
+      "step": 12119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0524754211940624e-06,
+      "loss": 0.7905,
+      "step": 12120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.051476985080937e-06,
+      "loss": 0.9407,
+      "step": 12121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0504789964977635e-06,
+      "loss": 0.8335,
+      "step": 12122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.049481455494451e-06,
+      "loss": 0.9536,
+      "step": 12123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0484843621208907e-06,
+      "loss": 0.9069,
+      "step": 12124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0474877164269436e-06,
+      "loss": 0.572,
+      "step": 12125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0464915184624603e-06,
+      "loss": 0.8009,
+      "step": 12126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0454957682772581e-06,
+      "loss": 0.8267,
+      "step": 12127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0445004659211388e-06,
+      "loss": 0.7727,
+      "step": 12128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0435056114438757e-06,
+      "loss": 0.8746,
+      "step": 12129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0425112048952236e-06,
+      "loss": 0.8381,
+      "step": 12130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0415172463249178e-06,
+      "loss": 0.8901,
+      "step": 12131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0405237357826648e-06,
+      "loss": 0.8967,
+      "step": 12132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395306733181531e-06,
+      "loss": 0.7945,
+      "step": 12133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0385380589810457e-06,
+      "loss": 0.8599,
+      "step": 12134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0375458928209848e-06,
+      "loss": 0.8933,
+      "step": 12135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0365541748875918e-06,
+      "loss": 0.6691,
+      "step": 12136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0355629052304627e-06,
+      "loss": 0.8275,
+      "step": 12137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0345720838991747e-06,
+      "loss": 0.9802,
+      "step": 12138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0335817109432788e-06,
+      "loss": 0.7542,
+      "step": 12139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0325917864123046e-06,
+      "loss": 0.7557,
+      "step": 12140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0316023103557593e-06,
+      "loss": 0.9536,
+      "step": 12141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0306132828231318e-06,
+      "loss": 0.8984,
+      "step": 12142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0296247038638806e-06,
+      "loss": 0.9493,
+      "step": 12143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0286365735274495e-06,
+      "loss": 0.8472,
+      "step": 12144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0276488918632533e-06,
+      "loss": 0.8113,
+      "step": 12145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0266616589206902e-06,
+      "loss": 0.8692,
+      "step": 12146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.025674874749132e-06,
+      "loss": 0.8039,
+      "step": 12147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.02468853939793e-06,
+      "loss": 0.9259,
+      "step": 12148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0237026529164119e-06,
+      "loss": 0.6416,
+      "step": 12149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0227172153538822e-06,
+      "loss": 0.8484,
+      "step": 12150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0217322267596275e-06,
+      "loss": 0.9503,
+      "step": 12151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0207476871829059e-06,
+      "loss": 0.8296,
+      "step": 12152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019763596672958e-06,
+      "loss": 0.9892,
+      "step": 12153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0187799552789967e-06,
+      "loss": 0.8709,
+      "step": 12154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0177967630502172e-06,
+      "loss": 0.871,
+      "step": 12155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0168140200357911e-06,
+      "loss": 0.8127,
+      "step": 12156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015831726284867e-06,
+      "loss": 0.8257,
+      "step": 12157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0148498818465703e-06,
+      "loss": 0.9419,
+      "step": 12158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0138684867700044e-06,
+      "loss": 0.8049,
+      "step": 12159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0128875411042482e-06,
+      "loss": 0.8783,
+      "step": 12160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0119070448983648e-06,
+      "loss": 0.8564,
+      "step": 12161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010926998201389e-06,
+      "loss": 0.9452,
+      "step": 12162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0099474010623344e-06,
+      "loss": 0.572,
+      "step": 12163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0089682535301915e-06,
+      "loss": 0.895,
+      "step": 12164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0079895556539266e-06,
+      "loss": 0.8664,
+      "step": 12165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0070113074824917e-06,
+      "loss": 0.8355,
+      "step": 12166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0060335090648066e-06,
+      "loss": 0.889,
+      "step": 12167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.005056160449771e-06,
+      "loss": 0.8481,
+      "step": 12168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0040792616862694e-06,
+      "loss": 0.8722,
+      "step": 12169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0031028128231523e-06,
+      "loss": 0.7816,
+      "step": 12170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0021268139092578e-06,
+      "loss": 0.9781,
+      "step": 12171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011512649933952e-06,
+      "loss": 0.7291,
+      "step": 12172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0001761661243537e-06,
+      "loss": 0.8318,
+      "step": 12173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992015173508995e-07,
+      "loss": 0.8371,
+      "step": 12174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.98227318721774e-07,
+      "loss": 0.8638,
+      "step": 12175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.972535702857022e-07,
+      "loss": 0.7732,
+      "step": 12176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.962802720913812e-07,
+      "loss": 0.8286,
+      "step": 12177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.953074241874872e-07,
+      "loss": 0.8507,
+      "step": 12178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.943350266226738e-07,
+      "loss": 0.8943,
+      "step": 12179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.933630794455695e-07,
+      "loss": 0.8498,
+      "step": 12180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.923915827047881e-07,
+      "loss": 0.9049,
+      "step": 12181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.914205364489138e-07,
+      "loss": 0.7908,
+      "step": 12182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.904499407265088e-07,
+      "loss": 0.8945,
+      "step": 12183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.894797955861158e-07,
+      "loss": 0.9159,
+      "step": 12184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.8851010107625e-07,
+      "loss": 0.8704,
+      "step": 12185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.875408572454114e-07,
+      "loss": 0.9002,
+      "step": 12186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.865720641420717e-07,
+      "loss": 0.8837,
+      "step": 12187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85603721814683e-07,
+      "loss": 0.9353,
+      "step": 12188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.846358303116698e-07,
+      "loss": 0.8439,
+      "step": 12189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.836683896814425e-07,
+      "loss": 0.8363,
+      "step": 12190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827013999723833e-07,
+      "loss": 0.873,
+      "step": 12191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.817348612328515e-07,
+      "loss": 0.8066,
+      "step": 12192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.80768773511186e-07,
+      "loss": 0.7915,
+      "step": 12193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.798031368557016e-07,
+      "loss": 0.8704,
+      "step": 12194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.788379513146928e-07,
+      "loss": 0.8625,
+      "step": 12195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77873216936429e-07,
+      "loss": 0.7834,
+      "step": 12196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.769089337691585e-07,
+      "loss": 0.9046,
+      "step": 12197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.759451018611067e-07,
+      "loss": 0.9294,
+      "step": 12198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74981721260475e-07,
+      "loss": 0.9253,
+      "step": 12199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.740187920154443e-07,
+      "loss": 0.8126,
+      "step": 12200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.730563141741745e-07,
+      "loss": 0.8571,
+      "step": 12201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720942877847994e-07,
+      "loss": 0.8737,
+      "step": 12202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.711327128954307e-07,
+      "loss": 0.9843,
+      "step": 12203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.701715895541586e-07,
+      "loss": 0.8818,
+      "step": 12204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69210917809048e-07,
+      "loss": 0.9165,
+      "step": 12205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.682506977081497e-07,
+      "loss": 0.8715,
+      "step": 12206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.672909292994804e-07,
+      "loss": 0.8212,
+      "step": 12207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.663316126310417e-07,
+      "loss": 0.7654,
+      "step": 12208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.653727477508113e-07,
+      "loss": 0.8407,
+      "step": 12209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.644143347067392e-07,
+      "loss": 0.9222,
+      "step": 12210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.63456373546764e-07,
+      "loss": 0.8159,
+      "step": 12211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.624988643187904e-07,
+      "loss": 0.9136,
+      "step": 12212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61541807070706e-07,
+      "loss": 0.8977,
+      "step": 12213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.605852018503736e-07,
+      "loss": 0.9254,
+      "step": 12214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59629048705637e-07,
+      "loss": 0.8671,
+      "step": 12215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586733476843135e-07,
+      "loss": 0.8453,
+      "step": 12216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.577180988341995e-07,
+      "loss": 0.795,
+      "step": 12217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.567633022030687e-07,
+      "loss": 0.8447,
+      "step": 12218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558089578386698e-07,
+      "loss": 0.8558,
+      "step": 12219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.548550657887345e-07,
+      "loss": 1.0369,
+      "step": 12220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.53901626100967e-07,
+      "loss": 0.844,
+      "step": 12221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.529486388230514e-07,
+      "loss": 0.8857,
+      "step": 12222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.519961040026459e-07,
+      "loss": 1.0479,
+      "step": 12223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.510440216873872e-07,
+      "loss": 0.798,
+      "step": 12224
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.50092391924895e-07,
+      "loss": 1.0031,
+      "step": 12225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.491412147627599e-07,
+      "loss": 0.8668,
+      "step": 12226
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.481904902485517e-07,
+      "loss": 0.9276,
+      "step": 12227
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.472402184298168e-07,
+      "loss": 0.6106,
+      "step": 12228
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.462903993540784e-07,
+      "loss": 0.8168,
+      "step": 12229
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.453410330688429e-07,
+      "loss": 0.8969,
+      "step": 12230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443921196215843e-07,
+      "loss": 0.8924,
+      "step": 12231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.43443659059764e-07,
+      "loss": 0.8576,
+      "step": 12232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.424956514308148e-07,
+      "loss": 0.9116,
+      "step": 12233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.415480967821478e-07,
+      "loss": 0.8227,
+      "step": 12234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.406009951611494e-07,
+      "loss": 0.8041,
+      "step": 12235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.396543466151887e-07,
+      "loss": 0.8324,
+      "step": 12236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.387081511916085e-07,
+      "loss": 0.6068,
+      "step": 12237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.377624089377291e-07,
+      "loss": 0.6332,
+      "step": 12238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.368171199008469e-07,
+      "loss": 0.8934,
+      "step": 12239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.358722841282408e-07,
+      "loss": 0.8992,
+      "step": 12240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.349279016671619e-07,
+      "loss": 0.9847,
+      "step": 12241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.339839725648403e-07,
+      "loss": 0.8439,
+      "step": 12242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.330404968684825e-07,
+      "loss": 0.9194,
+      "step": 12243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.32097474625272e-07,
+      "loss": 0.9322,
+      "step": 12244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.311549058823755e-07,
+      "loss": 0.7937,
+      "step": 12245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.302127906869296e-07,
+      "loss": 0.8545,
+      "step": 12246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.292711290860501e-07,
+      "loss": 0.868,
+      "step": 12247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.283299211268337e-07,
+      "loss": 0.873,
+      "step": 12248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.273891668563462e-07,
+      "loss": 0.8246,
+      "step": 12249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.264488663216431e-07,
+      "loss": 0.8916,
+      "step": 12250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.255090195697469e-07,
+      "loss": 0.8772,
+      "step": 12251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.245696266476611e-07,
+      "loss": 0.8878,
+      "step": 12252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.23630687602367e-07,
+      "loss": 0.906,
+      "step": 12253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.226922024808194e-07,
+      "loss": 0.6193,
+      "step": 12254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.217541713299571e-07,
+      "loss": 0.8352,
+      "step": 12255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.208165941966917e-07,
+      "loss": 0.8635,
+      "step": 12256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.198794711279124e-07,
+      "loss": 0.7513,
+      "step": 12257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.189428021704849e-07,
+      "loss": 1.0537,
+      "step": 12258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.180065873712562e-07,
+      "loss": 0.8765,
+      "step": 12259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.170708267770468e-07,
+      "loss": 0.8654,
+      "step": 12260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.161355204346556e-07,
+      "loss": 0.8898,
+      "step": 12261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.152006683908565e-07,
+      "loss": 0.8241,
+      "step": 12262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.142662706924077e-07,
+      "loss": 0.8322,
+      "step": 12263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.13332327386035e-07,
+      "loss": 0.8817,
+      "step": 12264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.123988385184501e-07,
+      "loss": 0.9091,
+      "step": 12265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.114658041363366e-07,
+      "loss": 0.9393,
+      "step": 12266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.105332242863574e-07,
+      "loss": 0.8724,
+      "step": 12267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.096010990151527e-07,
+      "loss": 0.8842,
+      "step": 12268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.086694283693365e-07,
+      "loss": 0.8032,
+      "step": 12269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.077382123955081e-07,
+      "loss": 0.8569,
+      "step": 12270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.06807451140237e-07,
+      "loss": 0.9138,
+      "step": 12271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.058771446500702e-07,
+      "loss": 0.9311,
+      "step": 12272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.049472929715363e-07,
+      "loss": 0.5807,
+      "step": 12273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.040178961511359e-07,
+      "loss": 0.9243,
+      "step": 12274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.030889542353539e-07,
+      "loss": 0.9575,
+      "step": 12275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.021604672706452e-07,
+      "loss": 0.8468,
+      "step": 12276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.012324353034452e-07,
+      "loss": 0.9277,
+      "step": 12277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.003048583801665e-07,
+      "loss": 0.8949,
+      "step": 12278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.993777365471978e-07,
+      "loss": 0.8838,
+      "step": 12279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.984510698509075e-07,
+      "loss": 0.9247,
+      "step": 12280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.975248583376395e-07,
+      "loss": 0.8771,
+      "step": 12281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.965991020537145e-07,
+      "loss": 0.8741,
+      "step": 12282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.956738010454302e-07,
+      "loss": 0.9193,
+      "step": 12283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.947489553590638e-07,
+      "loss": 0.8993,
+      "step": 12284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.938245650408694e-07,
+      "loss": 0.8933,
+      "step": 12285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.929006301370746e-07,
+      "loss": 0.8967,
+      "step": 12286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.919771506938879e-07,
+      "loss": 0.8307,
+      "step": 12287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.910541267574935e-07,
+      "loss": 0.79,
+      "step": 12288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901315583740544e-07,
+      "loss": 0.9805,
+      "step": 12289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.892094455897104e-07,
+      "loss": 0.8925,
+      "step": 12290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.882877884505759e-07,
+      "loss": 0.8949,
+      "step": 12291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87366587002746e-07,
+      "loss": 0.8678,
+      "step": 12292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.864458412922872e-07,
+      "loss": 0.9422,
+      "step": 12293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.855255513652528e-07,
+      "loss": 0.8464,
+      "step": 12294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.84605717267667e-07,
+      "loss": 0.9213,
+      "step": 12295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.836863390455308e-07,
+      "loss": 0.9698,
+      "step": 12296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827674167448252e-07,
+      "loss": 0.8334,
+      "step": 12297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.818489504115057e-07,
+      "loss": 0.9249,
+      "step": 12298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.809309400915044e-07,
+      "loss": 0.9354,
+      "step": 12299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.800133858307369e-07,
+      "loss": 0.9939,
+      "step": 12300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.790962876750886e-07,
+      "loss": 0.8283,
+      "step": 12301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.781796456704261e-07,
+      "loss": 0.9207,
+      "step": 12302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.772634598625917e-07,
+      "loss": 1.0382,
+      "step": 12303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.763477302974044e-07,
+      "loss": 0.579,
+      "step": 12304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.75432457020664e-07,
+      "loss": 0.9042,
+      "step": 12305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.745176400781418e-07,
+      "loss": 0.8189,
+      "step": 12306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.736032795155925e-07,
+      "loss": 0.9201,
+      "step": 12307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.726893753787391e-07,
+      "loss": 0.8498,
+      "step": 12308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.717759277132943e-07,
+      "loss": 0.7627,
+      "step": 12309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.708629365649368e-07,
+      "loss": 0.8853,
+      "step": 12310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.699504019793282e-07,
+      "loss": 0.7737,
+      "step": 12311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.690383240021061e-07,
+      "loss": 0.854,
+      "step": 12312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.68126702678882e-07,
+      "loss": 0.9051,
+      "step": 12313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672155380552516e-07,
+      "loss": 0.8381,
+      "step": 12314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.663048301767818e-07,
+      "loss": 0.8128,
+      "step": 12315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.653945790890183e-07,
+      "loss": 0.8824,
+      "step": 12316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.644847848374859e-07,
+      "loss": 0.8871,
+      "step": 12317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.635754474676805e-07,
+      "loss": 0.9101,
+      "step": 12318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.626665670250856e-07,
+      "loss": 0.9395,
+      "step": 12319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617581435551514e-07,
+      "loss": 0.6378,
+      "step": 12320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.608501771033118e-07,
+      "loss": 0.9088,
+      "step": 12321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.599426677149747e-07,
+      "loss": 0.9132,
+      "step": 12322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.59035615435525e-07,
+      "loss": 0.8608,
+      "step": 12323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.581290203103276e-07,
+      "loss": 0.8501,
+      "step": 12324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.572228823847218e-07,
+      "loss": 0.8496,
+      "step": 12325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.563172017040267e-07,
+      "loss": 0.7108,
+      "step": 12326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.554119783135351e-07,
+      "loss": 0.8223,
+      "step": 12327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.545072122585196e-07,
+      "loss": 0.6766,
+      "step": 12328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.536029035842264e-07,
+      "loss": 0.8957,
+      "step": 12329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.526990523358858e-07,
+      "loss": 0.7699,
+      "step": 12330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.517956585586983e-07,
+      "loss": 0.7449,
+      "step": 12331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.508927222978447e-07,
+      "loss": 0.751,
+      "step": 12332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.499902435984809e-07,
+      "loss": 0.8732,
+      "step": 12333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.490882225057428e-07,
+      "loss": 1.0003,
+      "step": 12334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.481866590647425e-07,
+      "loss": 0.8909,
+      "step": 12335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.472855533205671e-07,
+      "loss": 0.972,
+      "step": 12336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463849053182826e-07,
+      "loss": 0.863,
+      "step": 12337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.45484715102931e-07,
+      "loss": 0.9391,
+      "step": 12338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.445849827195352e-07,
+      "loss": 0.8054,
+      "step": 12339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.4368570821309e-07,
+      "loss": 0.7776,
+      "step": 12340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.427868916285698e-07,
+      "loss": 0.8654,
+      "step": 12341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.418885330109261e-07,
+      "loss": 0.9137,
+      "step": 12342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.409906324050865e-07,
+      "loss": 1.0037,
+      "step": 12343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.400931898559573e-07,
+      "loss": 0.916,
+      "step": 12344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.391962054084212e-07,
+      "loss": 0.9377,
+      "step": 12345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.382996791073383e-07,
+      "loss": 0.8613,
+      "step": 12346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.374036109975448e-07,
+      "loss": 0.7334,
+      "step": 12347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.365080011238513e-07,
+      "loss": 0.7794,
+      "step": 12348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.356128495310534e-07,
+      "loss": 0.8767,
+      "step": 12349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.347181562639183e-07,
+      "loss": 0.8305,
+      "step": 12350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338239213671894e-07,
+      "loss": 0.7555,
+      "step": 12351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.329301448855875e-07,
+      "loss": 0.9107,
+      "step": 12352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.320368268638157e-07,
+      "loss": 0.9507,
+      "step": 12353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311439673465482e-07,
+      "loss": 0.8229,
+      "step": 12354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.302515663784383e-07,
+      "loss": 0.9162,
+      "step": 12355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.293596240041146e-07,
+      "loss": 0.9316,
+      "step": 12356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.284681402681882e-07,
+      "loss": 0.8733,
+      "step": 12357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.275771152152412e-07,
+      "loss": 0.9356,
+      "step": 12358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.266865488898368e-07,
+      "loss": 0.8496,
+      "step": 12359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.257964413365127e-07,
+      "loss": 0.9007,
+      "step": 12360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.249067925997845e-07,
+      "loss": 0.9572,
+      "step": 12361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.240176027241453e-07,
+      "loss": 0.8368,
+      "step": 12362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.231288717540631e-07,
+      "loss": 0.9163,
+      "step": 12363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.222405997339878e-07,
+      "loss": 0.8931,
+      "step": 12364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.213527867083427e-07,
+      "loss": 0.7996,
+      "step": 12365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.204654327215267e-07,
+      "loss": 0.7641,
+      "step": 12366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.195785378179199e-07,
+      "loss": 0.7937,
+      "step": 12367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.186921020418736e-07,
+      "loss": 0.8755,
+      "step": 12368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.178061254377257e-07,
+      "loss": 0.8667,
+      "step": 12369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.169206080497816e-07,
+      "loss": 0.8375,
+      "step": 12370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.160355499223282e-07,
+      "loss": 0.8197,
+      "step": 12371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151509510996292e-07,
+      "loss": 0.7816,
+      "step": 12372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.142668116259223e-07,
+      "loss": 0.9287,
+      "step": 12373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.133831315454277e-07,
+      "loss": 0.8358,
+      "step": 12374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124999109023391e-07,
+      "loss": 0.8334,
+      "step": 12375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.116171497408277e-07,
+      "loss": 0.9373,
+      "step": 12376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.107348481050392e-07,
+      "loss": 0.9682,
+      "step": 12377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.09853006039103e-07,
+      "loss": 0.884,
+      "step": 12378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.089716235871192e-07,
+      "loss": 0.9133,
+      "step": 12379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.080907007931682e-07,
+      "loss": 0.8483,
+      "step": 12380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.072102377013047e-07,
+      "loss": 0.8352,
+      "step": 12381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.063302343555623e-07,
+      "loss": 0.8685,
+      "step": 12382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.054506907999537e-07,
+      "loss": 0.6208,
+      "step": 12383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.045716070784637e-07,
+      "loss": 0.9905,
+      "step": 12384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.036929832350582e-07,
+      "loss": 0.8356,
+      "step": 12385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.028148193136787e-07,
+      "loss": 0.8427,
+      "step": 12386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.0193711535824e-07,
+      "loss": 0.8399,
+      "step": 12387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.010598714126405e-07,
+      "loss": 0.8337,
+      "step": 12388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.00183087520755e-07,
+      "loss": 0.8754,
+      "step": 12389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.993067637264295e-07,
+      "loss": 0.878,
+      "step": 12390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.984309000734924e-07,
+      "loss": 0.8884,
+      "step": 12391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.975554966057453e-07,
+      "loss": 0.8511,
+      "step": 12392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.966805533669663e-07,
+      "loss": 0.8995,
+      "step": 12393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.958060704009185e-07,
+      "loss": 0.8111,
+      "step": 12394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.949320477513323e-07,
+      "loss": 0.8519,
+      "step": 12395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.940584854619204e-07,
+      "loss": 0.581,
+      "step": 12396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.931853835763703e-07,
+      "loss": 0.7736,
+      "step": 12397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.923127421383459e-07,
+      "loss": 0.8367,
+      "step": 12398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.914405611914921e-07,
+      "loss": 0.9071,
+      "step": 12399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.905688407794276e-07,
+      "loss": 0.8912,
+      "step": 12400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.896975809457475e-07,
+      "loss": 0.7948,
+      "step": 12401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.888267817340234e-07,
+      "loss": 0.9113,
+      "step": 12402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.879564431878084e-07,
+      "loss": 0.9916,
+      "step": 12403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.870865653506288e-07,
+      "loss": 0.9298,
+      "step": 12404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.862171482659874e-07,
+      "loss": 0.8287,
+      "step": 12405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.853481919773665e-07,
+      "loss": 0.9541,
+      "step": 12406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.84479696528222e-07,
+      "loss": 0.6265,
+      "step": 12407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.836116619619916e-07,
+      "loss": 0.8382,
+      "step": 12408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.827440883220849e-07,
+      "loss": 0.9195,
+      "step": 12409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.818769756518907e-07,
+      "loss": 0.8092,
+      "step": 12410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.810103239947763e-07,
+      "loss": 0.8266,
+      "step": 12411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.801441333940818e-07,
+      "loss": 0.9009,
+      "step": 12412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.792784038931289e-07,
+      "loss": 0.5951,
+      "step": 12413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.784131355352142e-07,
+      "loss": 0.9573,
+      "step": 12414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.77548328363611e-07,
+      "loss": 0.7979,
+      "step": 12415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.766839824215689e-07,
+      "loss": 0.8555,
+      "step": 12416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758200977523134e-07,
+      "loss": 0.9289,
+      "step": 12417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.749566743990522e-07,
+      "loss": 0.9634,
+      "step": 12418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.740937124049641e-07,
+      "loss": 0.8371,
+      "step": 12419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73231211813209e-07,
+      "loss": 0.9619,
+      "step": 12420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.723691726669225e-07,
+      "loss": 0.9312,
+      "step": 12421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.715075950092143e-07,
+      "loss": 0.849,
+      "step": 12422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.706464788831725e-07,
+      "loss": 1.0322,
+      "step": 12423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.697858243318667e-07,
+      "loss": 0.6229,
+      "step": 12424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689256313983373e-07,
+      "loss": 0.8936,
+      "step": 12425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.680659001256042e-07,
+      "loss": 0.7929,
+      "step": 12426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.672066305566616e-07,
+      "loss": 0.8389,
+      "step": 12427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.663478227344867e-07,
+      "loss": 0.8967,
+      "step": 12428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.654894767020293e-07,
+      "loss": 0.8231,
+      "step": 12429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.646315925022151e-07,
+      "loss": 0.9248,
+      "step": 12430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.6377417017795e-07,
+      "loss": 0.8481,
+      "step": 12431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.629172097721104e-07,
+      "loss": 0.9274,
+      "step": 12432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.620607113275614e-07,
+      "loss": 0.9072,
+      "step": 12433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-07,
+      "loss": 0.8893,
+      "step": 12434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.603491004936381e-07,
+      "loss": 0.9902,
+      "step": 12435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594939881898667e-07,
+      "loss": 0.8021,
+      "step": 12436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.586393380185797e-07,
+      "loss": 0.8684,
+      "step": 12437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.577851500225264e-07,
+      "loss": 0.8907,
+      "step": 12438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.569314242444215e-07,
+      "loss": 0.8206,
+      "step": 12439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.560781607269618e-07,
+      "loss": 0.8788,
+      "step": 12440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.552253595128223e-07,
+      "loss": 0.8692,
+      "step": 12441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543730206446487e-07,
+      "loss": 0.829,
+      "step": 12442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.535211441650725e-07,
+      "loss": 0.9113,
+      "step": 12443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.526697301166963e-07,
+      "loss": 0.6047,
+      "step": 12444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.518187785420994e-07,
+      "loss": 0.8759,
+      "step": 12445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.509682894838378e-07,
+      "loss": 0.8414,
+      "step": 12446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.501182629844506e-07,
+      "loss": 0.8769,
+      "step": 12447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.49268699086445e-07,
+      "loss": 0.8177,
+      "step": 12448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.484195978323105e-07,
+      "loss": 0.9771,
+      "step": 12449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.475709592645109e-07,
+      "loss": 0.8254,
+      "step": 12450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.467227834254898e-07,
+      "loss": 0.8222,
+      "step": 12451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.458750703576634e-07,
+      "loss": 0.8791,
+      "step": 12452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4502782010343e-07,
+      "loss": 0.9268,
+      "step": 12453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.441810327051613e-07,
+      "loss": 0.8677,
+      "step": 12454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.433347082052045e-07,
+      "loss": 0.8715,
+      "step": 12455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.424888466458879e-07,
+      "loss": 0.8283,
+      "step": 12456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.416434480695123e-07,
+      "loss": 0.9075,
+      "step": 12457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.407985125183592e-07,
+      "loss": 0.9522,
+      "step": 12458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.39954040034685e-07,
+      "loss": 0.867,
+      "step": 12459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.391100306607224e-07,
+      "loss": 0.965,
+      "step": 12460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.382664844386822e-07,
+      "loss": 0.8627,
+      "step": 12461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374234014107484e-07,
+      "loss": 0.8736,
+      "step": 12462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.365807816190907e-07,
+      "loss": 0.7971,
+      "step": 12463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.357386251058462e-07,
+      "loss": 0.7783,
+      "step": 12464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.348969319131338e-07,
+      "loss": 0.9372,
+      "step": 12465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.340557020830475e-07,
+      "loss": 0.8362,
+      "step": 12466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.332149356576556e-07,
+      "loss": 0.8158,
+      "step": 12467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.323746326790115e-07,
+      "loss": 1.0027,
+      "step": 12468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.31534793189137e-07,
+      "loss": 0.9021,
+      "step": 12469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.306954172300351e-07,
+      "loss": 0.8904,
+      "step": 12470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.298565048436812e-07,
+      "loss": 0.8999,
+      "step": 12471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.290180560720351e-07,
+      "loss": 0.7754,
+      "step": 12472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281800709570264e-07,
+      "loss": 0.9093,
+      "step": 12473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.273425495405651e-07,
+      "loss": 0.8554,
+      "step": 12474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.265054918645364e-07,
+      "loss": 0.8671,
+      "step": 12475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.256688979708015e-07,
+      "loss": 0.9159,
+      "step": 12476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.248327679012024e-07,
+      "loss": 0.7514,
+      "step": 12477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.239971016975544e-07,
+      "loss": 0.8453,
+      "step": 12478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.231618994016499e-07,
+      "loss": 0.9327,
+      "step": 12479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.223271610552596e-07,
+      "loss": 0.8889,
+      "step": 12480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21492886700128e-07,
+      "loss": 0.8237,
+      "step": 12481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.206590763779797e-07,
+      "loss": 0.9255,
+      "step": 12482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.198257301305167e-07,
+      "loss": 0.8649,
+      "step": 12483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.189928479994146e-07,
+      "loss": 0.815,
+      "step": 12484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.181604300263279e-07,
+      "loss": 0.8977,
+      "step": 12485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.173284762528853e-07,
+      "loss": 0.7794,
+      "step": 12486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.164969867206939e-07,
+      "loss": 0.8477,
+      "step": 12487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.156659614713413e-07,
+      "loss": 0.6251,
+      "step": 12488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.148354005463854e-07,
+      "loss": 0.8287,
+      "step": 12489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.140053039873652e-07,
+      "loss": 0.8547,
+      "step": 12490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.13175671835794e-07,
+      "loss": 0.7801,
+      "step": 12491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123465041331634e-07,
+      "loss": 0.8657,
+      "step": 12492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.115178009209423e-07,
+      "loss": 0.9134,
+      "step": 12493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.106895622405752e-07,
+      "loss": 0.976,
+      "step": 12494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098617881334824e-07,
+      "loss": 0.8454,
+      "step": 12495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090344786410619e-07,
+      "loss": 0.8571,
+      "step": 12496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.082076338046928e-07,
+      "loss": 0.9027,
+      "step": 12497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.073812536657232e-07,
+      "loss": 0.91,
+      "step": 12498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.065553382654822e-07,
+      "loss": 0.8712,
+      "step": 12499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.057298876452756e-07,
+      "loss": 0.948,
+      "step": 12500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.04904901846385e-07,
+      "loss": 0.8115,
+      "step": 12501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.040803809100705e-07,
+      "loss": 0.805,
+      "step": 12502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.032563248775659e-07,
+      "loss": 0.9545,
+      "step": 12503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.024327337900861e-07,
+      "loss": 0.8265,
+      "step": 12504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.016096076888179e-07,
+      "loss": 0.9441,
+      "step": 12505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.007869466149275e-07,
+      "loss": 0.6094,
+      "step": 12506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.999647506095586e-07,
+      "loss": 0.9184,
+      "step": 12507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991430197138294e-07,
+      "loss": 0.8368,
+      "step": 12508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.983217539688369e-07,
+      "loss": 0.805,
+      "step": 12509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.975009534156541e-07,
+      "loss": 0.9393,
+      "step": 12510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.966806180953279e-07,
+      "loss": 0.9751,
+      "step": 12511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.958607480488877e-07,
+      "loss": 0.8558,
+      "step": 12512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950413433173342e-07,
+      "loss": 0.8376,
+      "step": 12513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9422240394165e-07,
+      "loss": 0.5904,
+      "step": 12514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.934039299627903e-07,
+      "loss": 0.9901,
+      "step": 12515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.92585921421688e-07,
+      "loss": 0.8217,
+      "step": 12516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.917683783592511e-07,
+      "loss": 0.9681,
+      "step": 12517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.909513008163693e-07,
+      "loss": 0.8963,
+      "step": 12518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.901346888339045e-07,
+      "loss": 0.6068,
+      "step": 12519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.893185424526983e-07,
+      "loss": 0.7973,
+      "step": 12520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.885028617135636e-07,
+      "loss": 0.9184,
+      "step": 12521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.87687646657298e-07,
+      "loss": 0.9241,
+      "step": 12522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.868728973246696e-07,
+      "loss": 0.9776,
+      "step": 12523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860586137564273e-07,
+      "loss": 0.9108,
+      "step": 12524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.852447959932929e-07,
+      "loss": 0.9017,
+      "step": 12525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.844314440759648e-07,
+      "loss": 0.9153,
+      "step": 12526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.836185580451249e-07,
+      "loss": 0.8326,
+      "step": 12527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.828061379414242e-07,
+      "loss": 0.9981,
+      "step": 12528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.819941838054934e-07,
+      "loss": 0.8598,
+      "step": 12529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.811826956779389e-07,
+      "loss": 0.964,
+      "step": 12530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.803716735993449e-07,
+      "loss": 0.8799,
+      "step": 12531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.795611176102723e-07,
+      "loss": 0.9092,
+      "step": 12532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.787510277512599e-07,
+      "loss": 0.8225,
+      "step": 12533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.779414040628185e-07,
+      "loss": 0.8544,
+      "step": 12534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.771322465854402e-07,
+      "loss": 0.6049,
+      "step": 12535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.763235553595915e-07,
+      "loss": 0.8481,
+      "step": 12536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.755153304257179e-07,
+      "loss": 0.7561,
+      "step": 12537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.747075718242391e-07,
+      "loss": 0.7709,
+      "step": 12538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.739002795955518e-07,
+      "loss": 0.9212,
+      "step": 12539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.730934537800305e-07,
+      "loss": 0.8947,
+      "step": 12540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.72287094418026e-07,
+      "loss": 0.8721,
+      "step": 12541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.714812015498662e-07,
+      "loss": 0.9484,
+      "step": 12542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.706757752158544e-07,
+      "loss": 0.7789,
+      "step": 12543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.698708154562694e-07,
+      "loss": 0.9199,
+      "step": 12544
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.690663223113725e-07,
+      "loss": 1.0057,
+      "step": 12545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.682622958213947e-07,
+      "loss": 0.8289,
+      "step": 12546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.674587360265483e-07,
+      "loss": 0.9952,
+      "step": 12547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.66655642967019e-07,
+      "loss": 0.8847,
+      "step": 12548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.658530166829735e-07,
+      "loss": 0.7979,
+      "step": 12549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.650508572145486e-07,
+      "loss": 0.8556,
+      "step": 12550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.642491646018634e-07,
+      "loss": 0.9482,
+      "step": 12551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.634479388850135e-07,
+      "loss": 0.877,
+      "step": 12552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.626471801040679e-07,
+      "loss": 0.7942,
+      "step": 12553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.618468882990725e-07,
+      "loss": 0.9549,
+      "step": 12554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.610470635100541e-07,
+      "loss": 0.841,
+      "step": 12555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.602477057770096e-07,
+      "loss": 0.8642,
+      "step": 12556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.594488151399203e-07,
+      "loss": 0.8919,
+      "step": 12557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.586503916387366e-07,
+      "loss": 0.7585,
+      "step": 12558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.578524353133919e-07,
+      "loss": 0.8931,
+      "step": 12559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.570549462037901e-07,
+      "loss": 0.8118,
+      "step": 12560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.562579243498157e-07,
+      "loss": 0.8705,
+      "step": 12561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.554613697913315e-07,
+      "loss": 0.9231,
+      "step": 12562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546652825681732e-07,
+      "loss": 0.9122,
+      "step": 12563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.538696627201525e-07,
+      "loss": 0.8439,
+      "step": 12564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.530745102870606e-07,
+      "loss": 0.7644,
+      "step": 12565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.52279825308666e-07,
+      "loss": 0.857,
+      "step": 12566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.514856078247111e-07,
+      "loss": 0.7379,
+      "step": 12567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.506918578749155e-07,
+      "loss": 0.9353,
+      "step": 12568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.498985754989773e-07,
+      "loss": 0.8063,
+      "step": 12569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.491057607365659e-07,
+      "loss": 0.8558,
+      "step": 12570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.483134136273362e-07,
+      "loss": 0.737,
+      "step": 12571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.475215342109131e-07,
+      "loss": 0.7408,
+      "step": 12572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.467301225268996e-07,
+      "loss": 0.8259,
+      "step": 12573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.459391786148739e-07,
+      "loss": 0.8309,
+      "step": 12574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.451487025143932e-07,
+      "loss": 0.872,
+      "step": 12575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.443586942649905e-07,
+      "loss": 0.8195,
+      "step": 12576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.435691539061772e-07,
+      "loss": 0.9761,
+      "step": 12577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427800814774388e-07,
+      "loss": 0.9354,
+      "step": 12578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.419914770182367e-07,
+      "loss": 0.9119,
+      "step": 12579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.412033405680107e-07,
+      "loss": 0.9012,
+      "step": 12580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.404156721661758e-07,
+      "loss": 0.984,
+      "step": 12581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.396284718521272e-07,
+      "loss": 0.7336,
+      "step": 12582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.388417396652336e-07,
+      "loss": 0.86,
+      "step": 12583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.380554756448387e-07,
+      "loss": 0.8168,
+      "step": 12584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.372696798302658e-07,
+      "loss": 0.96,
+      "step": 12585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.364843522608133e-07,
+      "loss": 0.8276,
+      "step": 12586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.356994929757576e-07,
+      "loss": 0.8748,
+      "step": 12587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.349151020143507e-07,
+      "loss": 0.7888,
+      "step": 12588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.341311794158211e-07,
+      "loss": 0.7695,
+      "step": 12589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.333477252193731e-07,
+      "loss": 0.8601,
+      "step": 12590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325647394641898e-07,
+      "loss": 0.9191,
+      "step": 12591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.317822221894298e-07,
+      "loss": 0.7966,
+      "step": 12592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.310001734342263e-07,
+      "loss": 0.8697,
+      "step": 12593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.302185932376925e-07,
+      "loss": 0.8204,
+      "step": 12594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.294374816389148e-07,
+      "loss": 0.898,
+      "step": 12595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.286568386769598e-07,
+      "loss": 0.9041,
+      "step": 12596
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.278766643908674e-07,
+      "loss": 0.8601,
+      "step": 12597
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270969588196563e-07,
+      "loss": 0.8006,
+      "step": 12598
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.263177220023198e-07,
+      "loss": 0.8399,
+      "step": 12599
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.255389539778278e-07,
+      "loss": 0.8836,
+      "step": 12600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.247606547851303e-07,
+      "loss": 0.8113,
+      "step": 12601
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239828244631508e-07,
+      "loss": 0.9329,
+      "step": 12602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232054630507889e-07,
+      "loss": 0.6065,
+      "step": 12603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.224285705869216e-07,
+      "loss": 0.8486,
+      "step": 12604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.21652147110402e-07,
+      "loss": 0.7993,
+      "step": 12605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.208761926600626e-07,
+      "loss": 0.7473,
+      "step": 12606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.201007072747067e-07,
+      "loss": 0.8772,
+      "step": 12607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.193256909931211e-07,
+      "loss": 0.8897,
+      "step": 12608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185511438540648e-07,
+      "loss": 0.8259,
+      "step": 12609
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.177770658962723e-07,
+      "loss": 0.8462,
+      "step": 12610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170034571584571e-07,
+      "loss": 0.5894,
+      "step": 12611
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.162303176793094e-07,
+      "loss": 0.8969,
+      "step": 12612
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.154576474974961e-07,
+      "loss": 0.9064,
+      "step": 12613
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.146854466516572e-07,
+      "loss": 0.9356,
+      "step": 12614
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.13913715180412e-07,
+      "loss": 0.9142,
+      "step": 12615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.131424531223584e-07,
+      "loss": 0.9823,
+      "step": 12616
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.123716605160667e-07,
+      "loss": 0.8915,
+      "step": 12617
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.116013374000862e-07,
+      "loss": 0.8437,
+      "step": 12618
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.108314838129415e-07,
+      "loss": 0.8974,
+      "step": 12619
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.100620997931328e-07,
+      "loss": 0.9456,
+      "step": 12620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.092931853791406e-07,
+      "loss": 0.8383,
+      "step": 12621
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.085247406094197e-07,
+      "loss": 0.8132,
+      "step": 12622
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.077567655223993e-07,
+      "loss": 0.7728,
+      "step": 12623
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.069892601564875e-07,
+      "loss": 0.7522,
+      "step": 12624
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.062222245500682e-07,
+      "loss": 0.9077,
+      "step": 12625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054556587415039e-07,
+      "loss": 0.9179,
+      "step": 12626
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.046895627691318e-07,
+      "loss": 0.8775,
+      "step": 12627
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.039239366712635e-07,
+      "loss": 0.8483,
+      "step": 12628
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.031587804861894e-07,
+      "loss": 0.7928,
+      "step": 12629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.023940942521766e-07,
+      "loss": 0.8761,
+      "step": 12630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.016298780074692e-07,
+      "loss": 0.8902,
+      "step": 12631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008661317902864e-07,
+      "loss": 0.7153,
+      "step": 12632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.001028556388244e-07,
+      "loss": 0.8095,
+      "step": 12633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.993400495912549e-07,
+      "loss": 0.8476,
+      "step": 12634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.985777136857285e-07,
+      "loss": 0.9376,
+      "step": 12635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.978158479603702e-07,
+      "loss": 0.8176,
+      "step": 12636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970544524532828e-07,
+      "loss": 0.7897,
+      "step": 12637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962935272025416e-07,
+      "loss": 0.906,
+      "step": 12638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.955330722462071e-07,
+      "loss": 0.9199,
+      "step": 12639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.947730876223057e-07,
+      "loss": 0.94,
+      "step": 12640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940135733688501e-07,
+      "loss": 0.8101,
+      "step": 12641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.932545295238213e-07,
+      "loss": 0.8401,
+      "step": 12642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.924959561251831e-07,
+      "loss": 0.6206,
+      "step": 12643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.917378532108698e-07,
+      "loss": 0.9585,
+      "step": 12644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.909802208187964e-07,
+      "loss": 0.989,
+      "step": 12645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.902230589868552e-07,
+      "loss": 0.8953,
+      "step": 12646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.894663677529122e-07,
+      "loss": 0.8606,
+      "step": 12647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.887101471548096e-07,
+      "loss": 0.7594,
+      "step": 12648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.879543972303681e-07,
+      "loss": 0.8257,
+      "step": 12649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871991180173819e-07,
+      "loss": 0.7797,
+      "step": 12650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.864443095536276e-07,
+      "loss": 0.9178,
+      "step": 12651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.856899718768527e-07,
+      "loss": 0.8301,
+      "step": 12652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849361050247815e-07,
+      "loss": 0.7926,
+      "step": 12653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84182709035117e-07,
+      "loss": 0.9158,
+      "step": 12654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.834297839455372e-07,
+      "loss": 0.8508,
+      "step": 12655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826773297936994e-07,
+      "loss": 0.9614,
+      "step": 12656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.819253466172325e-07,
+      "loss": 1.0024,
+      "step": 12657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.811738344537454e-07,
+      "loss": 0.8562,
+      "step": 12658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.804227933408202e-07,
+      "loss": 0.9157,
+      "step": 12659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.796722233160223e-07,
+      "loss": 1.0051,
+      "step": 12660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.789221244168852e-07,
+      "loss": 0.9284,
+      "step": 12661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.781724966809244e-07,
+      "loss": 0.5905,
+      "step": 12662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.774233401456286e-07,
+      "loss": 0.8967,
+      "step": 12663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766746548484647e-07,
+      "loss": 0.7641,
+      "step": 12664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.75926440826876e-07,
+      "loss": 0.9061,
+      "step": 12665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.751786981182828e-07,
+      "loss": 0.8843,
+      "step": 12666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.744314267600792e-07,
+      "loss": 0.8951,
+      "step": 12667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.73684626789639e-07,
+      "loss": 0.8925,
+      "step": 12668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.729382982443077e-07,
+      "loss": 0.8159,
+      "step": 12669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.721924411614122e-07,
+      "loss": 0.8961,
+      "step": 12670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.71447055578257e-07,
+      "loss": 0.8979,
+      "step": 12671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.707021415321168e-07,
+      "loss": 0.6026,
+      "step": 12672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.699576990602474e-07,
+      "loss": 0.8741,
+      "step": 12673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.692137281998788e-07,
+      "loss": 0.8551,
+      "step": 12674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.68470228988216e-07,
+      "loss": 0.8307,
+      "step": 12675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.677272014624468e-07,
+      "loss": 0.9387,
+      "step": 12676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.669846456597284e-07,
+      "loss": 0.6468,
+      "step": 12677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.662425616171985e-07,
+      "loss": 0.812,
+      "step": 12678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.655009493719688e-07,
+      "loss": 0.8834,
+      "step": 12679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.647598089611284e-07,
+      "loss": 0.8823,
+      "step": 12680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.640191404217454e-07,
+      "loss": 0.9337,
+      "step": 12681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.632789437908592e-07,
+      "loss": 0.985,
+      "step": 12682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.62539219105489e-07,
+      "loss": 0.8584,
+      "step": 12683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.617999664026286e-07,
+      "loss": 0.7832,
+      "step": 12684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.610611857192516e-07,
+      "loss": 0.9341,
+      "step": 12685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.603228770923042e-07,
+      "loss": 0.8394,
+      "step": 12686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.595850405587111e-07,
+      "loss": 0.7711,
+      "step": 12687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.588476761553707e-07,
+      "loss": 0.8674,
+      "step": 12688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.581107839191613e-07,
+      "loss": 0.8622,
+      "step": 12689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.573743638869355e-07,
+      "loss": 0.7757,
+      "step": 12690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.566384160955252e-07,
+      "loss": 0.768,
+      "step": 12691
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.559029405817329e-07,
+      "loss": 0.9563,
+      "step": 12692
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.551679373823427e-07,
+      "loss": 0.8763,
+      "step": 12693
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.544334065341118e-07,
+      "loss": 0.8258,
+      "step": 12694
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.536993480737774e-07,
+      "loss": 0.9186,
+      "step": 12695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.529657620380502e-07,
+      "loss": 0.85,
+      "step": 12696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.522326484636176e-07,
+      "loss": 0.6997,
+      "step": 12697
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.515000073871446e-07,
+      "loss": 0.8546,
+      "step": 12698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.507678388452687e-07,
+      "loss": 0.9809,
+      "step": 12699
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.500361428746114e-07,
+      "loss": 0.8138,
+      "step": 12700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.493049195117628e-07,
+      "loss": 0.8823,
+      "step": 12701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.485741687932933e-07,
+      "loss": 0.8411,
+      "step": 12702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.478438907557504e-07,
+      "loss": 0.8595,
+      "step": 12703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.471140854356538e-07,
+      "loss": 0.9023,
+      "step": 12704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.463847528695043e-07,
+      "loss": 0.6147,
+      "step": 12705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.456558930937772e-07,
+      "loss": 0.9482,
+      "step": 12706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.449275061449233e-07,
+      "loss": 0.953,
+      "step": 12707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44199592059369e-07,
+      "loss": 0.7839,
+      "step": 12708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.434721508735197e-07,
+      "loss": 0.8015,
+      "step": 12709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427451826237562e-07,
+      "loss": 0.7588,
+      "step": 12710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420186873464351e-07,
+      "loss": 0.9569,
+      "step": 12711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.412926650778893e-07,
+      "loss": 0.8254,
+      "step": 12712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.40567115854429e-07,
+      "loss": 0.6053,
+      "step": 12713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.398420397123383e-07,
+      "loss": 0.9655,
+      "step": 12714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391174366878815e-07,
+      "loss": 0.9162,
+      "step": 12715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.383933068172975e-07,
+      "loss": 0.5891,
+      "step": 12716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376696501367984e-07,
+      "loss": 0.8232,
+      "step": 12717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369464666825785e-07,
+      "loss": 0.8994,
+      "step": 12718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.362237564908013e-07,
+      "loss": 0.9754,
+      "step": 12719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.355015195976154e-07,
+      "loss": 0.8519,
+      "step": 12720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.347797560391399e-07,
+      "loss": 1.0152,
+      "step": 12721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.340584658514692e-07,
+      "loss": 0.8433,
+      "step": 12722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.333376490706776e-07,
+      "loss": 0.9059,
+      "step": 12723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326173057328132e-07,
+      "loss": 0.9192,
+      "step": 12724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.318974358739026e-07,
+      "loss": 0.6202,
+      "step": 12725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.311780395299481e-07,
+      "loss": 0.8549,
+      "step": 12726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.304591167369277e-07,
+      "loss": 0.8905,
+      "step": 12727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.297406675307936e-07,
+      "loss": 0.8524,
+      "step": 12728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.290226919474795e-07,
+      "loss": 0.853,
+      "step": 12729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28305190022892e-07,
+      "loss": 0.88,
+      "step": 12730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275881617929136e-07,
+      "loss": 0.9238,
+      "step": 12731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.268716072934032e-07,
+      "loss": 0.9647,
+      "step": 12732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.261555265601992e-07,
+      "loss": 0.761,
+      "step": 12733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254399196291115e-07,
+      "loss": 0.9028,
+      "step": 12734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247247865359306e-07,
+      "loss": 0.8177,
+      "step": 12735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.240101273164222e-07,
+      "loss": 0.8546,
+      "step": 12736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.232959420063244e-07,
+      "loss": 0.9329,
+      "step": 12737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.225822306413575e-07,
+      "loss": 0.8965,
+      "step": 12738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.21868993257213e-07,
+      "loss": 0.6122,
+      "step": 12739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211562298895645e-07,
+      "loss": 0.8674,
+      "step": 12740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204439405740558e-07,
+      "loss": 0.8338,
+      "step": 12741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.197321253463095e-07,
+      "loss": 0.9223,
+      "step": 12742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.190207842419259e-07,
+      "loss": 0.6027,
+      "step": 12743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.183099172964778e-07,
+      "loss": 0.9632,
+      "step": 12744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175995245455212e-07,
+      "loss": 0.8161,
+      "step": 12745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.168896060245798e-07,
+      "loss": 0.6095,
+      "step": 12746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.161801617691608e-07,
+      "loss": 0.7948,
+      "step": 12747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.154711918147425e-07,
+      "loss": 0.8544,
+      "step": 12748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.147626961967811e-07,
+      "loss": 0.9015,
+      "step": 12749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.140546749507136e-07,
+      "loss": 0.837,
+      "step": 12750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.133471281119451e-07,
+      "loss": 0.8313,
+      "step": 12751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.126400557158628e-07,
+      "loss": 0.6158,
+      "step": 12752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.119334577978275e-07,
+      "loss": 0.75,
+      "step": 12753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.112273343931795e-07,
+      "loss": 0.8296,
+      "step": 12754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105216855372308e-07,
+      "loss": 0.8426,
+      "step": 12755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098165112652742e-07,
+      "loss": 0.8892,
+      "step": 12756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.091118116125749e-07,
+      "loss": 0.7713,
+      "step": 12757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084075866143745e-07,
+      "loss": 0.9879,
+      "step": 12758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.077038363058972e-07,
+      "loss": 0.8645,
+      "step": 12759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07000560722335e-07,
+      "loss": 0.8871,
+      "step": 12760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.062977598988617e-07,
+      "loss": 0.8385,
+      "step": 12761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.055954338706237e-07,
+      "loss": 0.8608,
+      "step": 12762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.048935826727453e-07,
+      "loss": 0.6599,
+      "step": 12763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.041922063403282e-07,
+      "loss": 0.9333,
+      "step": 12764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.034913049084511e-07,
+      "loss": 0.8942,
+      "step": 12765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027908784121649e-07,
+      "loss": 0.9404,
+      "step": 12766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.020909268865004e-07,
+      "loss": 0.8454,
+      "step": 12767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013914503664619e-07,
+      "loss": 0.6321,
+      "step": 12768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.006924488870313e-07,
+      "loss": 0.9108,
+      "step": 12769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.999939224831685e-07,
+      "loss": 0.6322,
+      "step": 12770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.992958711898066e-07,
+      "loss": 0.8586,
+      "step": 12771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.985982950418566e-07,
+      "loss": 0.9778,
+      "step": 12772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.979011940742051e-07,
+      "loss": 0.7566,
+      "step": 12773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.972045683217152e-07,
+      "loss": 0.7324,
+      "step": 12774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.965084178192269e-07,
+      "loss": 0.8719,
+      "step": 12775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958127426015557e-07,
+      "loss": 0.7429,
+      "step": 12776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.951175427034937e-07,
+      "loss": 0.885,
+      "step": 12777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.944228181598065e-07,
+      "loss": 0.9432,
+      "step": 12778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.937285690052418e-07,
+      "loss": 0.8736,
+      "step": 12779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.930347952745196e-07,
+      "loss": 0.8286,
+      "step": 12780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.923414970023343e-07,
+      "loss": 0.935,
+      "step": 12781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916486742233606e-07,
+      "loss": 0.7888,
+      "step": 12782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909563269722461e-07,
+      "loss": 0.629,
+      "step": 12783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.902644552836188e-07,
+      "loss": 0.8552,
+      "step": 12784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895730591920788e-07,
+      "loss": 1.0172,
+      "step": 12785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888821387322041e-07,
+      "loss": 0.8269,
+      "step": 12786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.881916939385489e-07,
+      "loss": 0.8536,
+      "step": 12787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875017248456415e-07,
+      "loss": 0.8095,
+      "step": 12788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.868122314879919e-07,
+      "loss": 0.8799,
+      "step": 12789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.861232139000804e-07,
+      "loss": 0.8738,
+      "step": 12790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854346721163672e-07,
+      "loss": 0.797,
+      "step": 12791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847466061712858e-07,
+      "loss": 0.8409,
+      "step": 12792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.840590160992475e-07,
+      "loss": 0.7908,
+      "step": 12793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833719019346417e-07,
+      "loss": 0.8254,
+      "step": 12794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.826852637118295e-07,
+      "loss": 0.57,
+      "step": 12795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.819991014651537e-07,
+      "loss": 0.8407,
+      "step": 12796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.813134152289278e-07,
+      "loss": 0.8041,
+      "step": 12797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.806282050374456e-07,
+      "loss": 0.9331,
+      "step": 12798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.799434709249751e-07,
+      "loss": 0.8438,
+      "step": 12799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.792592129257611e-07,
+      "loss": 0.8201,
+      "step": 12800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.785754310740254e-07,
+      "loss": 0.944,
+      "step": 12801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.778921254039625e-07,
+      "loss": 0.8408,
+      "step": 12802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772092959497465e-07,
+      "loss": 0.8748,
+      "step": 12803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7652694274552877e-07,
+      "loss": 0.9109,
+      "step": 12804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.758450658254332e-07,
+      "loss": 0.8229,
+      "step": 12805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.751636652235614e-07,
+      "loss": 0.6359,
+      "step": 12806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7448274097399163e-07,
+      "loss": 0.593,
+      "step": 12807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7380229311077775e-07,
+      "loss": 0.9017,
+      "step": 12808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.731223216679504e-07,
+      "loss": 0.8302,
+      "step": 12809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7244282667951667e-07,
+      "loss": 0.84,
+      "step": 12810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.717638081794584e-07,
+      "loss": 0.8674,
+      "step": 12811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7108526620173514e-07,
+      "loss": 0.8297,
+      "step": 12812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7040720078027844e-07,
+      "loss": 0.7557,
+      "step": 12813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.697296119490047e-07,
+      "loss": 0.8332,
+      "step": 12814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.690524997417989e-07,
+      "loss": 0.8925,
+      "step": 12815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.68375864192524e-07,
+      "loss": 0.8651,
+      "step": 12816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.676997053350196e-07,
+      "loss": 0.8629,
+      "step": 12817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.670240232031009e-07,
+      "loss": 0.7722,
+      "step": 12818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6634881783056196e-07,
+      "loss": 0.8727,
+      "step": 12819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6567408925116906e-07,
+      "loss": 1.027,
+      "step": 12820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.649998374986686e-07,
+      "loss": 0.8328,
+      "step": 12821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.643260626067769e-07,
+      "loss": 0.9789,
+      "step": 12822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6365276460919485e-07,
+      "loss": 0.9623,
+      "step": 12823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6297994353959317e-07,
+      "loss": 0.8095,
+      "step": 12824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.623075994316217e-07,
+      "loss": 0.7983,
+      "step": 12825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.616357323189036e-07,
+      "loss": 0.811,
+      "step": 12826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6096434223504184e-07,
+      "loss": 0.9218,
+      "step": 12827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6029342921361186e-07,
+      "loss": 0.9111,
+      "step": 12828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5962299328816907e-07,
+      "loss": 0.6596,
+      "step": 12829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5895303449224214e-07,
+      "loss": 0.8044,
+      "step": 12830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5828355285933765e-07,
+      "loss": 0.9106,
+      "step": 12831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5761454842293664e-07,
+      "loss": 0.9501,
+      "step": 12832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.569460212164956e-07,
+      "loss": 0.8279,
+      "step": 12833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.562779712734511e-07,
+      "loss": 0.9124,
+      "step": 12834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5561039862721313e-07,
+      "loss": 0.856,
+      "step": 12835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5494330331116723e-07,
+      "loss": 0.6042,
+      "step": 12836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5427668535867553e-07,
+      "loss": 0.6908,
+      "step": 12837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5361054480307587e-07,
+      "loss": 0.8151,
+      "step": 12838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.52944881677686e-07,
+      "loss": 0.7705,
+      "step": 12839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.522796960157949e-07,
+      "loss": 0.582,
+      "step": 12840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5161498785067036e-07,
+      "loss": 0.7847,
+      "step": 12841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.509507572155547e-07,
+      "loss": 0.8364,
+      "step": 12842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5028700414366576e-07,
+      "loss": 0.8039,
+      "step": 12843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4962372866820147e-07,
+      "loss": 0.8938,
+      "step": 12844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4896093082233305e-07,
+      "loss": 0.872,
+      "step": 12845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.482986106392073e-07,
+      "loss": 0.8855,
+      "step": 12846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.476367681519467e-07,
+      "loss": 0.8536,
+      "step": 12847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4697540339365353e-07,
+      "loss": 0.7486,
+      "step": 12848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4631451639740255e-07,
+      "loss": 0.8619,
+      "step": 12849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4565410719624503e-07,
+      "loss": 0.8476,
+      "step": 12850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4499417582321017e-07,
+      "loss": 0.8676,
+      "step": 12851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.443347223113015e-07,
+      "loss": 0.9144,
+      "step": 12852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.436757466934993e-07,
+      "loss": 0.8192,
+      "step": 12853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.430172490027618e-07,
+      "loss": 0.7768,
+      "step": 12854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4235922927201913e-07,
+      "loss": 0.8927,
+      "step": 12855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.417016875341817e-07,
+      "loss": 0.8023,
+      "step": 12856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.410446238221322e-07,
+      "loss": 0.8371,
+      "step": 12857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.40388038168732e-07,
+      "loss": 0.8143,
+      "step": 12858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3973193060681927e-07,
+      "loss": 0.824,
+      "step": 12859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3907630116920787e-07,
+      "loss": 0.8187,
+      "step": 12860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384211498886837e-07,
+      "loss": 0.9272,
+      "step": 12861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.37766476798015e-07,
+      "loss": 0.8952,
+      "step": 12862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3711228192994005e-07,
+      "loss": 0.8832,
+      "step": 12863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.364585653171782e-07,
+      "loss": 0.869,
+      "step": 12864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3580532699242337e-07,
+      "loss": 0.978,
+      "step": 12865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3515256698834384e-07,
+      "loss": 0.785,
+      "step": 12866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.345002853375846e-07,
+      "loss": 0.8039,
+      "step": 12867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.338484820727673e-07,
+      "loss": 0.9208,
+      "step": 12868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.331971572264926e-07,
+      "loss": 0.8491,
+      "step": 12869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.32546310831331e-07,
+      "loss": 0.9216,
+      "step": 12870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3189594291983436e-07,
+      "loss": 0.925,
+      "step": 12871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3124605352452666e-07,
+      "loss": 0.9095,
+      "step": 12872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.305966426779118e-07,
+      "loss": 0.8119,
+      "step": 12873
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.299477104124672e-07,
+      "loss": 0.8273,
+      "step": 12874
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292992567606469e-07,
+      "loss": 0.8698,
+      "step": 12875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.286512817548805e-07,
+      "loss": 0.7379,
+      "step": 12876
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.280037854275743e-07,
+      "loss": 0.9556,
+      "step": 12877
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2735676781111235e-07,
+      "loss": 0.9219,
+      "step": 12878
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.267102289378522e-07,
+      "loss": 0.8062,
+      "step": 12879
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.260641688401279e-07,
+      "loss": 0.7773,
+      "step": 12880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2541858755025035e-07,
+      "loss": 0.8214,
+      "step": 12881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2477348510050364e-07,
+      "loss": 0.8488,
+      "step": 12882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2412886152315425e-07,
+      "loss": 0.5996,
+      "step": 12883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234847168504386e-07,
+      "loss": 0.7974,
+      "step": 12884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2284105111457306e-07,
+      "loss": 0.9386,
+      "step": 12885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2219786434774645e-07,
+      "loss": 0.8847,
+      "step": 12886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.215551565821252e-07,
+      "loss": 0.8948,
+      "step": 12887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2091292784985474e-07,
+      "loss": 0.8809,
+      "step": 12888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.202711781830515e-07,
+      "loss": 0.7486,
+      "step": 12889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1962990761381215e-07,
+      "loss": 0.8702,
+      "step": 12890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1898911617420656e-07,
+      "loss": 0.9333,
+      "step": 12891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.183488038962813e-07,
+      "loss": 0.8485,
+      "step": 12892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1770897081206184e-07,
+      "loss": 0.7707,
+      "step": 12893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1706961695354596e-07,
+      "loss": 0.8488,
+      "step": 12894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.16430742352707e-07,
+      "loss": 1.0037,
+      "step": 12895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.157923470414993e-07,
+      "loss": 0.8301,
+      "step": 12896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.151544310518452e-07,
+      "loss": 0.8257,
+      "step": 12897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.145169944156535e-07,
+      "loss": 0.7741,
+      "step": 12898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.138800371648011e-07,
+      "loss": 0.9873,
+      "step": 12899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1324355933114235e-07,
+      "loss": 0.8336,
+      "step": 12900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.126075609465097e-07,
+      "loss": 0.6386,
+      "step": 12901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1197204204270756e-07,
+      "loss": 0.863,
+      "step": 12902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.11337002651524e-07,
+      "loss": 0.8829,
+      "step": 12903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.107024428047157e-07,
+      "loss": 1.056,
+      "step": 12904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100683625340185e-07,
+      "loss": 0.8349,
+      "step": 12905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0943476187114363e-07,
+      "loss": 0.8073,
+      "step": 12906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0880164084777683e-07,
+      "loss": 0.8355,
+      "step": 12907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0816899949558506e-07,
+      "loss": 0.8826,
+      "step": 12908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0753683784620526e-07,
+      "loss": 0.8578,
+      "step": 12909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0690515593125315e-07,
+      "loss": 0.809,
+      "step": 12910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0627395378232126e-07,
+      "loss": 0.9083,
+      "step": 12911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.056432314309744e-07,
+      "loss": 0.8983,
+      "step": 12912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.050129889087584e-07,
+      "loss": 0.9036,
+      "step": 12913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.043832262471925e-07,
+      "loss": 0.8292,
+      "step": 12914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0375394347777154e-07,
+      "loss": 0.8569,
+      "step": 12915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.031251406319658e-07,
+      "loss": 0.9457,
+      "step": 12916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0249681774122583e-07,
+      "loss": 0.9309,
+      "step": 12917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.018689748369731e-07,
+      "loss": 0.866,
+      "step": 12918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0124161195060684e-07,
+      "loss": 0.9435,
+      "step": 12919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.006147291135032e-07,
+      "loss": 0.8987,
+      "step": 12920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9998832635701146e-07,
+      "loss": 0.7947,
+      "step": 12921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9936240371245993e-07,
+      "loss": 0.8902,
+      "step": 12922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.987369612111558e-07,
+      "loss": 0.8647,
+      "step": 12923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9811199888437403e-07,
+      "loss": 0.924,
+      "step": 12924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.974875167633729e-07,
+      "loss": 0.8344,
+      "step": 12925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.968635148793809e-07,
+      "loss": 0.9216,
+      "step": 12926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9623999326360627e-07,
+      "loss": 0.8706,
+      "step": 12927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9561695194723413e-07,
+      "loss": 0.7866,
+      "step": 12928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949943909614218e-07,
+      "loss": 0.7978,
+      "step": 12929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9437231033730537e-07,
+      "loss": 0.8783,
+      "step": 12930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9375071010599677e-07,
+      "loss": 0.9116,
+      "step": 12931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.931295902985799e-07,
+      "loss": 0.8603,
+      "step": 12932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9250895094612327e-07,
+      "loss": 0.8818,
+      "step": 12933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9188879207966213e-07,
+      "loss": 0.8462,
+      "step": 12934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.912691137302127e-07,
+      "loss": 0.6059,
+      "step": 12935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.906499159287658e-07,
+      "loss": 0.9281,
+      "step": 12936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.900311987062877e-07,
+      "loss": 0.7928,
+      "step": 12937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.894129620937248e-07,
+      "loss": 0.8437,
+      "step": 12938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.887952061219924e-07,
+      "loss": 0.9276,
+      "step": 12939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8817793082198796e-07,
+      "loss": 0.8571,
+      "step": 12940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8756113622457904e-07,
+      "loss": 0.7859,
+      "step": 12941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8694482236061646e-07,
+      "loss": 0.8564,
+      "step": 12942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8632898926092234e-07,
+      "loss": 0.843,
+      "step": 12943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8571363695629304e-07,
+      "loss": 0.8706,
+      "step": 12944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8509876547750624e-07,
+      "loss": 0.9681,
+      "step": 12945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.844843748553095e-07,
+      "loss": 0.9826,
+      "step": 12946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.838704651204317e-07,
+      "loss": 0.9676,
+      "step": 12947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.832570363035759e-07,
+      "loss": 0.8637,
+      "step": 12948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8264408843541877e-07,
+      "loss": 0.8085,
+      "step": 12949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.820316215466169e-07,
+      "loss": 0.9432,
+      "step": 12950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8141963566779795e-07,
+      "loss": 0.8622,
+      "step": 12951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.808081308295697e-07,
+      "loss": 1.0075,
+      "step": 12952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8019710706251655e-07,
+      "loss": 0.8353,
+      "step": 12953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.795865643971941e-07,
+      "loss": 0.7601,
+      "step": 12954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.789765028641379e-07,
+      "loss": 0.7762,
+      "step": 12955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.78366922493858e-07,
+      "loss": 0.8118,
+      "step": 12956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7775782331683776e-07,
+      "loss": 0.9313,
+      "step": 12957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7714920536354393e-07,
+      "loss": 0.8883,
+      "step": 12958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.765410686644122e-07,
+      "loss": 0.882,
+      "step": 12959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7593341324985487e-07,
+      "loss": 0.8513,
+      "step": 12960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7532623915026436e-07,
+      "loss": 0.7866,
+      "step": 12961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7471954639600405e-07,
+      "loss": 0.7396,
+      "step": 12962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7411333501741863e-07,
+      "loss": 0.8075,
+      "step": 12963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.735076050448239e-07,
+      "loss": 0.9105,
+      "step": 12964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7290235650851213e-07,
+      "loss": 0.863,
+      "step": 12965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.722975894387537e-07,
+      "loss": 0.8531,
+      "step": 12966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.716933038657955e-07,
+      "loss": 0.8653,
+      "step": 12967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7108949981985775e-07,
+      "loss": 0.8768,
+      "step": 12968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7048617733113856e-07,
+      "loss": 0.8335,
+      "step": 12969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6988333642980824e-07,
+      "loss": 0.8521,
+      "step": 12970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.692809771460182e-07,
+      "loss": 0.8565,
+      "step": 12971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6867909950989324e-07,
+      "loss": 0.9268,
+      "step": 12972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6807770355153373e-07,
+      "loss": 0.9241,
+      "step": 12973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.674767893010156e-07,
+      "loss": 0.9868,
+      "step": 12974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6687635678839373e-07,
+      "loss": 0.7878,
+      "step": 12975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.662764060436941e-07,
+      "loss": 0.7414,
+      "step": 12976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6567693709692374e-07,
+      "loss": 0.657,
+      "step": 12977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.650779499780621e-07,
+      "loss": 0.7355,
+      "step": 12978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.644794447170652e-07,
+      "loss": 0.9009,
+      "step": 12979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6388142134386464e-07,
+      "loss": 0.5842,
+      "step": 12980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.632838798883687e-07,
+      "loss": 0.8508,
+      "step": 12981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.626868203804623e-07,
+      "loss": 0.8822,
+      "step": 12982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6209024285000504e-07,
+      "loss": 0.645,
+      "step": 12983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6149414732683295e-07,
+      "loss": 0.9059,
+      "step": 12984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.608985338407567e-07,
+      "loss": 0.9066,
+      "step": 12985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6030340242156456e-07,
+      "loss": 0.8843,
+      "step": 12986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5970875309902065e-07,
+      "loss": 0.8556,
+      "step": 12987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.591145859028633e-07,
+      "loss": 0.8241,
+      "step": 12988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5852090086280877e-07,
+      "loss": 0.9137,
+      "step": 12989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5792769800854777e-07,
+      "loss": 0.8033,
+      "step": 12990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5733497736974543e-07,
+      "loss": 0.9412,
+      "step": 12991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5674273897604807e-07,
+      "loss": 0.966,
+      "step": 12992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5615098285707197e-07,
+      "loss": 0.8672,
+      "step": 12993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.555597090424123e-07,
+      "loss": 0.9435,
+      "step": 12994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.549689175616411e-07,
+      "loss": 0.9236,
+      "step": 12995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.543786084443013e-07,
+      "loss": 0.8262,
+      "step": 12996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5378878171991926e-07,
+      "loss": 0.9575,
+      "step": 12997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5319943741799147e-07,
+      "loss": 0.8558,
+      "step": 12998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.526105755679921e-07,
+      "loss": 0.935,
+      "step": 12999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.520221961993697e-07,
+      "loss": 0.5838,
+      "step": 13000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5143429934155094e-07,
+      "loss": 0.86,
+      "step": 13001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508468850239388e-07,
+      "loss": 0.9483,
+      "step": 13002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5025995327590987e-07,
+      "loss": 0.9215,
+      "step": 13003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4967350412681845e-07,
+      "loss": 0.7615,
+      "step": 13004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.49087537605991e-07,
+      "loss": 0.6002,
+      "step": 13005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4850205374273414e-07,
+      "loss": 0.6013,
+      "step": 13006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.479170525663311e-07,
+      "loss": 0.9146,
+      "step": 13007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473325341060352e-07,
+      "loss": 0.8676,
+      "step": 13008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4674849839108184e-07,
+      "loss": 0.9268,
+      "step": 13009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461649454506777e-07,
+      "loss": 0.8105,
+      "step": 13010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455818753140083e-07,
+      "loss": 0.8186,
+      "step": 13011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4499928801023374e-07,
+      "loss": 0.8542,
+      "step": 13012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.444171835684895e-07,
+      "loss": 0.8124,
+      "step": 13013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4383556201788905e-07,
+      "loss": 0.8671,
+      "step": 13014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.432544233875168e-07,
+      "loss": 0.8968,
+      "step": 13015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4267376770643955e-07,
+      "loss": 0.8131,
+      "step": 13016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4209359500369744e-07,
+      "loss": 0.8596,
+      "step": 13017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.41513905308305e-07,
+      "loss": 0.8065,
+      "step": 13018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.409346986492523e-07,
+      "loss": 0.9562,
+      "step": 13019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.403559750555063e-07,
+      "loss": 0.9168,
+      "step": 13020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.397777345560105e-07,
+      "loss": 0.95,
+      "step": 13021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.39199977179685e-07,
+      "loss": 0.5908,
+      "step": 13022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.386227029554223e-07,
+      "loss": 0.96,
+      "step": 13023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380459119120938e-07,
+      "loss": 0.8993,
+      "step": 13024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3746960407854633e-07,
+      "loss": 0.7443,
+      "step": 13025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3689377948359916e-07,
+      "loss": 0.8624,
+      "step": 13026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3631843815605357e-07,
+      "loss": 0.8566,
+      "step": 13027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.357435801246822e-07,
+      "loss": 0.919,
+      "step": 13028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.351692054182354e-07,
+      "loss": 0.9513,
+      "step": 13029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3459531406543566e-07,
+      "loss": 0.8617,
+      "step": 13030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3402190609498675e-07,
+      "loss": 0.8418,
+      "step": 13031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.334489815355646e-07,
+      "loss": 0.8181,
+      "step": 13032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328765404158241e-07,
+      "loss": 0.9217,
+      "step": 13033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3230458276439115e-07,
+      "loss": 0.8837,
+      "step": 13034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3173310860987073e-07,
+      "loss": 0.7882,
+      "step": 13035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3116211798084553e-07,
+      "loss": 0.8783,
+      "step": 13036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3059161090586935e-07,
+      "loss": 0.8555,
+      "step": 13037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.30021587413476e-07,
+      "loss": 0.9333,
+      "step": 13038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294520475321716e-07,
+      "loss": 0.8909,
+      "step": 13039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.28882991290439e-07,
+      "loss": 0.8054,
+      "step": 13040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2831441871674085e-07,
+      "loss": 0.7787,
+      "step": 13041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2774632983951005e-07,
+      "loss": 0.9043,
+      "step": 13042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2717872468715826e-07,
+      "loss": 0.8075,
+      "step": 13043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2661160328807175e-07,
+      "loss": 0.8803,
+      "step": 13044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.260449656706133e-07,
+      "loss": 0.9055,
+      "step": 13045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.254788118631202e-07,
+      "loss": 0.8217,
+      "step": 13046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.249131418939111e-07,
+      "loss": 0.8801,
+      "step": 13047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.243479557912721e-07,
+      "loss": 0.8544,
+      "step": 13048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2378325358346954e-07,
+      "loss": 0.8706,
+      "step": 13049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.232190352987463e-07,
+      "loss": 0.9142,
+      "step": 13050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.226553009653188e-07,
+      "loss": 0.815,
+      "step": 13051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2209205061138004e-07,
+      "loss": 0.9002,
+      "step": 13052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2152928426510076e-07,
+      "loss": 1.0054,
+      "step": 13053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2096700195462516e-07,
+      "loss": 0.8292,
+      "step": 13054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.204052037080718e-07,
+      "loss": 0.7892,
+      "step": 13055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.198438895535394e-07,
+      "loss": 0.627,
+      "step": 13056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1928305951909987e-07,
+      "loss": 0.7401,
+      "step": 13057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187227136328008e-07,
+      "loss": 0.8878,
+      "step": 13058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.181628519226665e-07,
+      "loss": 0.899,
+      "step": 13059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1760347441669446e-07,
+      "loss": 0.8835,
+      "step": 13060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.170445811428635e-07,
+      "loss": 0.7903,
+      "step": 13061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1648617212912237e-07,
+      "loss": 0.8355,
+      "step": 13062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.159282474033987e-07,
+      "loss": 0.828,
+      "step": 13063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.153708069935946e-07,
+      "loss": 0.8484,
+      "step": 13064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.148138509275889e-07,
+      "loss": 0.9679,
+      "step": 13065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1425737923323595e-07,
+      "loss": 0.613,
+      "step": 13066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.137013919383669e-07,
+      "loss": 0.9355,
+      "step": 13067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1314588907078616e-07,
+      "loss": 0.8961,
+      "step": 13068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.125908706582759e-07,
+      "loss": 0.8657,
+      "step": 13069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1203633672859167e-07,
+      "loss": 0.8415,
+      "step": 13070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1148228730946917e-07,
+      "loss": 0.8397,
+      "step": 13071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1092872242861727e-07,
+      "loss": 0.603,
+      "step": 13072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.103756421137183e-07,
+      "loss": 0.9008,
+      "step": 13073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0982304639243454e-07,
+      "loss": 0.8566,
+      "step": 13074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.092709352924017e-07,
+      "loss": 0.9162,
+      "step": 13075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0871930884123215e-07,
+      "loss": 0.8591,
+      "step": 13076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.081681670665115e-07,
+      "loss": 0.8297,
+      "step": 13077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0761750999580673e-07,
+      "loss": 0.7578,
+      "step": 13078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0706733765665465e-07,
+      "loss": 0.8542,
+      "step": 13079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0651765007656984e-07,
+      "loss": 1.0077,
+      "step": 13080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0596844728304597e-07,
+      "loss": 0.908,
+      "step": 13081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0541972930354766e-07,
+      "loss": 0.5791,
+      "step": 13082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0487149616551636e-07,
+      "loss": 0.9163,
+      "step": 13083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.043237478963712e-07,
+      "loss": 0.9753,
+      "step": 13084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0377648452350584e-07,
+      "loss": 0.8145,
+      "step": 13085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.032297060742895e-07,
+      "loss": 0.9588,
+      "step": 13086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0268341257606695e-07,
+      "loss": 0.7943,
+      "step": 13087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.021376040561608e-07,
+      "loss": 0.7708,
+      "step": 13088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0159228054186694e-07,
+      "loss": 1.0248,
+      "step": 13089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.01047442060457e-07,
+      "loss": 0.9319,
+      "step": 13090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.005030886391802e-07,
+      "loss": 0.8363,
+      "step": 13091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.999592203052604e-07,
+      "loss": 0.8916,
+      "step": 13092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9941583708589793e-07,
+      "loss": 0.8376,
+      "step": 13093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9887293900826677e-07,
+      "loss": 0.955,
+      "step": 13094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.983305260995184e-07,
+      "loss": 0.9337,
+      "step": 13095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.977885983867812e-07,
+      "loss": 0.9091,
+      "step": 13096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.972471558971568e-07,
+      "loss": 0.589,
+      "step": 13097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9670619865772356e-07,
+      "loss": 0.7729,
+      "step": 13098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9616572669553644e-07,
+      "loss": 0.8242,
+      "step": 13099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.956257400376228e-07,
+      "loss": 0.8063,
+      "step": 13100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.950862387109921e-07,
+      "loss": 0.8643,
+      "step": 13101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.945472227426227e-07,
+      "loss": 0.8529,
+      "step": 13102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9400869215947316e-07,
+      "loss": 0.8358,
+      "step": 13103
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9347064698847406e-07,
+      "loss": 0.8113,
+      "step": 13104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9293308725653726e-07,
+      "loss": 0.8811,
+      "step": 13105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9239601299054567e-07,
+      "loss": 0.7495,
+      "step": 13106
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.918594242173578e-07,
+      "loss": 0.8247,
+      "step": 13107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.913233209638111e-07,
+      "loss": 0.8581,
+      "step": 13108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9078770325671414e-07,
+      "loss": 0.8954,
+      "step": 13109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.902525711228565e-07,
+      "loss": 0.8146,
+      "step": 13110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.897179245890025e-07,
+      "loss": 0.8854,
+      "step": 13111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.891837636818873e-07,
+      "loss": 0.9126,
+      "step": 13112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.886500884282273e-07,
+      "loss": 0.8694,
+      "step": 13113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.881168988547123e-07,
+      "loss": 0.8952,
+      "step": 13114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.875841949880054e-07,
+      "loss": 0.7449,
+      "step": 13115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8705197685475195e-07,
+      "loss": 0.9656,
+      "step": 13116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.865202444815662e-07,
+      "loss": 0.985,
+      "step": 13117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.859889978950414e-07,
+      "loss": 0.9086,
+      "step": 13118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8545823712174735e-07,
+      "loss": 0.9023,
+      "step": 13119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8492796218822614e-07,
+      "loss": 0.8262,
+      "step": 13120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8439817312099995e-07,
+      "loss": 0.7614,
+      "step": 13121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8386886994656306e-07,
+      "loss": 0.8516,
+      "step": 13122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8334005269138654e-07,
+      "loss": 0.8673,
+      "step": 13123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8281172138191813e-07,
+      "loss": 0.8815,
+      "step": 13124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8228387604457896e-07,
+      "loss": 0.8723,
+      "step": 13125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.817565167057701e-07,
+      "loss": 0.9132,
+      "step": 13126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.812296433918649e-07,
+      "loss": 0.9982,
+      "step": 13127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8070325612921114e-07,
+      "loss": 0.8888,
+      "step": 13128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.801773549441356e-07,
+      "loss": 0.9786,
+      "step": 13129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7965193986293936e-07,
+      "loss": 0.9308,
+      "step": 13130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.791270109119004e-07,
+      "loss": 0.9182,
+      "step": 13131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7860256811726993e-07,
+      "loss": 0.8869,
+      "step": 13132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.780786115052758e-07,
+      "loss": 0.898,
+      "step": 13133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.775551411021227e-07,
+      "loss": 0.9159,
+      "step": 13134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7703215693399065e-07,
+      "loss": 0.8572,
+      "step": 13135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.765096590270344e-07,
+      "loss": 0.8324,
+      "step": 13136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7598764740738416e-07,
+      "loss": 1.0257,
+      "step": 13137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.754661221011479e-07,
+      "loss": 0.789,
+      "step": 13138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.749450831344058e-07,
+      "loss": 0.765,
+      "step": 13139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7442453053321826e-07,
+      "loss": 0.7336,
+      "step": 13140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.739044643236177e-07,
+      "loss": 1.0429,
+      "step": 13141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733848845316156e-07,
+      "loss": 0.8975,
+      "step": 13142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7286579118319333e-07,
+      "loss": 0.8317,
+      "step": 13143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.723471843043146e-07,
+      "loss": 0.6338,
+      "step": 13144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7182906392091313e-07,
+      "loss": 0.7642,
+      "step": 13145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.713114300589037e-07,
+      "loss": 0.9622,
+      "step": 13146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.707942827441734e-07,
+      "loss": 0.7043,
+      "step": 13147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.702776220025838e-07,
+      "loss": 0.7863,
+      "step": 13148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6976144785997414e-07,
+      "loss": 0.8502,
+      "step": 13149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6924576034216164e-07,
+      "loss": 0.9753,
+      "step": 13150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6873055947493456e-07,
+      "loss": 0.9985,
+      "step": 13151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6821584528406e-07,
+      "loss": 0.835,
+      "step": 13152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.677016177952785e-07,
+      "loss": 0.7917,
+      "step": 13153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6718787703430724e-07,
+      "loss": 0.8214,
+      "step": 13154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6667462302684134e-07,
+      "loss": 0.8784,
+      "step": 13155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.661618557985468e-07,
+      "loss": 0.756,
+      "step": 13156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6564957537507095e-07,
+      "loss": 0.8648,
+      "step": 13157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6513778178203e-07,
+      "loss": 0.8848,
+      "step": 13158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6462647504502227e-07,
+      "loss": 0.7728,
+      "step": 13159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6411565518961846e-07,
+      "loss": 0.8808,
+      "step": 13160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6360532224136483e-07,
+      "loss": 0.8452,
+      "step": 13161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6309547622578533e-07,
+      "loss": 0.9847,
+      "step": 13162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6258611716837746e-07,
+      "loss": 0.8703,
+      "step": 13163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.62077245094613e-07,
+      "loss": 0.87,
+      "step": 13164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6156886002994486e-07,
+      "loss": 0.9753,
+      "step": 13165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6106096199979616e-07,
+      "loss": 0.8493,
+      "step": 13166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605535510295687e-07,
+      "loss": 0.8568,
+      "step": 13167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.600466271446389e-07,
+      "loss": 0.8532,
+      "step": 13168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.595401903703565e-07,
+      "loss": 0.8772,
+      "step": 13169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.590342407320523e-07,
+      "loss": 0.9353,
+      "step": 13170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.585287782550283e-07,
+      "loss": 0.641,
+      "step": 13171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5802380296456433e-07,
+      "loss": 0.9475,
+      "step": 13172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.575193148859134e-07,
+      "loss": 0.9452,
+      "step": 13173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5701531404430655e-07,
+      "loss": 0.9704,
+      "step": 13174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.565118004649503e-07,
+      "loss": 0.9301,
+      "step": 13175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.560087741730255e-07,
+      "loss": 0.9002,
+      "step": 13176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.555062351936899e-07,
+      "loss": 0.9044,
+      "step": 13177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5500418355207556e-07,
+      "loss": 0.8677,
+      "step": 13178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5450261927329135e-07,
+      "loss": 0.7658,
+      "step": 13179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5400154238242045e-07,
+      "loss": 0.8718,
+      "step": 13180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5350095290452404e-07,
+      "loss": 0.9001,
+      "step": 13181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.530008508646364e-07,
+      "loss": 0.8929,
+      "step": 13182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.525012362877688e-07,
+      "loss": 0.7699,
+      "step": 13183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.520021091989067e-07,
+      "loss": 0.586,
+      "step": 13184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.515034696230134e-07,
+      "loss": 0.8285,
+      "step": 13185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.510053175850269e-07,
+      "loss": 0.9047,
+      "step": 13186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5050765310985937e-07,
+      "loss": 0.8848,
+      "step": 13187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.500104762224009e-07,
+      "loss": 0.8851,
+      "step": 13188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49513786947515e-07,
+      "loss": 0.8624,
+      "step": 13189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4901758531004293e-07,
+      "loss": 1.0132,
+      "step": 13190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.485218713348003e-07,
+      "loss": 0.922,
+      "step": 13191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.480266450465785e-07,
+      "loss": 0.9709,
+      "step": 13192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4753190647014535e-07,
+      "loss": 0.7342,
+      "step": 13193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.470376556302412e-07,
+      "loss": 0.8254,
+      "step": 13194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.465438925515862e-07,
+      "loss": 0.7728,
+      "step": 13195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4605061725887394e-07,
+      "loss": 0.9469,
+      "step": 13196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.455578297767747e-07,
+      "loss": 0.9694,
+      "step": 13197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4506553012993093e-07,
+      "loss": 0.9698,
+      "step": 13198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.445737183429664e-07,
+      "loss": 0.8738,
+      "step": 13199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4408239444047577e-07,
+      "loss": 0.8316,
+      "step": 13200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4359155844703054e-07,
+      "loss": 0.9186,
+      "step": 13201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.431012103871799e-07,
+      "loss": 0.8012,
+      "step": 13202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4261135028544433e-07,
+      "loss": 0.918,
+      "step": 13203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.421219781663253e-07,
+      "loss": 0.7829,
+      "step": 13204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.416330940542955e-07,
+      "loss": 0.6262,
+      "step": 13205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4114469797380635e-07,
+      "loss": 0.9466,
+      "step": 13206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4065678994928177e-07,
+      "loss": 0.8492,
+      "step": 13207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.401693700051233e-07,
+      "loss": 0.8793,
+      "step": 13208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.396824381657059e-07,
+      "loss": 0.8874,
+      "step": 13209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3919599445538566e-07,
+      "loss": 0.8297,
+      "step": 13210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3871003889848753e-07,
+      "loss": 0.8293,
+      "step": 13211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3822457151931543e-07,
+      "loss": 0.9454,
+      "step": 13212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3773959234214883e-07,
+      "loss": 0.94,
+      "step": 13213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3725510139124053e-07,
+      "loss": 0.8873,
+      "step": 13214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3677109869082338e-07,
+      "loss": 0.8531,
+      "step": 13215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3628758426510244e-07,
+      "loss": 1.0119,
+      "step": 13216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3580455813825732e-07,
+      "loss": 0.8258,
+      "step": 13217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.353220203344475e-07,
+      "loss": 0.8206,
+      "step": 13218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3483997087780264e-07,
+      "loss": 0.7306,
+      "step": 13219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.343584097924323e-07,
+      "loss": 0.8008,
+      "step": 13220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3387733710242166e-07,
+      "loss": 0.9429,
+      "step": 13221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3339675283182705e-07,
+      "loss": 0.824,
+      "step": 13222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329166570046848e-07,
+      "loss": 0.6358,
+      "step": 13223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3243704964500458e-07,
+      "loss": 0.9025,
+      "step": 13224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3195793077677387e-07,
+      "loss": 0.7378,
+      "step": 13225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.314793004239535e-07,
+      "loss": 0.8327,
+      "step": 13226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3100115861047878e-07,
+      "loss": 0.8306,
+      "step": 13227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.305235053602639e-07,
+      "loss": 0.8647,
+      "step": 13228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3004634069719646e-07,
+      "loss": 0.836,
+      "step": 13229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2956966464514175e-07,
+      "loss": 0.8519,
+      "step": 13230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290934772279374e-07,
+      "loss": 0.8846,
+      "step": 13231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.286177784693988e-07,
+      "loss": 0.898,
+      "step": 13232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2814256839331583e-07,
+      "loss": 0.8923,
+      "step": 13233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2766784702345611e-07,
+      "loss": 0.9307,
+      "step": 13234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271936143835596e-07,
+      "loss": 0.9218,
+      "step": 13235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2671987049734502e-07,
+      "loss": 0.8649,
+      "step": 13236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2624661538850345e-07,
+      "loss": 0.9186,
+      "step": 13237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2577384908070488e-07,
+      "loss": 0.9385,
+      "step": 13238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2530157159759037e-07,
+      "loss": 0.7622,
+      "step": 13239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2482978296278213e-07,
+      "loss": 0.9123,
+      "step": 13240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2435848319987353e-07,
+      "loss": 0.9386,
+      "step": 13241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2388767233243568e-07,
+      "loss": 0.6137,
+      "step": 13242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2341735038401313e-07,
+      "loss": 0.8483,
+      "step": 13243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2294751737812926e-07,
+      "loss": 0.9821,
+      "step": 13244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224781733382808e-07,
+      "loss": 0.8478,
+      "step": 13245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2200931828794014e-07,
+      "loss": 0.9541,
+      "step": 13246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2154095225055629e-07,
+      "loss": 0.8481,
+      "step": 13247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2107307524955046e-07,
+      "loss": 0.8195,
+      "step": 13248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.20605687308324e-07,
+      "loss": 0.8543,
+      "step": 13249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.201387884502515e-07,
+      "loss": 0.9287,
+      "step": 13250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1967237869868318e-07,
+      "loss": 0.7893,
+      "step": 13251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1920645807694485e-07,
+      "loss": 0.8859,
+      "step": 13252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1874102660833785e-07,
+      "loss": 0.8795,
+      "step": 13253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1827608431613911e-07,
+      "loss": 0.9066,
+      "step": 13254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1781163122360228e-07,
+      "loss": 0.896,
+      "step": 13255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.173476673539543e-07,
+      "loss": 0.8059,
+      "step": 13256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1688419273039885e-07,
+      "loss": 0.8374,
+      "step": 13257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1642120737611295e-07,
+      "loss": 0.9263,
+      "step": 13258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1595871131425583e-07,
+      "loss": 0.898,
+      "step": 13259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1549670456795457e-07,
+      "loss": 0.8796,
+      "step": 13260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1503518716031513e-07,
+      "loss": 0.8479,
+      "step": 13261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1457415911442014e-07,
+      "loss": 0.9293,
+      "step": 13262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.141136204533234e-07,
+      "loss": 0.9395,
+      "step": 13263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.136535712000598e-07,
+      "loss": 0.8579,
+      "step": 13264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1319401137763652e-07,
+      "loss": 0.8721,
+      "step": 13265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1273494100903734e-07,
+      "loss": 0.8227,
+      "step": 13266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1227636011722065e-07,
+      "loss": 0.8862,
+      "step": 13267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1181826872511912e-07,
+      "loss": 0.7712,
+      "step": 13268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.113606668556456e-07,
+      "loss": 0.8188,
+      "step": 13269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1090355453168397e-07,
+      "loss": 0.942,
+      "step": 13270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.10446931776096e-07,
+      "loss": 0.8577,
+      "step": 13271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0999079861171669e-07,
+      "loss": 0.7737,
+      "step": 13272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.095351550613578e-07,
+      "loss": 0.5981,
+      "step": 13273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.090800011478089e-07,
+      "loss": 0.5936,
+      "step": 13274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.086253368938318e-07,
+      "loss": 0.851,
+      "step": 13275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0817116232216496e-07,
+      "loss": 0.8817,
+      "step": 13276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0771747745552129e-07,
+      "loss": 1.0184,
+      "step": 13277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0726428231659157e-07,
+      "loss": 0.988,
+      "step": 13278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068115769280421e-07,
+      "loss": 0.7599,
+      "step": 13279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0635936131251144e-07,
+      "loss": 0.9353,
+      "step": 13280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0590763549261593e-07,
+      "loss": 0.8945,
+      "step": 13281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.054563994909464e-07,
+      "loss": 0.8372,
+      "step": 13282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.050056533300715e-07,
+      "loss": 0.798,
+      "step": 13283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0455539703253314e-07,
+      "loss": 0.9931,
+      "step": 13284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0410563062084886e-07,
+      "loss": 0.996,
+      "step": 13285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0365635411751295e-07,
+      "loss": 0.8344,
+      "step": 13286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0320756754499404e-07,
+      "loss": 0.8781,
+      "step": 13287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0275927092573644e-07,
+      "loss": 0.8461,
+      "step": 13288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0231146428215997e-07,
+      "loss": 0.8504,
+      "step": 13289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.018641476366623e-07,
+      "loss": 0.8483,
+      "step": 13290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0141732101161106e-07,
+      "loss": 0.8783,
+      "step": 13291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0097098442935502e-07,
+      "loss": 0.9018,
+      "step": 13292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.005251379122164e-07,
+      "loss": 0.5922,
+      "step": 13293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0007978148249064e-07,
+      "loss": 0.8549,
+      "step": 13294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9963491516245325e-07,
+      "loss": 0.8213,
+      "step": 13295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.991905389743509e-07,
+      "loss": 0.9279,
+      "step": 13296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9874665294040807e-07,
+      "loss": 0.7853,
+      "step": 13297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.983032570828236e-07,
+      "loss": 0.9169,
+      "step": 13298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9786035142377424e-07,
+      "loss": 0.9373,
+      "step": 13299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9741793598540894e-07,
+      "loss": 0.5905,
+      "step": 13300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.969760107898544e-07,
+      "loss": 0.903,
+      "step": 13301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9653457585921188e-07,
+      "loss": 0.8728,
+      "step": 13302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9609363121555701e-07,
+      "loss": 0.9424,
+      "step": 13303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9565317688094332e-07,
+      "loss": 0.8096,
+      "step": 13304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952132128773987e-07,
+      "loss": 0.8775,
+      "step": 13305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9477373922692555e-07,
+      "loss": 0.7902,
+      "step": 13306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.943347559515041e-07,
+      "loss": 0.8781,
+      "step": 13307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9389626307308563e-07,
+      "loss": 0.9484,
+      "step": 13308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9345826061360374e-07,
+      "loss": 0.8548,
+      "step": 13309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.93020748594962e-07,
+      "loss": 0.9861,
+      "step": 13310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9258372703904072e-07,
+      "loss": 0.7653,
+      "step": 13311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.921471959676957e-07,
+      "loss": 0.9182,
+      "step": 13312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9171115540275953e-07,
+      "loss": 0.7683,
+      "step": 13313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9127560536603917e-07,
+      "loss": 0.8035,
+      "step": 13314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.908405458793161e-07,
+      "loss": 0.825,
+      "step": 13315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.904059769643507e-07,
+      "loss": 0.7358,
+      "step": 13316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8997189864287225e-07,
+      "loss": 0.7633,
+      "step": 13317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8953831093659448e-07,
+      "loss": 0.8475,
+      "step": 13318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8910521386719893e-07,
+      "loss": 0.9544,
+      "step": 13319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.886726074563472e-07,
+      "loss": 0.8797,
+      "step": 13320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8824049172567304e-07,
+      "loss": 0.807,
+      "step": 13321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8780886669678699e-07,
+      "loss": 0.8773,
+      "step": 13322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8737773239127732e-07,
+      "loss": 0.8728,
+      "step": 13323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8694708883070455e-07,
+      "loss": 0.8509,
+      "step": 13324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.865169360366048e-07,
+      "loss": 0.8932,
+      "step": 13325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8608727403049309e-07,
+      "loss": 0.8264,
+      "step": 13326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8565810283385556e-07,
+      "loss": 0.8036,
+      "step": 13327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.85229422468155e-07,
+      "loss": 0.9216,
+      "step": 13328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.848012329548343e-07,
+      "loss": 0.8599,
+      "step": 13329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8437353431530413e-07,
+      "loss": 0.8313,
+      "step": 13330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8394632657095623e-07,
+      "loss": 0.8511,
+      "step": 13331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8351960974315574e-07,
+      "loss": 0.7289,
+      "step": 13332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8309338385324117e-07,
+      "loss": 0.8225,
+      "step": 13333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.826676489225321e-07,
+      "loss": 0.9179,
+      "step": 13334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8224240497231927e-07,
+      "loss": 0.8621,
+      "step": 13335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8181765202386902e-07,
+      "loss": 0.8089,
+      "step": 13336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8139339009842327e-07,
+      "loss": 0.8178,
+      "step": 13337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8096961921720167e-07,
+      "loss": 0.6279,
+      "step": 13338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.805463394013973e-07,
+      "loss": 0.8696,
+      "step": 13339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8012355067217879e-07,
+      "loss": 0.889,
+      "step": 13340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.797012530506903e-07,
+      "loss": 0.9337,
+      "step": 13341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7927944655805162e-07,
+      "loss": 0.8227,
+      "step": 13342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.788581312153581e-07,
+      "loss": 0.979,
+      "step": 13343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7843730704368067e-07,
+      "loss": 0.9438,
+      "step": 13344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7801697406406582e-07,
+      "loss": 0.5935,
+      "step": 13345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.775971322975334e-07,
+      "loss": 0.9702,
+      "step": 13346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7717778176508105e-07,
+      "loss": 0.7917,
+      "step": 13347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.76758922487682e-07,
+      "loss": 0.8282,
+      "step": 13348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7634055448628395e-07,
+      "loss": 0.8549,
+      "step": 13349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7592267778181017e-07,
+      "loss": 0.8438,
+      "step": 13350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7550529239515835e-07,
+      "loss": 0.8809,
+      "step": 13351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7508839834720182e-07,
+      "loss": 0.9177,
+      "step": 13352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.746719956587939e-07,
+      "loss": 0.7813,
+      "step": 13353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.742560843507557e-07,
+      "loss": 0.8521,
+      "step": 13354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7384066444388947e-07,
+      "loss": 0.9033,
+      "step": 13355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7342573595897084e-07,
+      "loss": 0.8724,
+      "step": 13356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7301129891674983e-07,
+      "loss": 0.8559,
+      "step": 13357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7259735333795546e-07,
+      "loss": 0.7918,
+      "step": 13358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7218389924328783e-07,
+      "loss": 0.6243,
+      "step": 13359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7177093665342482e-07,
+      "loss": 0.9307,
+      "step": 13360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7135846558901992e-07,
+      "loss": 0.9619,
+      "step": 13361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7094648607069998e-07,
+      "loss": 0.8108,
+      "step": 13362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705349981190707e-07,
+      "loss": 0.631,
+      "step": 13363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7012400175471123e-07,
+      "loss": 0.8568,
+      "step": 13364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6971349699817508e-07,
+      "loss": 0.7682,
+      "step": 13365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.693034838699925e-07,
+      "loss": 0.8899,
+      "step": 13366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6889396239066824e-07,
+      "loss": 0.8373,
+      "step": 13367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6848493258068477e-07,
+      "loss": 0.8642,
+      "step": 13368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6807639446049685e-07,
+      "loss": 0.7718,
+      "step": 13369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.67668348050537e-07,
+      "loss": 0.8124,
+      "step": 13370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6726079337121226e-07,
+      "loss": 0.9762,
+      "step": 13371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.668537304429041e-07,
+      "loss": 0.8809,
+      "step": 13372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6644715928597066e-07,
+      "loss": 0.9685,
+      "step": 13373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.660410799207468e-07,
+      "loss": 0.8266,
+      "step": 13374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.656354923675396e-07,
+      "loss": 0.9371,
+      "step": 13375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.65230396646634e-07,
+      "loss": 0.8321,
+      "step": 13376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6482579277828702e-07,
+      "loss": 0.8141,
+      "step": 13377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6442168078273814e-07,
+      "loss": 0.8392,
+      "step": 13378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6401806068019333e-07,
+      "loss": 0.9001,
+      "step": 13379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6361493249084092e-07,
+      "loss": 0.7983,
+      "step": 13380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6321229623484148e-07,
+      "loss": 0.7618,
+      "step": 13381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6281015193232996e-07,
+      "loss": 0.9342,
+      "step": 13382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.624084996034203e-07,
+      "loss": 0.8638,
+      "step": 13383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6200733926819868e-07,
+      "loss": 0.8457,
+      "step": 13384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6160667094672787e-07,
+      "loss": 0.7219,
+      "step": 13385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6120649465904525e-07,
+      "loss": 0.8881,
+      "step": 13386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080681042516588e-07,
+      "loss": 0.9043,
+      "step": 13387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040761826507711e-07,
+      "loss": 0.8307,
+      "step": 13388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.600089181987452e-07,
+      "loss": 0.8732,
+      "step": 13389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5961071024610753e-07,
+      "loss": 0.613,
+      "step": 13390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5921299442707927e-07,
+      "loss": 0.6265,
+      "step": 13391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.588157707615523e-07,
+      "loss": 0.9201,
+      "step": 13392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5841903926939183e-07,
+      "loss": 0.7579,
+      "step": 13393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5802279997043868e-07,
+      "loss": 0.8769,
+      "step": 13394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.576270528845103e-07,
+      "loss": 0.8753,
+      "step": 13395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5723179803139755e-07,
+      "loss": 0.9084,
+      "step": 13396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5683703543086683e-07,
+      "loss": 0.8826,
+      "step": 13397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5644276510266343e-07,
+      "loss": 0.9039,
+      "step": 13398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5604898706650495e-07,
+      "loss": 0.8418,
+      "step": 13399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5565570134208342e-07,
+      "loss": 0.9444,
+      "step": 13400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.552629079490675e-07,
+      "loss": 0.8784,
+      "step": 13401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5487060690710266e-07,
+      "loss": 0.8709,
+      "step": 13402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.544787982358087e-07,
+      "loss": 0.8594,
+      "step": 13403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5408748195477995e-07,
+      "loss": 0.8742,
+      "step": 13404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5369665808358636e-07,
+      "loss": 1.0162,
+      "step": 13405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5330632664177447e-07,
+      "loss": 0.889,
+      "step": 13406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5291648764886536e-07,
+      "loss": 0.872,
+      "step": 13407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.525271411243545e-07,
+      "loss": 0.6531,
+      "step": 13408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5213828708771528e-07,
+      "loss": 0.9067,
+      "step": 13409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.517499255583932e-07,
+      "loss": 0.5795,
+      "step": 13410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.513620565558116e-07,
+      "loss": 0.9225,
+      "step": 13411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5097468009936945e-07,
+      "loss": 0.7777,
+      "step": 13412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.505877962084379e-07,
+      "loss": 0.7496,
+      "step": 13413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5020140490236812e-07,
+      "loss": 0.9227,
+      "step": 13414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4981550620048246e-07,
+      "loss": 0.8293,
+      "step": 13415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4943010012207992e-07,
+      "loss": 0.9089,
+      "step": 13416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.490451866864373e-07,
+      "loss": 0.8845,
+      "step": 13417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4866076591280366e-07,
+      "loss": 0.8251,
+      "step": 13418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.482768378204036e-07,
+      "loss": 0.8298,
+      "step": 13419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4789340242843952e-07,
+      "loss": 0.789,
+      "step": 13420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4751045975608615e-07,
+      "loss": 0.9017,
+      "step": 13421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4712800982249475e-07,
+      "loss": 0.9348,
+      "step": 13422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4674605264679564e-07,
+      "loss": 0.8149,
+      "step": 13423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4636458824808797e-07,
+      "loss": 0.8814,
+      "step": 13424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.459836166454498e-07,
+      "loss": 0.8414,
+      "step": 13425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.456031378579348e-07,
+      "loss": 0.9056,
+      "step": 13426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4522315190457104e-07,
+      "loss": 0.8364,
+      "step": 13427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4484365880436223e-07,
+      "loss": 0.8324,
+      "step": 13428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4446465857628767e-07,
+      "loss": 0.853,
+      "step": 13429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4408615123930102e-07,
+      "loss": 0.723,
+      "step": 13430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4370813681233276e-07,
+      "loss": 0.9799,
+      "step": 13431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4333061531428772e-07,
+      "loss": 0.8474,
+      "step": 13432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4295358676404747e-07,
+      "loss": 0.9871,
+      "step": 13433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4257705118046584e-07,
+      "loss": 0.8762,
+      "step": 13434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.422010085823744e-07,
+      "loss": 0.9126,
+      "step": 13435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4182545898858037e-07,
+      "loss": 0.79,
+      "step": 13436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4145040241786646e-07,
+      "loss": 0.796,
+      "step": 13437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.410758388889877e-07,
+      "loss": 0.8168,
+      "step": 13438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4070176842067795e-07,
+      "loss": 0.8153,
+      "step": 13439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4032819103164564e-07,
+      "loss": 0.6054,
+      "step": 13440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3995510674057133e-07,
+      "loss": 0.9001,
+      "step": 13441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.395825155661168e-07,
+      "loss": 0.8312,
+      "step": 13442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3921041752691378e-07,
+      "loss": 0.9091,
+      "step": 13443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3883881264157184e-07,
+      "loss": 0.973,
+      "step": 13444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3846770092867723e-07,
+      "loss": 0.8682,
+      "step": 13445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.380970824067873e-07,
+      "loss": 0.8388,
+      "step": 13446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.377269570944384e-07,
+      "loss": 0.888,
+      "step": 13447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.373573250101423e-07,
+      "loss": 0.8634,
+      "step": 13448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.369881861723832e-07,
+      "loss": 0.7775,
+      "step": 13449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.366195405996229e-07,
+      "loss": 0.7836,
+      "step": 13450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3625138831029673e-07,
+      "loss": 0.8995,
+      "step": 13451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3588372932281879e-07,
+      "loss": 0.644,
+      "step": 13452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3551656365557553e-07,
+      "loss": 0.5903,
+      "step": 13453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.351498913269289e-07,
+      "loss": 0.5739,
+      "step": 13454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3478371235521649e-07,
+      "loss": 0.9087,
+      "step": 13455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.344180267587525e-07,
+      "loss": 0.8139,
+      "step": 13456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3405283455582564e-07,
+      "loss": 0.8122,
+      "step": 13457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3368813576469908e-07,
+      "loss": 0.8797,
+      "step": 13458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.333239304036116e-07,
+      "loss": 0.8453,
+      "step": 13459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3296021849077856e-07,
+      "loss": 0.8911,
+      "step": 13460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.325970000443888e-07,
+      "loss": 0.9491,
+      "step": 13461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3223427508260888e-07,
+      "loss": 0.9643,
+      "step": 13462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3187204362357875e-07,
+      "loss": 0.8556,
+      "step": 13463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.315103056854128e-07,
+      "loss": 0.923,
+      "step": 13464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.311490612862043e-07,
+      "loss": 0.8066,
+      "step": 13465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3078831044401664e-07,
+      "loss": 0.8805,
+      "step": 13466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3042805317689533e-07,
+      "loss": 0.9002,
+      "step": 13467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3006828950285489e-07,
+      "loss": 0.8962,
+      "step": 13468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2970901943988867e-07,
+      "loss": 0.9088,
+      "step": 13469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2935024300596344e-07,
+      "loss": 0.8625,
+      "step": 13470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.289919602190226e-07,
+      "loss": 0.8437,
+      "step": 13471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2863417109698516e-07,
+      "loss": 0.7944,
+      "step": 13472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2827687565774462e-07,
+      "loss": 0.7805,
+      "step": 13473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2792007391916884e-07,
+      "loss": 0.9672,
+      "step": 13474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2756376589910246e-07,
+      "loss": 0.8132,
+      "step": 13475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2720795161536458e-07,
+      "loss": 0.7897,
+      "step": 13476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2685263108575096e-07,
+      "loss": 0.8991,
+      "step": 13477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2649780432803294e-07,
+      "loss": 0.7841,
+      "step": 13478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2614347135995299e-07,
+      "loss": 0.8199,
+      "step": 13479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2578963219923356e-07,
+      "loss": 0.9592,
+      "step": 13480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2543628686357057e-07,
+      "loss": 0.5793,
+      "step": 13481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2508343537063538e-07,
+      "loss": 0.8318,
+      "step": 13482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2473107773807502e-07,
+      "loss": 0.7982,
+      "step": 13483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2437921398350982e-07,
+      "loss": 0.8651,
+      "step": 13484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2402784412453906e-07,
+      "loss": 0.9388,
+      "step": 13485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2367696817873421e-07,
+      "loss": 0.9038,
+      "step": 13486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2332658616364347e-07,
+      "loss": 0.8735,
+      "step": 13487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2297669809679058e-07,
+      "loss": 0.8362,
+      "step": 13488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.226273039956727e-07,
+      "loss": 0.9155,
+      "step": 13489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.222784038777647e-07,
+      "loss": 0.8038,
+      "step": 13490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2192999776051483e-07,
+      "loss": 1.0011,
+      "step": 13491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.215820856613481e-07,
+      "loss": 0.7614,
+      "step": 13492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.212346675976639e-07,
+      "loss": 0.8048,
+      "step": 13493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.208877435868372e-07,
+      "loss": 0.8279,
+      "step": 13494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2054131364621858e-07,
+      "loss": 0.8867,
+      "step": 13495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2019537779313194e-07,
+      "loss": 0.5852,
+      "step": 13496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1984993604488016e-07,
+      "loss": 0.9133,
+      "step": 13497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1950498841873825e-07,
+      "loss": 0.926,
+      "step": 13498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.19160534931958e-07,
+      "loss": 0.6127,
+      "step": 13499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1881657560176563e-07,
+      "loss": 0.5802,
+      "step": 13500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1847311044536292e-07,
+      "loss": 0.9009,
+      "step": 13501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1813013947992835e-07,
+      "loss": 0.8671,
+      "step": 13502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1778766272261377e-07,
+      "loss": 0.863,
+      "step": 13503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1744568019054658e-07,
+      "loss": 0.8433,
+      "step": 13504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1710419190082978e-07,
+      "loss": 0.8777,
+      "step": 13505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1676319787054191e-07,
+      "loss": 0.8809,
+      "step": 13506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1642269811673713e-07,
+      "loss": 0.8692,
+      "step": 13507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1608269265644512e-07,
+      "loss": 0.9536,
+      "step": 13508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1574318150666786e-07,
+      "loss": 0.8015,
+      "step": 13509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.154041646843862e-07,
+      "loss": 0.9254,
+      "step": 13510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1506564220655436e-07,
+      "loss": 0.9583,
+      "step": 13511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1472761409010324e-07,
+      "loss": 0.8088,
+      "step": 13512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1439008035193821e-07,
+      "loss": 0.8653,
+      "step": 13513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1405304100893911e-07,
+      "loss": 0.9237,
+      "step": 13514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1371649607796132e-07,
+      "loss": 0.9032,
+      "step": 13515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1338044557583694e-07,
+      "loss": 0.8872,
+      "step": 13516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1304488951937253e-07,
+      "loss": 0.8716,
+      "step": 13517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1270982792534913e-07,
+      "loss": 0.883,
+      "step": 13518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1237526081052553e-07,
+      "loss": 0.9227,
+      "step": 13519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1204118819163057e-07,
+      "loss": 0.8468,
+      "step": 13520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1170761008537424e-07,
+      "loss": 0.9214,
+      "step": 13521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1137452650843872e-07,
+      "loss": 0.8788,
+      "step": 13522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1104193747748293e-07,
+      "loss": 0.8753,
+      "step": 13523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1070984300913912e-07,
+      "loss": 0.8792,
+      "step": 13524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1037824312001511e-07,
+      "loss": 0.9204,
+      "step": 13525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1004713782669541e-07,
+      "loss": 0.9577,
+      "step": 13526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0971652714574121e-07,
+      "loss": 0.8765,
+      "step": 13527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0938641109368376e-07,
+      "loss": 0.8458,
+      "step": 13528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0905678968703426e-07,
+      "loss": 0.844,
+      "step": 13529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0872766294227733e-07,
+      "loss": 0.5866,
+      "step": 13530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0839903087587312e-07,
+      "loss": 0.8285,
+      "step": 13531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0807089350425737e-07,
+      "loss": 0.8988,
+      "step": 13532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.077432508438403e-07,
+      "loss": 0.8728,
+      "step": 13533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0741610291100768e-07,
+      "loss": 0.9266,
+      "step": 13534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0708944972212088e-07,
+      "loss": 0.8763,
+      "step": 13535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0676329129351682e-07,
+      "loss": 0.8704,
+      "step": 13536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.06437627641508e-07,
+      "loss": 0.8944,
+      "step": 13537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0611245878237919e-07,
+      "loss": 0.8376,
+      "step": 13538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0578778473239293e-07,
+      "loss": 0.7235,
+      "step": 13539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0546360550778845e-07,
+      "loss": 0.621,
+      "step": 13540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0513992112477722e-07,
+      "loss": 0.9598,
+      "step": 13541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0481673159954741e-07,
+      "loss": 0.9115,
+      "step": 13542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0449403694826276e-07,
+      "loss": 0.8633,
+      "step": 13543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0417183718706036e-07,
+      "loss": 0.828,
+      "step": 13544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.038501323320551e-07,
+      "loss": 0.8263,
+      "step": 13545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0352892239933631e-07,
+      "loss": 0.8662,
+      "step": 13546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0320820740496673e-07,
+      "loss": 0.9577,
+      "step": 13547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0288798736498683e-07,
+      "loss": 0.8509,
+      "step": 13548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0256826229541161e-07,
+      "loss": 0.7681,
+      "step": 13549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0224903221222937e-07,
+      "loss": 0.8698,
+      "step": 13550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0193029713140845e-07,
+      "loss": 0.8645,
+      "step": 13551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0161205706888611e-07,
+      "loss": 0.8996,
+      "step": 13552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0129431204058071e-07,
+      "loss": 0.8627,
+      "step": 13553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0097706206238067e-07,
+      "loss": 0.7998,
+      "step": 13554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.006603071501533e-07,
+      "loss": 0.8643,
+      "step": 13555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0034404731974034e-07,
+      "loss": 0.9144,
+      "step": 13556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0002828258695808e-07,
+      "loss": 0.956,
+      "step": 13557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.971301296759938e-08,
+      "loss": 0.8376,
+      "step": 13558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.939823847742946e-08,
+      "loss": 0.8739,
+      "step": 13559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.908395913219127e-08,
+      "loss": 0.8478,
+      "step": 13560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.877017494760444e-08,
+      "loss": 0.9366,
+      "step": 13561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.845688593935976e-08,
+      "loss": 0.9541,
+      "step": 13562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.814409212312471e-08,
+      "loss": 0.857,
+      "step": 13563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.783179351454564e-08,
+      "loss": 0.9642,
+      "step": 13564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.751999012923675e-08,
+      "loss": 0.8088,
+      "step": 13565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.720868198279553e-08,
+      "loss": 0.9518,
+      "step": 13566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.689786909079068e-08,
+      "loss": 0.9626,
+      "step": 13567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.65875514687642e-08,
+      "loss": 0.9211,
+      "step": 13568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.627772913223699e-08,
+      "loss": 0.9181,
+      "step": 13569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.596840209670332e-08,
+      "loss": 0.6296,
+      "step": 13570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.565957037763418e-08,
+      "loss": 0.7796,
+      "step": 13571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.535123399047386e-08,
+      "loss": 0.7824,
+      "step": 13572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.504339295064336e-08,
+      "loss": 0.8506,
+      "step": 13573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.473604727353703e-08,
+      "loss": 0.8438,
+      "step": 13574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.442919697452701e-08,
+      "loss": 0.9093,
+      "step": 13575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.412284206895883e-08,
+      "loss": 0.8447,
+      "step": 13576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.381698257215355e-08,
+      "loss": 0.5935,
+      "step": 13577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.351161849940782e-08,
+      "loss": 0.7872,
+      "step": 13578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.320674986599387e-08,
+      "loss": 0.8234,
+      "step": 13579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.290237668715729e-08,
+      "loss": 0.8355,
+      "step": 13580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.259849897812256e-08,
+      "loss": 0.5677,
+      "step": 13581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.229511675408642e-08,
+      "loss": 0.8502,
+      "step": 13582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.199223003022007e-08,
+      "loss": 0.8659,
+      "step": 13583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.168983882167248e-08,
+      "loss": 0.886,
+      "step": 13584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.138794314356603e-08,
+      "loss": 0.6395,
+      "step": 13585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.108654301099973e-08,
+      "loss": 0.8895,
+      "step": 13586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.078563843904709e-08,
+      "loss": 0.8921,
+      "step": 13587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.048522944275717e-08,
+      "loss": 0.8886,
+      "step": 13588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.018531603715242e-08,
+      "loss": 0.9653,
+      "step": 13589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.988589823723193e-08,
+      "loss": 0.8879,
+      "step": 13590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.958697605797261e-08,
+      "loss": 0.7953,
+      "step": 13591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.928854951432141e-08,
+      "loss": 0.832,
+      "step": 13592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.899061862120528e-08,
+      "loss": 0.9169,
+      "step": 13593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.869318339352117e-08,
+      "loss": 0.9585,
+      "step": 13594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.83962438461472e-08,
+      "loss": 0.8655,
+      "step": 13595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80997999939337e-08,
+      "loss": 0.8397,
+      "step": 13596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.780385185170437e-08,
+      "loss": 0.8651,
+      "step": 13597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.750839943426181e-08,
+      "loss": 1.0067,
+      "step": 13598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.721344275637978e-08,
+      "loss": 0.8274,
+      "step": 13599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.6918981832812e-08,
+      "loss": 0.8396,
+      "step": 13600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.662501667828449e-08,
+      "loss": 0.8312,
+      "step": 13601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.633154730749772e-08,
+      "loss": 0.9474,
+      "step": 13602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.603857373512991e-08,
+      "loss": 0.872,
+      "step": 13603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.57460959758316e-08,
+      "loss": 0.8647,
+      "step": 13604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.545411404423109e-08,
+      "loss": 0.8226,
+      "step": 13605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.516262795493113e-08,
+      "loss": 0.7953,
+      "step": 13606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.487163772250895e-08,
+      "loss": 0.962,
+      "step": 13607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.458114336151735e-08,
+      "loss": 0.8661,
+      "step": 13608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.429114488648471e-08,
+      "loss": 0.9342,
+      "step": 13609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.400164231191387e-08,
+      "loss": 0.8983,
+      "step": 13610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.371263565228327e-08,
+      "loss": 0.8254,
+      "step": 13611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.342412492204799e-08,
+      "loss": 0.8926,
+      "step": 13612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.313611013563539e-08,
+      "loss": 0.8653,
+      "step": 13613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.28485913074506e-08,
+      "loss": 0.9115,
+      "step": 13614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.256156845187102e-08,
+      "loss": 0.9017,
+      "step": 13615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.227504158325405e-08,
+      "loss": 0.9046,
+      "step": 13616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.198901071592713e-08,
+      "loss": 0.9438,
+      "step": 13617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.170347586419547e-08,
+      "loss": 0.8932,
+      "step": 13618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.141843704233876e-08,
+      "loss": 0.8765,
+      "step": 13619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.113389426461337e-08,
+      "loss": 0.9176,
+      "step": 13620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.084984754525016e-08,
+      "loss": 0.8684,
+      "step": 13621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.05662968984533e-08,
+      "loss": 0.9537,
+      "step": 13622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.028324233840367e-08,
+      "loss": 0.7445,
+      "step": 13623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.000068387925775e-08,
+      "loss": 0.9315,
+      "step": 13624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.971862153514642e-08,
+      "loss": 0.7671,
+      "step": 13625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.943705532017732e-08,
+      "loss": 0.7458,
+      "step": 13626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.915598524843137e-08,
+      "loss": 0.7735,
+      "step": 13627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.887541133396404e-08,
+      "loss": 0.8398,
+      "step": 13628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.859533359080851e-08,
+      "loss": 0.867,
+      "step": 13629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.831575203297248e-08,
+      "loss": 0.7853,
+      "step": 13630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.803666667443699e-08,
+      "loss": 0.7985,
+      "step": 13631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.775807752916087e-08,
+      "loss": 0.8621,
+      "step": 13632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.74799846110752e-08,
+      "loss": 0.8321,
+      "step": 13633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.720238793408774e-08,
+      "loss": 0.8672,
+      "step": 13634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.692528751208406e-08,
+      "loss": 0.8749,
+      "step": 13635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.664868335891973e-08,
+      "loss": 0.8649,
+      "step": 13636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.637257548843035e-08,
+      "loss": 0.9653,
+      "step": 13637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.609696391442267e-08,
+      "loss": 0.8154,
+      "step": 13638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.58218486506812e-08,
+      "loss": 0.6737,
+      "step": 13639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.554722971096496e-08,
+      "loss": 0.9092,
+      "step": 13640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.527310710900738e-08,
+      "loss": 0.9533,
+      "step": 13641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.49994808585186e-08,
+      "loss": 0.7877,
+      "step": 13642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.472635097318326e-08,
+      "loss": 0.6061,
+      "step": 13643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.445371746666042e-08,
+      "loss": 0.816,
+      "step": 13644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.418158035258582e-08,
+      "loss": 0.8714,
+      "step": 13645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.39099396445686e-08,
+      "loss": 0.873,
+      "step": 13646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.363879535619456e-08,
+      "loss": 0.932,
+      "step": 13647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.336814750102284e-08,
+      "loss": 0.8046,
+      "step": 13648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.309799609258927e-08,
+      "loss": 0.7959,
+      "step": 13649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.28283411444064e-08,
+      "loss": 0.9729,
+      "step": 13650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.255918266995898e-08,
+      "loss": 0.9065,
+      "step": 13651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.229052068270737e-08,
+      "loss": 0.9571,
+      "step": 13652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.202235519608746e-08,
+      "loss": 0.8484,
+      "step": 13653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.17546862235119e-08,
+      "loss": 0.7829,
+      "step": 13654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.148751377836771e-08,
+      "loss": 0.8907,
+      "step": 13655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.122083787401423e-08,
+      "loss": 0.5556,
+      "step": 13656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.09546585237908e-08,
+      "loss": 0.8827,
+      "step": 13657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.068897574100897e-08,
+      "loss": 0.8175,
+      "step": 13658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.042378953895368e-08,
+      "loss": 0.8926,
+      "step": 13659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.015909993089098e-08,
+      "loss": 0.7825,
+      "step": 13660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.989490693005586e-08,
+      "loss": 0.8773,
+      "step": 13661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.963121054966104e-08,
+      "loss": 0.8594,
+      "step": 13662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.936801080289602e-08,
+      "loss": 0.7976,
+      "step": 13663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.910530770292245e-08,
+      "loss": 0.62,
+      "step": 13664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.884310126287985e-08,
+      "loss": 0.9506,
+      "step": 13665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.858139149587994e-08,
+      "loss": 0.8588,
+      "step": 13666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.832017841501227e-08,
+      "loss": 0.9171,
+      "step": 13667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.805946203333969e-08,
+      "loss": 0.8717,
+      "step": 13668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.779924236390178e-08,
+      "loss": 0.9235,
+      "step": 13669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.753951941971259e-08,
+      "loss": 0.9204,
+      "step": 13670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.728029321376173e-08,
+      "loss": 0.9411,
+      "step": 13671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.702156375901215e-08,
+      "loss": 0.8553,
+      "step": 13672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.67633310684046e-08,
+      "loss": 0.9181,
+      "step": 13673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.650559515485211e-08,
+      "loss": 0.8962,
+      "step": 13674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.624835603124658e-08,
+      "loss": 0.8223,
+      "step": 13675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.599161371045215e-08,
+      "loss": 0.8965,
+      "step": 13676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.573536820530746e-08,
+      "loss": 0.8866,
+      "step": 13677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.547961952863003e-08,
+      "loss": 0.9434,
+      "step": 13678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.52243676932085e-08,
+      "loss": 0.87,
+      "step": 13679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.496961271180935e-08,
+      "loss": 0.7226,
+      "step": 13680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.471535459717237e-08,
+      "loss": 0.7941,
+      "step": 13681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.446159336201519e-08,
+      "loss": 0.8697,
+      "step": 13682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.420832901902763e-08,
+      "loss": 0.7554,
+      "step": 13683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.395556158087512e-08,
+      "loss": 0.5851,
+      "step": 13684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.370329106019979e-08,
+      "loss": 0.8846,
+      "step": 13685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.345151746961931e-08,
+      "loss": 0.8386,
+      "step": 13686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.32002408217236e-08,
+      "loss": 0.9378,
+      "step": 13687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.294946112907818e-08,
+      "loss": 0.7772,
+      "step": 13688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.269917840422856e-08,
+      "loss": 0.7467,
+      "step": 13689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.244939265968919e-08,
+      "loss": 0.889,
+      "step": 13690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.220010390795228e-08,
+      "loss": 0.9996,
+      "step": 13691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.195131216148676e-08,
+      "loss": 0.8223,
+      "step": 13692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.170301743273377e-08,
+      "loss": 0.9825,
+      "step": 13693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.145521973411118e-08,
+      "loss": 0.9222,
+      "step": 13694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.120791907801238e-08,
+      "loss": 0.7296,
+      "step": 13695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.096111547680417e-08,
+      "loss": 0.9126,
+      "step": 13696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.071480894283e-08,
+      "loss": 0.8414,
+      "step": 13697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.04689994884078e-08,
+      "loss": 0.5947,
+      "step": 13698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.022368712583216e-08,
+      "loss": 0.621,
+      "step": 13699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.997887186737106e-08,
+      "loss": 0.854,
+      "step": 13700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.973455372526694e-08,
+      "loss": 0.8785,
+      "step": 13701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9490732711739995e-08,
+      "loss": 0.8669,
+      "step": 13702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9247408838982725e-08,
+      "loss": 0.8686,
+      "step": 13703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9004582119164265e-08,
+      "loss": 0.8873,
+      "step": 13704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8762252564430464e-08,
+      "loss": 0.9589,
+      "step": 13705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.852042018689941e-08,
+      "loss": 0.7273,
+      "step": 13706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8279084998665856e-08,
+      "loss": 0.8312,
+      "step": 13707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.803824701179794e-08,
+      "loss": 0.8878,
+      "step": 13708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.779790623834269e-08,
+      "loss": 0.9459,
+      "step": 13709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.755806269031827e-08,
+      "loss": 0.9893,
+      "step": 13710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.731871637972064e-08,
+      "loss": 0.864,
+      "step": 13711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.707986731851911e-08,
+      "loss": 0.8633,
+      "step": 13712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.684151551865858e-08,
+      "loss": 0.8855,
+      "step": 13713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6603660992061716e-08,
+      "loss": 0.7563,
+      "step": 13714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6366303750621246e-08,
+      "loss": 0.8926,
+      "step": 13715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.612944380620877e-08,
+      "loss": 0.8407,
+      "step": 13716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.589308117067038e-08,
+      "loss": 0.7896,
+      "step": 13717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.565721585582662e-08,
+      "loss": 0.7734,
+      "step": 13718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.54218478734736e-08,
+      "loss": 0.9232,
+      "step": 13719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.518697723538191e-08,
+      "loss": 0.9066,
+      "step": 13720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.495260395329882e-08,
+      "loss": 0.8385,
+      "step": 13721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4718728038944955e-08,
+      "loss": 0.9115,
+      "step": 13722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.448534950401763e-08,
+      "loss": 0.9018,
+      "step": 13723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.425246836018749e-08,
+      "loss": 0.587,
+      "step": 13724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.402008461910191e-08,
+      "loss": 0.7999,
+      "step": 13725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3788198292381574e-08,
+      "loss": 0.9747,
+      "step": 13726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.355680939162611e-08,
+      "loss": 0.8069,
+      "step": 13727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.332591792840513e-08,
+      "loss": 0.8235,
+      "step": 13728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3095523914266086e-08,
+      "loss": 1.0141,
+      "step": 13729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.286562736073308e-08,
+      "loss": 0.7939,
+      "step": 13730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.263622827930248e-08,
+      "loss": 0.8882,
+      "step": 13731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.240732668144732e-08,
+      "loss": 0.809,
+      "step": 13732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.217892257861401e-08,
+      "loss": 0.8345,
+      "step": 13733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.195101598222785e-08,
+      "loss": 0.9521,
+      "step": 13734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1723606903684166e-08,
+      "loss": 0.8002,
+      "step": 13735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.149669535435719e-08,
+      "loss": 0.8938,
+      "step": 13736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.127028134559564e-08,
+      "loss": 0.7396,
+      "step": 13737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1044364888722665e-08,
+      "loss": 0.8693,
+      "step": 13738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.081894599503589e-08,
+      "loss": 0.5983,
+      "step": 13739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.059402467580965e-08,
+      "loss": 0.881,
+      "step": 13740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0369600942292706e-08,
+      "loss": 0.8745,
+      "step": 13741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0145674805708314e-08,
+      "loss": 0.9603,
+      "step": 13742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.992224627725528e-08,
+      "loss": 0.853,
+      "step": 13743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.969931536810912e-08,
+      "loss": 0.9179,
+      "step": 13744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.947688208941648e-08,
+      "loss": 0.8592,
+      "step": 13745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9254946452303996e-08,
+      "loss": 0.8811,
+      "step": 13746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.903350846786947e-08,
+      "loss": 0.8589,
+      "step": 13747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8812568147187376e-08,
+      "loss": 0.9639,
+      "step": 13748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.859212550130776e-08,
+      "loss": 0.916,
+      "step": 13749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.837218054125625e-08,
+      "loss": 0.8757,
+      "step": 13750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.815273327803183e-08,
+      "loss": 0.8491,
+      "step": 13751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.793378372260793e-08,
+      "loss": 0.851,
+      "step": 13752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.771533188593691e-08,
+      "loss": 0.851,
+      "step": 13753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7497377778942254e-08,
+      "loss": 0.8676,
+      "step": 13754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7279921412525244e-08,
+      "loss": 0.9041,
+      "step": 13755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7062962797561615e-08,
+      "loss": 0.8379,
+      "step": 13756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.684650194489937e-08,
+      "loss": 0.7868,
+      "step": 13757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6630538865366506e-08,
+      "loss": 0.92,
+      "step": 13758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.641507356976327e-08,
+      "loss": 0.8529,
+      "step": 13759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.620010606886549e-08,
+      "loss": 0.595,
+      "step": 13760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.598563637342235e-08,
+      "loss": 0.8499,
+      "step": 13761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5771664494161926e-08,
+      "loss": 0.8273,
+      "step": 13762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5558190441783445e-08,
+      "loss": 0.7422,
+      "step": 13763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.534521422696503e-08,
+      "loss": 0.8141,
+      "step": 13764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.513273586035705e-08,
+      "loss": 0.8797,
+      "step": 13765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4920755352585445e-08,
+      "loss": 0.92,
+      "step": 13766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.470927271425174e-08,
+      "loss": 0.8296,
+      "step": 13767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.449828795593303e-08,
+      "loss": 0.8323,
+      "step": 13768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.428780108818198e-08,
+      "loss": 0.7789,
+      "step": 13769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.407781212152351e-08,
+      "loss": 0.8614,
+      "step": 13770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.386832106646033e-08,
+      "loss": 0.903,
+      "step": 13771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.365932793346961e-08,
+      "loss": 0.8553,
+      "step": 13772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.345083273300188e-08,
+      "loss": 0.907,
+      "step": 13773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.324283547548658e-08,
+      "loss": 0.8977,
+      "step": 13774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.303533617132538e-08,
+      "loss": 0.9996,
+      "step": 13775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2828334830894434e-08,
+      "loss": 0.913,
+      "step": 13776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.262183146454768e-08,
+      "loss": 0.6206,
+      "step": 13777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.24158260826113e-08,
+      "loss": 0.7979,
+      "step": 13778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2210318695389275e-08,
+      "loss": 0.7721,
+      "step": 13779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.200530931315894e-08,
+      "loss": 0.7371,
+      "step": 13780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.18007979461732e-08,
+      "loss": 0.8534,
+      "step": 13781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.159678460465832e-08,
+      "loss": 0.8177,
+      "step": 13782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.139326929882059e-08,
+      "loss": 0.8342,
+      "step": 13783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.11902520388352e-08,
+      "loss": 0.8537,
+      "step": 13784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0987732834857355e-08,
+      "loss": 0.8691,
+      "step": 13785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.078571169701451e-08,
+      "loss": 0.7686,
+      "step": 13786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.058418863540858e-08,
+      "loss": 0.8354,
+      "step": 13787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0383163660121516e-08,
+      "loss": 0.7843,
+      "step": 13788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0182636781204154e-08,
+      "loss": 0.6397,
+      "step": 13789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.998260800868625e-08,
+      "loss": 0.8785,
+      "step": 13790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.978307735257092e-08,
+      "loss": 0.8562,
+      "step": 13791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.958404482283684e-08,
+      "loss": 0.9508,
+      "step": 13792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.938551042943828e-08,
+      "loss": 0.8692,
+      "step": 13793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9187474182305066e-08,
+      "loss": 0.8018,
+      "step": 13794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.89899360913415e-08,
+      "loss": 0.7876,
+      "step": 13795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8792896166425234e-08,
+      "loss": 0.8807,
+      "step": 13796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.859635441741061e-08,
+      "loss": 0.8957,
+      "step": 13797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8400310854128654e-08,
+      "loss": 0.8725,
+      "step": 13798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8204765486381523e-08,
+      "loss": 0.8589,
+      "step": 13799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8009718323951394e-08,
+      "loss": 0.986,
+      "step": 13800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.781516937659158e-08,
+      "loss": 0.7728,
+      "step": 13801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.762111865403095e-08,
+      "loss": 0.8435,
+      "step": 13802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.742756616597509e-08,
+      "loss": 0.9136,
+      "step": 13803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.723451192210403e-08,
+      "loss": 0.7679,
+      "step": 13804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.704195593207338e-08,
+      "loss": 0.982,
+      "step": 13805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.68498982055121e-08,
+      "loss": 0.7706,
+      "step": 13806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6658338752024734e-08,
+      "loss": 0.9251,
+      "step": 13807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.646727758119251e-08,
+      "loss": 0.9683,
+      "step": 13808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6276714702571104e-08,
+      "loss": 0.9474,
+      "step": 13809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.608665012568957e-08,
+      "loss": 0.9636,
+      "step": 13810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5897083860054746e-08,
+      "loss": 0.8479,
+      "step": 13811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5708015915145725e-08,
+      "loss": 0.8296,
+      "step": 13812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.551944630041937e-08,
+      "loss": 0.8931,
+      "step": 13813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.533137502530593e-08,
+      "loss": 0.8285,
+      "step": 13814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.51438020992112e-08,
+      "loss": 0.8913,
+      "step": 13815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4956727531515466e-08,
+      "loss": 0.791,
+      "step": 13816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.477015133157347e-08,
+      "loss": 0.833,
+      "step": 13817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.458407350871995e-08,
+      "loss": 0.9963,
+      "step": 13818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.439849407225748e-08,
+      "loss": 0.9321,
+      "step": 13819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4213413031468637e-08,
+      "loss": 0.9358,
+      "step": 13820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.402883039560934e-08,
+      "loss": 0.8998,
+      "step": 13821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384474617391109e-08,
+      "loss": 0.7609,
+      "step": 13822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.366116037557987e-08,
+      "loss": 0.7731,
+      "step": 13823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.347807300979722e-08,
+      "loss": 0.8462,
+      "step": 13824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3295484085720254e-08,
+      "loss": 0.9193,
+      "step": 13825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.311339361247945e-08,
+      "loss": 0.7566,
+      "step": 13826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.293180159918197e-08,
+      "loss": 0.9037,
+      "step": 13827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.275070805490943e-08,
+      "loss": 0.8327,
+      "step": 13828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2570112988719036e-08,
+      "loss": 0.7505,
+      "step": 13829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2390016409641345e-08,
+      "loss": 0.8519,
+      "step": 13830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.221041832668581e-08,
+      "loss": 0.838,
+      "step": 13831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2031318748830806e-08,
+      "loss": 0.9017,
+      "step": 13832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.185271768503695e-08,
+      "loss": 0.9079,
+      "step": 13833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1674615144234865e-08,
+      "loss": 0.9169,
+      "step": 13834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1497011135330766e-08,
+      "loss": 0.9713,
+      "step": 13835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.131990566720866e-08,
+      "loss": 0.9176,
+      "step": 13836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1143298748724793e-08,
+      "loss": 0.8354,
+      "step": 13837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.09671903887121e-08,
+      "loss": 0.8324,
+      "step": 13838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0791580595977974e-08,
+      "loss": 0.7881,
+      "step": 13839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0616469379304294e-08,
+      "loss": 0.9046,
+      "step": 13840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0441856747448484e-08,
+      "loss": 0.8809,
+      "step": 13841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.026774270914468e-08,
+      "loss": 0.8115,
+      "step": 13842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.009412727309924e-08,
+      "loss": 0.9247,
+      "step": 13843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.992101044799633e-08,
+      "loss": 0.8617,
+      "step": 13844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9748392242492376e-08,
+      "loss": 0.9454,
+      "step": 13845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9576272665220452e-08,
+      "loss": 0.8147,
+      "step": 13846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9404651724789234e-08,
+      "loss": 0.9264,
+      "step": 13847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9233529429781858e-08,
+      "loss": 0.8106,
+      "step": 13848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9062905788755925e-08,
+      "loss": 0.8275,
+      "step": 13849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8892780810244604e-08,
+      "loss": 0.8599,
+      "step": 13850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8723154502755534e-08,
+      "loss": 0.8174,
+      "step": 13851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8554026874773045e-08,
+      "loss": 0.8204,
+      "step": 13852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.838539793475481e-08,
+      "loss": 0.9942,
+      "step": 13853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8217267691135196e-08,
+      "loss": 0.8996,
+      "step": 13854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8049636152321924e-08,
+      "loss": 0.9372,
+      "step": 13855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7882503326698284e-08,
+      "loss": 0.8765,
+      "step": 13856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7715869222623148e-08,
+      "loss": 0.8048,
+      "step": 13857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.754973384842985e-08,
+      "loss": 0.849,
+      "step": 13858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7384097212428405e-08,
+      "loss": 0.8867,
+      "step": 13859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7218959322901083e-08,
+      "loss": 0.8509,
+      "step": 13860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.705432018810683e-08,
+      "loss": 0.885,
+      "step": 13861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.689017981627906e-08,
+      "loss": 0.8409,
+      "step": 13862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6726538215628982e-08,
+      "loss": 0.7787,
+      "step": 13863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.656339539433783e-08,
+      "loss": 0.8321,
+      "step": 13864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6400751360566857e-08,
+      "loss": 0.8129,
+      "step": 13865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.623860612244955e-08,
+      "loss": 0.8585,
+      "step": 13866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.607695968809387e-08,
+      "loss": 0.8651,
+      "step": 13867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5915812065584466e-08,
+      "loss": 0.8723,
+      "step": 13868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5755163262981552e-08,
+      "loss": 0.8433,
+      "step": 13869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5595013288318705e-08,
+      "loss": 0.6211,
+      "step": 13870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.543536214960618e-08,
+      "loss": 0.9065,
+      "step": 13871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5276209854826482e-08,
+      "loss": 0.8624,
+      "step": 13872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5117556411941025e-08,
+      "loss": 0.6181,
+      "step": 13873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4959401828882346e-08,
+      "loss": 0.9461,
+      "step": 13874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4801746113561897e-08,
+      "loss": 0.7362,
+      "step": 13875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4644589273863372e-08,
+      "loss": 0.8956,
+      "step": 13876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.448793131764715e-08,
+      "loss": 0.8777,
+      "step": 13877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4331772252746964e-08,
+      "loss": 0.7978,
+      "step": 13878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4176112086973237e-08,
+      "loss": 0.6129,
+      "step": 13879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4020950828110844e-08,
+      "loss": 0.8309,
+      "step": 13880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3866288483918032e-08,
+      "loss": 0.9199,
+      "step": 13881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3712125062131942e-08,
+      "loss": 0.6076,
+      "step": 13882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3558460570461962e-08,
+      "loss": 0.8163,
+      "step": 13883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3405295016593056e-08,
+      "loss": 0.5684,
+      "step": 13884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.325262840818465e-08,
+      "loss": 0.8705,
+      "step": 13885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3100460752871755e-08,
+      "loss": 0.8953,
+      "step": 13886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2948792058264947e-08,
+      "loss": 0.9172,
+      "step": 13887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2797622331949266e-08,
+      "loss": 0.8834,
+      "step": 13888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2646951581486443e-08,
+      "loss": 0.7668,
+      "step": 13889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2496779814409343e-08,
+      "loss": 0.877,
+      "step": 13890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2347107038229733e-08,
+      "loss": 0.92,
+      "step": 13891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2197933260432737e-08,
+      "loss": 0.8098,
+      "step": 13892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2049258488477942e-08,
+      "loss": 0.8458,
+      "step": 13893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1901082729801627e-08,
+      "loss": 0.8669,
+      "step": 13894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1753405991815634e-08,
+      "loss": 0.8886,
+      "step": 13895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1606228281902953e-08,
+      "loss": 0.7776,
+      "step": 13896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1459549607425468e-08,
+      "loss": 0.8835,
+      "step": 13897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1313369975718423e-08,
+      "loss": 0.9499,
+      "step": 13898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.116768939409375e-08,
+      "loss": 0.5865,
+      "step": 13899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1022507869835618e-08,
+      "loss": 0.8202,
+      "step": 13900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0877825410204888e-08,
+      "loss": 0.884,
+      "step": 13901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0733642022437993e-08,
+      "loss": 0.8455,
+      "step": 13902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0589957713745835e-08,
+      "loss": 0.8364,
+      "step": 13903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.044677249131488e-08,
+      "loss": 0.8443,
+      "step": 13904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0304086362303853e-08,
+      "loss": 0.8815,
+      "step": 13905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0161899333850376e-08,
+      "loss": 0.802,
+      "step": 13906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0020211413065428e-08,
+      "loss": 0.8338,
+      "step": 13907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.987902260703445e-08,
+      "loss": 0.7793,
+      "step": 13908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.973833292281957e-08,
+      "loss": 0.8601,
+      "step": 13909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9598142367456273e-08,
+      "loss": 0.8817,
+      "step": 13910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.945845094795451e-08,
+      "loss": 0.9749,
+      "step": 13911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.931925867130202e-08,
+      "loss": 0.749,
+      "step": 13912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9180565544459906e-08,
+      "loss": 0.8379,
+      "step": 13913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9042371574364838e-08,
+      "loss": 0.8402,
+      "step": 13914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8904676767926845e-08,
+      "loss": 0.7887,
+      "step": 13915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.876748113203153e-08,
+      "loss": 0.9315,
+      "step": 13916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86307846735434e-08,
+      "loss": 0.8987,
+      "step": 13917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8494587399295883e-08,
+      "loss": 0.9004,
+      "step": 13918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8358889316102413e-08,
+      "loss": 0.8331,
+      "step": 13919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8223690430747566e-08,
+      "loss": 0.8817,
+      "step": 13920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8088990749994815e-08,
+      "loss": 0.9021,
+      "step": 13921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.795479028057878e-08,
+      "loss": 1.0406,
+      "step": 13922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7821089029212978e-08,
+      "loss": 0.6078,
+      "step": 13923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7687887002583172e-08,
+      "loss": 0.8437,
+      "step": 13924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7555184207350694e-08,
+      "loss": 0.9692,
+      "step": 13925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7422980650152467e-08,
+      "loss": 0.9949,
+      "step": 13926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.729127633759986e-08,
+      "loss": 0.9087,
+      "step": 13927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7160071276279832e-08,
+      "loss": 0.8199,
+      "step": 13928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.702936547275491e-08,
+      "loss": 0.858,
+      "step": 13929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6899158933560978e-08,
+      "loss": 0.9007,
+      "step": 13930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.67694516652106e-08,
+      "loss": 0.915,
+      "step": 13931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6640243674189704e-08,
+      "loss": 0.9142,
+      "step": 13932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6511534966960897e-08,
+      "loss": 0.8979,
+      "step": 13933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6383325549961247e-08,
+      "loss": 0.7497,
+      "step": 13934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62556154296023e-08,
+      "loss": 0.7052,
+      "step": 13935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6128404612270054e-08,
+      "loss": 0.8723,
+      "step": 13936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.600169310432942e-08,
+      "loss": 0.8795,
+      "step": 13937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5875480912114217e-08,
+      "loss": 0.9837,
+      "step": 13938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749768041938286e-08,
+      "loss": 0.8607,
+      "step": 13939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5624554500088817e-08,
+      "loss": 0.8626,
+      "step": 13940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5499840292826362e-08,
+      "loss": 1.0391,
+      "step": 13941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.537562542639037e-08,
+      "loss": 0.9367,
+      "step": 13942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5251909906991436e-08,
+      "loss": 0.8648,
+      "step": 13943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5128693740816823e-08,
+      "loss": 0.8467,
+      "step": 13944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5005976934029386e-08,
+      "loss": 0.7661,
+      "step": 13945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4883759492766437e-08,
+      "loss": 0.8108,
+      "step": 13946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.476204142313975e-08,
+      "loss": 0.8547,
+      "step": 13947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4640822731237791e-08,
+      "loss": 0.906,
+      "step": 13948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4520103423120158e-08,
+      "loss": 0.8699,
+      "step": 13949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4399883504827572e-08,
+      "loss": 0.8434,
+      "step": 13950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4280162982369671e-08,
+      "loss": 0.8132,
+      "step": 13951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.416094186173611e-08,
+      "loss": 0.5671,
+      "step": 13952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4042220148887674e-08,
+      "loss": 0.8834,
+      "step": 13953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.392399784976295e-08,
+      "loss": 0.9241,
+      "step": 13954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.380627497027276e-08,
+      "loss": 0.8622,
+      "step": 13955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.368905151630684e-08,
+      "loss": 0.9296,
+      "step": 13956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3572327493726056e-08,
+      "loss": 0.849,
+      "step": 13957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3456102908369074e-08,
+      "loss": 0.827,
+      "step": 13958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3340377766047907e-08,
+      "loss": 0.8844,
+      "step": 13959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3225152072550152e-08,
+      "loss": 0.8448,
+      "step": 13960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3110425833638974e-08,
+      "loss": 0.91,
+      "step": 13961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2996199055050895e-08,
+      "loss": 0.8253,
+      "step": 13962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2882471742499126e-08,
+      "loss": 0.8687,
+      "step": 13963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2769243901671335e-08,
+      "loss": 0.8266,
+      "step": 13964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2656515538230774e-08,
+      "loss": 0.8824,
+      "step": 13965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2544286657815152e-08,
+      "loss": 0.9795,
+      "step": 13966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2432557266035538e-08,
+      "loss": 0.6078,
+      "step": 13967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2321327368481906e-08,
+      "loss": 0.8503,
+      "step": 13968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2210596970716471e-08,
+      "loss": 0.9518,
+      "step": 13969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2100366078275916e-08,
+      "loss": 0.8098,
+      "step": 13970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1990634696673608e-08,
+      "loss": 0.6175,
+      "step": 13971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1881402831397382e-08,
+      "loss": 0.9779,
+      "step": 13972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1772670487910643e-08,
+      "loss": 0.8311,
+      "step": 13973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1664437671651263e-08,
+      "loss": 0.8726,
+      "step": 13974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.155670438803158e-08,
+      "loss": 0.8784,
+      "step": 13975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1449470642439509e-08,
+      "loss": 0.8538,
+      "step": 13976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1342736440238533e-08,
+      "loss": 0.7866,
+      "step": 13977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1236501786766607e-08,
+      "loss": 0.9005,
+      "step": 13978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1130766687336148e-08,
+      "loss": 0.883,
+      "step": 13979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1025531147235146e-08,
+      "loss": 0.8934,
+      "step": 13980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0920795171726062e-08,
+      "loss": 0.8745,
+      "step": 13981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0816558766049146e-08,
+      "loss": 0.8887,
+      "step": 13982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0712821935414675e-08,
+      "loss": 0.8593,
+      "step": 13983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0609584685012942e-08,
+      "loss": 0.7875,
+      "step": 13984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0506847020005373e-08,
+      "loss": 0.9429,
+      "step": 13985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0404608945530081e-08,
+      "loss": 0.8789,
+      "step": 13986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0302870466701864e-08,
+      "loss": 0.8389,
+      "step": 13987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0201631588606653e-08,
+      "loss": 0.8255,
+      "step": 13988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0100892316309285e-08,
+      "loss": 0.8011,
+      "step": 13989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0000652654847954e-08,
+      "loss": 0.7422,
+      "step": 13990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.900912609234203e-09,
+      "loss": 0.8063,
+      "step": 13991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801672184457378e-09,
+      "loss": 0.8944,
+      "step": 13992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.702931385480175e-09,
+      "loss": 0.9144,
+      "step": 13993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.604690217240864e-09,
+      "loss": 0.8329,
+      "step": 13994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.506948684652184e-09,
+      "loss": 0.8582,
+      "step": 13995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.409706792603556e-09,
+      "loss": 0.8422,
+      "step": 13996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.312964545957758e-09,
+      "loss": 0.8444,
+      "step": 13997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.216721949553143e-09,
+      "loss": 0.8149,
+      "step": 13998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.120979008201414e-09,
+      "loss": 0.7778,
+      "step": 13999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.025735726693186e-09,
+      "loss": 0.8848,
+      "step": 14000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.930992109790204e-09,
+      "loss": 0.7394,
+      "step": 14001
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.836748162230901e-09,
+      "loss": 0.8621,
+      "step": 14002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.743003888728174e-09,
+      "loss": 0.8752,
+      "step": 14003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.649759293972714e-09,
+      "loss": 0.9403,
+      "step": 14004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.55701438262413e-09,
+      "loss": 0.8642,
+      "step": 14005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.46476915932426e-09,
+      "loss": 0.8106,
+      "step": 14006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.373023628683863e-09,
+      "loss": 0.9994,
+      "step": 14007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.2817777952926e-09,
+      "loss": 0.8987,
+      "step": 14008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.19103166371349e-09,
+      "loss": 0.9224,
+      "step": 14009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.10078523848401e-09,
+      "loss": 0.8303,
+      "step": 14010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.01103852411944e-09,
+      "loss": 0.8833,
+      "step": 14011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.921791525107303e-09,
+      "loss": 0.9827,
+      "step": 14012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.833044245910692e-09,
+      "loss": 0.9376,
+      "step": 14013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.744796690967172e-09,
+      "loss": 0.8875,
+      "step": 14014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.657048864692096e-09,
+      "loss": 0.8601,
+      "step": 14015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.569800771471957e-09,
+      "loss": 0.6217,
+      "step": 14016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.483052415672154e-09,
+      "loss": 0.6319,
+      "step": 14017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.396803801629215e-09,
+      "loss": 0.8439,
+      "step": 14018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.311054933657469e-09,
+      "loss": 0.5866,
+      "step": 14019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.2258058160457056e-09,
+      "loss": 0.6059,
+      "step": 14020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.141056453056072e-09,
+      "loss": 0.7868,
+      "step": 14021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.05680684892851e-09,
+      "loss": 0.8821,
+      "step": 14022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.973057007876316e-09,
+      "loss": 0.7762,
+      "step": 14023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.889806934087251e-09,
+      "loss": 0.6053,
+      "step": 14024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.807056631724651e-09,
+      "loss": 0.8243,
+      "step": 14025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.724806104927428e-09,
+      "loss": 0.8933,
+      "step": 14026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.643055357808959e-09,
+      "loss": 0.9298,
+      "step": 14027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5618043944581935e-09,
+      "loss": 0.88,
+      "step": 14028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.481053218937439e-09,
+      "loss": 0.8212,
+      "step": 14029
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.400801835286796e-09,
+      "loss": 0.7567,
+      "step": 14030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.321050247518612e-09,
+      "loss": 0.955,
+      "step": 14031
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.241798459620807e-09,
+      "loss": 0.97,
+      "step": 14032
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.163046475559098e-09,
+      "loss": 0.7857,
+      "step": 14033
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.084794299270336e-09,
+      "loss": 0.8292,
+      "step": 14034
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0070419346680566e-09,
+      "loss": 0.9117,
+      "step": 14035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.929789385641371e-09,
+      "loss": 0.8873,
+      "step": 14036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8530366560527465e-09,
+      "loss": 0.9422,
+      "step": 14037
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.776783749742443e-09,
+      "loss": 0.827,
+      "step": 14038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.701030670522967e-09,
+      "loss": 0.8415,
+      "step": 14039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6257774221823995e-09,
+      "loss": 0.9572,
+      "step": 14040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.551024008485506e-09,
+      "loss": 0.9777,
+      "step": 14041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.476770433169299e-09,
+      "loss": 0.9797,
+      "step": 14042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4030166999485825e-09,
+      "loss": 0.8105,
+      "step": 14043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.329762812511519e-09,
+      "loss": 0.8674,
+      "step": 14044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2570087745218435e-09,
+      "loss": 0.8219,
+      "step": 14045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.184754589617758e-09,
+      "loss": 0.8828,
+      "step": 14046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.113000261411927e-09,
+      "loss": 0.8171,
+      "step": 14047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.041745793494812e-09,
+      "loss": 0.8772,
+      "step": 14048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.97099118942912e-09,
+      "loss": 0.8934,
+      "step": 14049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.90073645275313e-09,
+      "loss": 0.9144,
+      "step": 14050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8309815869806986e-09,
+      "loss": 0.8678,
+      "step": 14051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7617265956001465e-09,
+      "loss": 0.8707,
+      "step": 14052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.69297148207426e-09,
+      "loss": 0.9011,
+      "step": 14053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.62471624984362e-09,
+      "loss": 0.8537,
+      "step": 14054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.55696090231994e-09,
+      "loss": 0.8161,
+      "step": 14055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.489705442892733e-09,
+      "loss": 0.854,
+      "step": 14056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.422949874925975e-09,
+      "loss": 0.7915,
+      "step": 14057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.356694201756994e-09,
+      "loss": 0.8327,
+      "step": 14058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.290938426698699e-09,
+      "loss": 0.7923,
+      "step": 14059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.225682553041788e-09,
+      "loss": 0.8465,
+      "step": 14060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.160926584049207e-09,
+      "loss": 0.5737,
+      "step": 14061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0966705229594784e-09,
+      "loss": 0.9147,
+      "step": 14062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.032914372984475e-09,
+      "loss": 0.7942,
+      "step": 14063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.96965813731498e-09,
+      "loss": 0.8613,
+      "step": 14064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.906901819112907e-09,
+      "loss": 0.8232,
+      "step": 14065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.844645421517967e-09,
+      "loss": 0.9679,
+      "step": 14066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.782888947644336e-09,
+      "loss": 0.6012,
+      "step": 14067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7216324005784342e-09,
+      "loss": 0.6095,
+      "step": 14068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6608757833844766e-09,
+      "loss": 0.8376,
+      "step": 14069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6006190991011434e-09,
+      "loss": 1.0741,
+      "step": 14070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5408623507438013e-09,
+      "loss": 0.9106,
+      "step": 14071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.481605541297839e-09,
+      "loss": 0.8485,
+      "step": 14072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4228486737286627e-09,
+      "loss": 0.9126,
+      "step": 14073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3645917509750325e-09,
+      "loss": 0.8517,
+      "step": 14074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3068347759490637e-09,
+      "loss": 0.8867,
+      "step": 14075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.249577751540667e-09,
+      "loss": 0.8818,
+      "step": 14076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.192820680613107e-09,
+      "loss": 0.8504,
+      "step": 14077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.136563566004114e-09,
+      "loss": 0.9319,
+      "step": 14078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0808064105269932e-09,
+      "loss": 0.7606,
+      "step": 14079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0255492169717348e-09,
+      "loss": 0.8326,
+      "step": 14080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9707919881016844e-09,
+      "loss": 0.8787,
+      "step": 14081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9165347266535414e-09,
+      "loss": 0.9087,
+      "step": 14082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8627774353418015e-09,
+      "loss": 0.843,
+      "step": 14083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8095201168565343e-09,
+      "loss": 0.926,
+      "step": 14084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.756762773857835e-09,
+      "loss": 0.8504,
+      "step": 14085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.704505408988034e-09,
+      "loss": 0.8615,
+      "step": 14086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6527480248572656e-09,
+      "loss": 0.7543,
+      "step": 14087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6014906240556803e-09,
+      "loss": 0.8318,
+      "step": 14088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.550733209146783e-09,
+      "loss": 0.8289,
+      "step": 14089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5004757826685432e-09,
+      "loss": 0.8078,
+      "step": 14090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4507183471356165e-09,
+      "loss": 0.865,
+      "step": 14091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4014609050337923e-09,
+      "loss": 0.7746,
+      "step": 14092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3527034588288757e-09,
+      "loss": 0.8767,
+      "step": 14093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.304446010958916e-09,
+      "loss": 0.7921,
+      "step": 14094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2566885638364287e-09,
+      "loss": 0.8103,
+      "step": 14095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2094311198517237e-09,
+      "loss": 0.9503,
+      "step": 14096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.162673681365135e-09,
+      "loss": 0.8081,
+      "step": 14097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.116416250718123e-09,
+      "loss": 0.549,
+      "step": 14098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.070658830222172e-09,
+      "loss": 0.8992,
+      "step": 14099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.025401422167672e-09,
+      "loss": 0.8523,
+      "step": 14100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9806440288150376e-09,
+      "loss": 0.8182,
+      "step": 14101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9363866524046983e-09,
+      "loss": 0.9622,
+      "step": 14102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8926292951504387e-09,
+      "loss": 0.8007,
+      "step": 14103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8493719592393988e-09,
+      "loss": 0.8837,
+      "step": 14104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8066146468354029e-09,
+      "loss": 0.8941,
+      "step": 14105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7643573600767406e-09,
+      "loss": 0.8324,
+      "step": 14106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7226001010772765e-09,
+      "loss": 0.9112,
+      "step": 14107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6813428719242298e-09,
+      "loss": 0.8159,
+      "step": 14108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6405856746826154e-09,
+      "loss": 0.845,
+      "step": 14109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6003285113885825e-09,
+      "loss": 0.6917,
+      "step": 14110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605713840571857e-09,
+      "loss": 0.8371,
+      "step": 14111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5213142946768344e-09,
+      "loss": 0.799,
+      "step": 14112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.482557245209293e-09,
+      "loss": 0.7969,
+      "step": 14113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4443002375941207e-09,
+      "loss": 0.911,
+      "step": 14114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4065432737442319e-09,
+      "loss": 0.9918,
+      "step": 14115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3692863555481163e-09,
+      "loss": 0.8066,
+      "step": 14116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3325294848698378e-09,
+      "loss": 0.9004,
+      "step": 14117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2962726635457057e-09,
+      "loss": 0.8824,
+      "step": 14118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2605158933898242e-09,
+      "loss": 0.577,
+      "step": 14119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2252591761907628e-09,
+      "loss": 0.9253,
+      "step": 14120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1905025137115555e-09,
+      "loss": 0.7902,
+      "step": 14121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1562459076908118e-09,
+      "loss": 0.9061,
+      "step": 14122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1224893598416054e-09,
+      "loss": 0.8119,
+      "step": 14123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0892328718514756e-09,
+      "loss": 0.6489,
+      "step": 14124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0564764453846465e-09,
+      "loss": 0.7922,
+      "step": 14125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0242200820786974e-09,
+      "loss": 0.9186,
+      "step": 14126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.924637835467821e-10,
+      "loss": 0.9375,
+      "step": 14127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.612075513776297e-10,
+      "loss": 0.8402,
+      "step": 14128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.304513871333242e-10,
+      "loss": 0.7952,
+      "step": 14129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.001952923537449e-10,
+      "loss": 0.9235,
+      "step": 14130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.704392685510154e-10,
+      "loss": 0.9133,
+      "step": 14131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.411833172128348e-10,
+      "loss": 0.9718,
+      "step": 14132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.124274398024767e-10,
+      "loss": 0.8851,
+      "step": 14133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.841716377587905e-10,
+      "loss": 0.9371,
+      "step": 14134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.564159124950899e-10,
+      "loss": 0.8643,
+      "step": 14135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.291602653980434e-10,
+      "loss": 0.845,
+      "step": 14136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.024046978321153e-10,
+      "loss": 0.8977,
+      "step": 14137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.761492111351242e-10,
+      "loss": 0.8005,
+      "step": 14138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.503938066193538e-10,
+      "loss": 0.7995,
+      "step": 14139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.25138485573773e-10,
+      "loss": 0.7724,
+      "step": 14140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.003832492607053e-10,
+      "loss": 0.8414,
+      "step": 14141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.761280989191598e-10,
+      "loss": 0.9102,
+      "step": 14142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.523730357614998e-10,
+      "loss": 0.8423,
+      "step": 14143
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.291180609756641e-10,
+      "loss": 0.7984,
+      "step": 14144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.063631757251664e-10,
+      "loss": 0.8424,
+      "step": 14145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.841083811479852e-10,
+      "loss": 0.9086,
+      "step": 14146
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.623536783565641e-10,
+      "loss": 0.8478,
+      "step": 14147
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.410990684389216e-10,
+      "loss": 0.9106,
+      "step": 14148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2034455245865134e-10,
+      "loss": 0.9261,
+      "step": 14149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.000901314527017e-10,
+      "loss": 0.8826,
+      "step": 14150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8033580643581647e-10,
+      "loss": 0.8886,
+      "step": 14151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.610815783938737e-10,
+      "loss": 0.7617,
+      "step": 14152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.423274482916572e-10,
+      "loss": 0.8389,
+      "step": 14153
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.24073417065085e-10,
+      "loss": 0.8327,
+      "step": 14154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0631948562898086e-10,
+      "loss": 0.6063,
+      "step": 14155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8906565487041295e-10,
+      "loss": 0.7717,
+      "step": 14156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7231192565202456e-10,
+      "loss": 0.819,
+      "step": 14157
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5605829881203414e-10,
+      "loss": 0.746,
+      "step": 14158
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.403047751631249e-10,
+      "loss": 0.8654,
+      "step": 14159
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2505135549355518e-10,
+      "loss": 0.8175,
+      "step": 14160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.102980405660482e-10,
+      "loss": 0.8149,
+      "step": 14161
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9604483111890226e-10,
+      "loss": 0.5947,
+      "step": 14162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8229172786266014e-10,
+      "loss": 0.9092,
+      "step": 14163
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6903873148788052e-10,
+      "loss": 0.7982,
+      "step": 14164
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5628584265625635e-10,
+      "loss": 0.8677,
+      "step": 14165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4403306200616585e-10,
+      "loss": 0.8093,
+      "step": 14166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3228039014934191e-10,
+      "loss": 0.9384,
+      "step": 14167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2102782767309252e-10,
+      "loss": 0.8595,
+      "step": 14168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1027537514252118e-10,
+      "loss": 0.9308,
+      "step": 14169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0002303309386563e-10,
+      "loss": 0.7993,
+      "step": 14170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.027080203893867e-11,
+      "loss": 0.9198,
+      "step": 14171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.101868246734868e-11,
+      "loss": 0.8325,
+      "step": 14172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.22666748409484e-11,
+      "loss": 0.6206,
+      "step": 14173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.401477959716573e-11,
+      "loss": 0.8603,
+      "step": 14174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.626299715011385e-11,
+      "loss": 0.9351,
+      "step": 14175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.901132788504015e-11,
+      "loss": 0.8174,
+      "step": 14176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.225977216609778e-11,
+      "loss": 0.9211,
+      "step": 14177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.600833033190476e-11,
+      "loss": 0.8077,
+      "step": 14178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.025700269221332e-11,
+      "loss": 0.92,
+      "step": 14179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5005789537901893e-11,
+      "loss": 0.8696,
+      "step": 14180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0254691129872883e-11,
+      "loss": 0.8287,
+      "step": 14181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6003707705714022e-11,
+      "loss": 0.577,
+      "step": 14182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2252839477477907e-11,
+      "loss": 0.849,
+      "step": 14183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.002086635012675e-12,
+      "loss": 0.92,
+      "step": 14184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.251449338190441e-12,
+      "loss": 0.9039,
+      "step": 14185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.000927726899306e-12,
+      "loss": 0.8205,
+      "step": 14186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.250521912161574e-12,
+      "loss": 0.6133,
+      "step": 14187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0002319816848627e-12,
+      "loss": 0.8903,
+      "step": 14188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5005799875188473e-13,
+      "loss": 0.8668,
+      "step": 14189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7352,
+      "step": 14190
+    },
+    {
+      "epoch": 1.0,
+      "step": 14190,
+      "total_flos": 2.1750800480259277e+17,
+      "train_loss": 0.9389607099515607,
+      "train_runtime": 80626.4831,
+      "train_samples_per_second": 22.528,
+      "train_steps_per_second": 0.176
+    }
+  ],
+  "max_steps": 14190,
+  "num_train_epochs": 1,
+  "total_flos": 2.1750800480259277e+17,
+  "trial_name": null,
+  "trial_params": null
+}