diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,70872 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999576540334533,
+  "eval_steps": 500,
+  "global_step": 11807,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-08,
+      "loss": 2.7681,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802818e-07,
+      "loss": 2.9693,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6901408450704225e-07,
+      "loss": 2.9468,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2535211267605636e-07,
+      "loss": 2.9982,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 2.9272,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.380281690140845e-07,
+      "loss": 2.7491,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9436619718309864e-07,
+      "loss": 3.077,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.507042253521127e-07,
+      "loss": 2.8191,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.070422535211268e-07,
+      "loss": 2.9243,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 2.8198,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.197183098591549e-07,
+      "loss": 3.0111,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.76056338028169e-07,
+      "loss": 2.8129,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.323943661971832e-07,
+      "loss": 2.7719,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.887323943661973e-07,
+      "loss": 2.8399,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 2.811,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.014084507042254e-07,
+      "loss": 2.8347,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.577464788732395e-07,
+      "loss": 3.3533,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0140845070422536e-06,
+      "loss": 3.0644,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0704225352112677e-06,
+      "loss": 3.7213,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 3.4961,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1830985915492958e-06,
+      "loss": 4.1043,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2394366197183099e-06,
+      "loss": 3.7088,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.295774647887324e-06,
+      "loss": 4.3867,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.352112676056338e-06,
+      "loss": 3.8109,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 3.8093,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4647887323943664e-06,
+      "loss": 4.0909,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5211267605633803e-06,
+      "loss": 3.9205,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5774647887323946e-06,
+      "loss": 3.4115,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6338028169014086e-06,
+      "loss": 3.5513,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 3.3566,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7464788732394366e-06,
+      "loss": 3.331,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8028169014084509e-06,
+      "loss": 3.3178,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.859154929577465e-06,
+      "loss": 3.118,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.915492957746479e-06,
+      "loss": 2.7772,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 2.6527,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.028169014084507e-06,
+      "loss": 2.51,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0845070422535212e-06,
+      "loss": 2.4016,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1408450704225353e-06,
+      "loss": 2.2789,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.19718309859155e-06,
+      "loss": 2.2039,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 2.2333,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3098591549295775e-06,
+      "loss": 2.3098,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3661971830985916e-06,
+      "loss": 2.2617,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.422535211267606e-06,
+      "loss": 2.1875,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4788732394366198e-06,
+      "loss": 2.0279,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 2.135,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.591549295774648e-06,
+      "loss": 2.1417,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.647887323943662e-06,
+      "loss": 2.2273,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.704225352112676e-06,
+      "loss": 2.0827,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7605633802816906e-06,
+      "loss": 2.0426,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 2.1192,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8732394366197187e-06,
+      "loss": 2.1116,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.929577464788733e-06,
+      "loss": 2.2034,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9859154929577465e-06,
+      "loss": 2.0058,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0422535211267605e-06,
+      "loss": 2.2411,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 2.0255,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.154929577464789e-06,
+      "loss": 2.0777,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.211267605633803e-06,
+      "loss": 2.0245,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2676056338028173e-06,
+      "loss": 2.0609,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3239436619718313e-06,
+      "loss": 2.0186,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 1.9153,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4366197183098595e-06,
+      "loss": 1.9111,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.492957746478873e-06,
+      "loss": 1.9662,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.549295774647888e-06,
+      "loss": 1.8458,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6056338028169017e-06,
+      "loss": 1.7621,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 1.9421,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71830985915493e-06,
+      "loss": 1.9112,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.774647887323944e-06,
+      "loss": 1.72,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.830985915492958e-06,
+      "loss": 1.7189,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.887323943661972e-06,
+      "loss": 1.823,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 1.7879,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.767,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.056338028169014e-06,
+      "loss": 1.5989,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.112676056338029e-06,
+      "loss": 1.6675,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1690140845070425e-06,
+      "loss": 7.2376,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 1.6167,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.281690140845071e-06,
+      "loss": 1.6149,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.338028169014084e-06,
+      "loss": 1.6063,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3943661971831e-06,
+      "loss": 1.5994,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.450704225352113e-06,
+      "loss": 1.5998,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 1.563,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5633802816901414e-06,
+      "loss": 1.5559,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.619718309859155e-06,
+      "loss": 1.5829,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.67605633802817e-06,
+      "loss": 1.5213,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.732394366197183e-06,
+      "loss": 1.5364,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 1.5303,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845070422535212e-06,
+      "loss": 1.4997,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.901408450704226e-06,
+      "loss": 1.4641,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9577464788732395e-06,
+      "loss": 1.4878,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.014084507042254e-06,
+      "loss": 1.4892,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 1.4205,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.126760563380282e-06,
+      "loss": 1.4925,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.183098591549296e-06,
+      "loss": 1.436,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.23943661971831e-06,
+      "loss": 1.4496,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.295774647887324e-06,
+      "loss": 1.4823,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 1.4211,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.408450704225352e-06,
+      "loss": 1.4152,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4647887323943675e-06,
+      "loss": 1.4142,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.521126760563381e-06,
+      "loss": 1.389,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.577464788732395e-06,
+      "loss": 1.4294,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 1.3983,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.690140845070423e-06,
+      "loss": 1.4193,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7464788732394374e-06,
+      "loss": 1.4363,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.802816901408451e-06,
+      "loss": 1.4176,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.859154929577466e-06,
+      "loss": 1.3762,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 1.363,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.971830985915493e-06,
+      "loss": 1.3764,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.028169014084507e-06,
+      "loss": 1.3516,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.084507042253521e-06,
+      "loss": 1.4018,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1408450704225356e-06,
+      "loss": 1.4142,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 1.3542,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2535211267605646e-06,
+      "loss": 1.3272,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.309859154929578e-06,
+      "loss": 1.3812,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.366197183098593e-06,
+      "loss": 1.3403,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.422535211267606e-06,
+      "loss": 1.4074,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 1.4027,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5352112676056345e-06,
+      "loss": 1.3014,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.591549295774648e-06,
+      "loss": 1.329,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.647887323943663e-06,
+      "loss": 1.3629,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.704225352112676e-06,
+      "loss": 1.3423,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 1.3959,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.8169014084507045e-06,
+      "loss": 1.3032,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.873239436619719e-06,
+      "loss": 6.8262,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.929577464788733e-06,
+      "loss": 1.3416,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.985915492957746e-06,
+      "loss": 1.3897,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 1.3435,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.098591549295776e-06,
+      "loss": 1.3382,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.15492957746479e-06,
+      "loss": 1.3398,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.211267605633803e-06,
+      "loss": 1.3499,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.267605633802818e-06,
+      "loss": 1.2258,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 1.336,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.380281690140846e-06,
+      "loss": 1.3142,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.43661971830986e-06,
+      "loss": 1.3386,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.492957746478874e-06,
+      "loss": 1.2826,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.549295774647888e-06,
+      "loss": 6.724,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 1.2855,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.661971830985916e-06,
+      "loss": 1.2833,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.71830985915493e-06,
+      "loss": 1.283,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.774647887323943e-06,
+      "loss": 1.3436,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.830985915492958e-06,
+      "loss": 1.3461,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 1.2958,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.943661971830987e-06,
+      "loss": 1.279,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.329,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.056338028169016e-06,
+      "loss": 1.2737,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.112676056338029e-06,
+      "loss": 1.3035,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 1.2332,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.225352112676058e-06,
+      "loss": 1.2836,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.28169014084507e-06,
+      "loss": 1.3235,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.338028169014085e-06,
+      "loss": 1.3608,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.3943661971831e-06,
+      "loss": 1.2169,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 1.2845,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.507042253521127e-06,
+      "loss": 1.2784,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.563380281690141e-06,
+      "loss": 1.2503,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.619718309859156e-06,
+      "loss": 1.3223,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.676056338028169e-06,
+      "loss": 1.2317,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 1.2686,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.7887323943662e-06,
+      "loss": 1.24,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.845070422535212e-06,
+      "loss": 1.2965,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.901408450704227e-06,
+      "loss": 1.2548,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.957746478873241e-06,
+      "loss": 1.2672,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 1.2729,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.070422535211268e-06,
+      "loss": 1.2554,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.126760563380283e-06,
+      "loss": 1.237,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.183098591549296e-06,
+      "loss": 1.2013,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.23943661971831e-06,
+      "loss": 1.2793,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 1.2444,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.35211267605634e-06,
+      "loss": 1.1995,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.408450704225352e-06,
+      "loss": 1.2421,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.464788732394366e-06,
+      "loss": 1.2352,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.521126760563381e-06,
+      "loss": 1.248,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 1.2579,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.63380281690141e-06,
+      "loss": 1.2222,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.690140845070424e-06,
+      "loss": 1.2235,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.746478873239437e-06,
+      "loss": 1.2903,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.802816901408452e-06,
+      "loss": 1.2314,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 1.2501,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.915492957746479e-06,
+      "loss": 1.2561,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.971830985915494e-06,
+      "loss": 1.2246,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0028169014084508e-05,
+      "loss": 1.2853,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0084507042253523e-05,
+      "loss": 1.2962,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 6.8509,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.019718309859155e-05,
+      "loss": 1.201,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0253521126760564e-05,
+      "loss": 1.2214,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0309859154929577e-05,
+      "loss": 1.247,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0366197183098592e-05,
+      "loss": 1.2534,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 1.2559,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.047887323943662e-05,
+      "loss": 1.2658,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0535211267605633e-05,
+      "loss": 1.1781,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0591549295774648e-05,
+      "loss": 1.2008,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0647887323943662e-05,
+      "loss": 1.1984,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 1.2465,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.076056338028169e-05,
+      "loss": 1.2155,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0816901408450704e-05,
+      "loss": 1.1693,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.087323943661972e-05,
+      "loss": 6.2666,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0929577464788735e-05,
+      "loss": 6.1364,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 1.204,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1042253521126762e-05,
+      "loss": 1.2128,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1098591549295777e-05,
+      "loss": 1.1934,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.115492957746479e-05,
+      "loss": 1.2255,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1211267605633804e-05,
+      "loss": 1.2123,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 1.2204,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1323943661971833e-05,
+      "loss": 1.2379,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1380281690140846e-05,
+      "loss": 1.1631,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.143661971830986e-05,
+      "loss": 1.2181,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1492957746478875e-05,
+      "loss": 1.2106,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 1.231,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1605633802816902e-05,
+      "loss": 1.2502,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1661971830985917e-05,
+      "loss": 1.202,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1718309859154931e-05,
+      "loss": 1.2021,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1774647887323944e-05,
+      "loss": 1.1754,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 1.2074,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1887323943661973e-05,
+      "loss": 1.1944,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1943661971830986e-05,
+      "loss": 1.1697,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 1.1972,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2056338028169015e-05,
+      "loss": 1.2012,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 6.2109,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2169014084507042e-05,
+      "loss": 1.1801,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2225352112676057e-05,
+      "loss": 1.1775,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2281690140845071e-05,
+      "loss": 1.2071,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2338028169014084e-05,
+      "loss": 1.2476,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 1.1671,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2450704225352113e-05,
+      "loss": 1.2423,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2507042253521129e-05,
+      "loss": 1.158,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2563380281690144e-05,
+      "loss": 1.154,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2619718309859156e-05,
+      "loss": 1.2171,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 6.2641,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2732394366197185e-05,
+      "loss": 1.2058,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2788732394366198e-05,
+      "loss": 5.8665,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2845070422535213e-05,
+      "loss": 1.1936,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2901408450704227e-05,
+      "loss": 1.2153,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 1.1598,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3014084507042255e-05,
+      "loss": 1.1434,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3070422535211269e-05,
+      "loss": 1.1946,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3126760563380284e-05,
+      "loss": 1.1983,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3183098591549296e-05,
+      "loss": 1.2181,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 1.1556,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3295774647887325e-05,
+      "loss": 1.1856,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.335211267605634e-05,
+      "loss": 1.1814,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3408450704225353e-05,
+      "loss": 1.1342,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3464788732394367e-05,
+      "loss": 1.1704,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 1.1986,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3577464788732394e-05,
+      "loss": 1.2058,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3633802816901409e-05,
+      "loss": 1.2105,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3690140845070423e-05,
+      "loss": 1.1827,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3746478873239438e-05,
+      "loss": 1.1591,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 1.1405,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3859154929577465e-05,
+      "loss": 1.1762,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391549295774648e-05,
+      "loss": 1.1897,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3971830985915493e-05,
+      "loss": 1.2576,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4028169014084507e-05,
+      "loss": 1.1975,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 1.191,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4140845070422536e-05,
+      "loss": 6.4414,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4197183098591552e-05,
+      "loss": 6.0461,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4253521126760565e-05,
+      "loss": 1.2252,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.430985915492958e-05,
+      "loss": 1.1235,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 1.1237,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4422535211267607e-05,
+      "loss": 1.1566,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4478873239436621e-05,
+      "loss": 1.1476,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4535211267605636e-05,
+      "loss": 1.1675,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.459154929577465e-05,
+      "loss": 1.1827,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 1.182,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4704225352112678e-05,
+      "loss": 1.2161,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4760563380281692e-05,
+      "loss": 1.1515,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4816901408450705e-05,
+      "loss": 1.1913,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.487323943661972e-05,
+      "loss": 1.2028,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 1.1659,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4985915492957748e-05,
+      "loss": 1.1808,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5042253521126761e-05,
+      "loss": 1.2098,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5098591549295776e-05,
+      "loss": 1.1748,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.515492957746479e-05,
+      "loss": 1.1583,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 1.1483,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5267605633802818e-05,
+      "loss": 1.1836,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5323943661971832e-05,
+      "loss": 1.1694,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5380281690140847e-05,
+      "loss": 1.1844,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.543661971830986e-05,
+      "loss": 1.189,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 1.1532,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5549295774647887e-05,
+      "loss": 1.141,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56056338028169e-05,
+      "loss": 1.1683,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5661971830985916e-05,
+      "loss": 1.1317,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.571830985915493e-05,
+      "loss": 1.1366,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 1.227,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.583098591549296e-05,
+      "loss": 1.1728,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5887323943661974e-05,
+      "loss": 1.1958,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5943661971830988e-05,
+      "loss": 1.1673,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 7.653,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 1.2041,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.611267605633803e-05,
+      "loss": 1.1395,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6169014084507043e-05,
+      "loss": 1.122,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6225352112676057e-05,
+      "loss": 1.1549,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6281690140845072e-05,
+      "loss": 1.1768,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 1.1673,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.63943661971831e-05,
+      "loss": 1.1047,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6450704225352115e-05,
+      "loss": 1.1117,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.650704225352113e-05,
+      "loss": 1.154,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.656338028169014e-05,
+      "loss": 1.1683,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 1.1544,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.667605633802817e-05,
+      "loss": 1.2377,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6732394366197184e-05,
+      "loss": 1.1451,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.67887323943662e-05,
+      "loss": 1.1818,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6845070422535213e-05,
+      "loss": 1.2047,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 1.2753,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.695774647887324e-05,
+      "loss": 1.1427,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7014084507042253e-05,
+      "loss": 1.1279,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7070422535211268e-05,
+      "loss": 1.1961,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7126760563380282e-05,
+      "loss": 1.1342,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 1.1418,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.723943661971831e-05,
+      "loss": 1.1788,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7295774647887326e-05,
+      "loss": 1.2082,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7352112676056337e-05,
+      "loss": 1.1224,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.740845070422535e-05,
+      "loss": 1.1588,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 1.1622,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7521126760563384e-05,
+      "loss": 1.138,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.75774647887324e-05,
+      "loss": 1.1806,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.763380281690141e-05,
+      "loss": 1.1247,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7690140845070424e-05,
+      "loss": 1.124,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 1.1677,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7802816901408453e-05,
+      "loss": 1.1164,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7859154929577468e-05,
+      "loss": 1.1423,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7915492957746482e-05,
+      "loss": 1.1741,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7971830985915497e-05,
+      "loss": 1.1758,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 9.3816,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8084507042253522e-05,
+      "loss": 1.1002,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8140845070422537e-05,
+      "loss": 1.1498,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.819718309859155e-05,
+      "loss": 1.0954,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8253521126760566e-05,
+      "loss": 1.1544,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 1.1604,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.836619718309859e-05,
+      "loss": 1.172,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8422535211267606e-05,
+      "loss": 1.1502,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.847887323943662e-05,
+      "loss": 1.2058,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8535211267605635e-05,
+      "loss": 1.2084,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 1.2362,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8647887323943664e-05,
+      "loss": 1.1817,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.870422535211268e-05,
+      "loss": 1.1293,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.876056338028169e-05,
+      "loss": 1.1587,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8816901408450704e-05,
+      "loss": 1.1418,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 1.1307,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8929577464788733e-05,
+      "loss": 1.1705,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8985915492957747e-05,
+      "loss": 1.1825,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9042253521126762e-05,
+      "loss": 1.1506,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9098591549295776e-05,
+      "loss": 1.1251,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 1.1073,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9211267605633805e-05,
+      "loss": 1.1727,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.926760563380282e-05,
+      "loss": 1.1547,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9323943661971834e-05,
+      "loss": 1.1291,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.938028169014085e-05,
+      "loss": 1.1331,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 1.1269,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9492957746478875e-05,
+      "loss": 1.1198,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.954929577464789e-05,
+      "loss": 1.131,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9605633802816904e-05,
+      "loss": 1.1756,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9661971830985918e-05,
+      "loss": 1.1297,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 1.1479,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9774647887323947e-05,
+      "loss": 1.1406,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9830985915492958e-05,
+      "loss": 1.1592,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9887323943661973e-05,
+      "loss": 1.1614,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943661971830987e-05,
+      "loss": 1.136,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.1454,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999623723524e-05,
+      "loss": 1.1516,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998494894127e-05,
+      "loss": 1.1474,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996613511883e-05,
+      "loss": 1.122,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993979576947e-05,
+      "loss": 1.1588,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999990593089513e-05,
+      "loss": 1.1419,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986454049833e-05,
+      "loss": 1.1311,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981562458215e-05,
+      "loss": 1.1644,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999975918315035e-05,
+      "loss": 1.1533,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999969521620714e-05,
+      "loss": 1.1536,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999962372375736e-05,
+      "loss": 1.1346,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995447058064e-05,
+      "loss": 1.129,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999945816236013e-05,
+      "loss": 1.1687,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999936409342512e-05,
+      "loss": 1.1563,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999926249900844e-05,
+      "loss": 1.1064,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999915337911776e-05,
+      "loss": 1.1052,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999903673376126e-05,
+      "loss": 1.1446,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989125629477e-05,
+      "loss": 1.1456,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999878086668646e-05,
+      "loss": 1.1481,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999864164498745e-05,
+      "loss": 1.1747,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999849489786114e-05,
+      "loss": 1.131,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999834062531856e-05,
+      "loss": 1.1059,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999817882737132e-05,
+      "loss": 1.128,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999800950403162e-05,
+      "loss": 1.136,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978326553122e-05,
+      "loss": 1.1155,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999764828122638e-05,
+      "loss": 1.1224,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999745638178798e-05,
+      "loss": 1.1484,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972569570115e-05,
+      "loss": 7.1005,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970500069119e-05,
+      "loss": 1.1115,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999683553150476e-05,
+      "loss": 1.1122,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999966135308063e-05,
+      "loss": 1.0752,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999963840048331e-05,
+      "loss": 1.1547,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999614695360252e-05,
+      "loss": 1.1617,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999590237713237e-05,
+      "loss": 1.1622,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956502754411e-05,
+      "loss": 1.1294,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999539064854762e-05,
+      "loss": 1.1569,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999512349647147e-05,
+      "loss": 1.1573,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999484881923285e-05,
+      "loss": 1.1593,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999945666168523e-05,
+      "loss": 1.2001,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999427688935112e-05,
+      "loss": 6.8806,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999397963675112e-05,
+      "loss": 1.129,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999367485907464e-05,
+      "loss": 6.5445,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999336255634464e-05,
+      "loss": 1.1524,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999304272858465e-05,
+      "loss": 1.1207,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999271537581868e-05,
+      "loss": 1.0942,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999238049807137e-05,
+      "loss": 1.1203,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99992038095368e-05,
+      "loss": 1.0914,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999168816773423e-05,
+      "loss": 1.09,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999133071519644e-05,
+      "loss": 1.1375,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999096573778155e-05,
+      "loss": 1.1359,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999059323551703e-05,
+      "loss": 1.0529,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999902132084309e-05,
+      "loss": 1.096,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998982565655175e-05,
+      "loss": 1.1344,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999894305799087e-05,
+      "loss": 1.1366,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998902797853155e-05,
+      "loss": 1.141,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999886178524506e-05,
+      "loss": 1.1615,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998820020169668e-05,
+      "loss": 1.1454,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998777502630126e-05,
+      "loss": 1.1242,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998734232629625e-05,
+      "loss": 1.1228,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999869021017143e-05,
+      "loss": 1.1381,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998645435258856e-05,
+      "loss": 1.1041,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999859990789526e-05,
+      "loss": 1.1296,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999855362808408e-05,
+      "loss": 6.7313,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998506595828794e-05,
+      "loss": 1.1433,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999845881113294e-05,
+      "loss": 1.1785,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998410274000117e-05,
+      "loss": 1.1107,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999836098443398e-05,
+      "loss": 1.1357,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999831094243823e-05,
+      "loss": 1.1479,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999826014801664e-05,
+      "loss": 1.126,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998208601173035e-05,
+      "loss": 1.1101,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998156301911285e-05,
+      "loss": 1.1482,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999810325023533e-05,
+      "loss": 1.1005,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998049446149165e-05,
+      "loss": 1.1406,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997994889656835e-05,
+      "loss": 1.1001,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997939580762452e-05,
+      "loss": 1.143,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997883519470172e-05,
+      "loss": 1.1259,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997826705784214e-05,
+      "loss": 1.1529,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997769139708857e-05,
+      "loss": 1.1523,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999771082124843e-05,
+      "loss": 1.125,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997651750407324e-05,
+      "loss": 1.1132,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997591927189984e-05,
+      "loss": 1.1142,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999753135160091e-05,
+      "loss": 1.0961,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997470023644665e-05,
+      "loss": 1.1031,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999740794332586e-05,
+      "loss": 1.1106,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999734511064917e-05,
+      "loss": 1.1334,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999728152561932e-05,
+      "loss": 1.1291,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997217188241098e-05,
+      "loss": 1.098,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997152098519345e-05,
+      "loss": 1.1297,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999708625645896e-05,
+      "loss": 1.0999,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997019662064896e-05,
+      "loss": 1.1177,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996952315342167e-05,
+      "loss": 1.1462,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996884216295838e-05,
+      "loss": 1.1298,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996815364931036e-05,
+      "loss": 1.1864,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996745761252942e-05,
+      "loss": 1.1128,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996675405266796e-05,
+      "loss": 1.1123,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999660429697789e-05,
+      "loss": 1.1072,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996532436391573e-05,
+      "loss": 1.1794,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999645982351326e-05,
+      "loss": 1.1124,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996386458348412e-05,
+      "loss": 1.1329,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996312340902548e-05,
+      "loss": 1.0875,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996237471181245e-05,
+      "loss": 1.0976,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996161849190144e-05,
+      "loss": 1.1312,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996085474934926e-05,
+      "loss": 1.1166,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996008348421353e-05,
+      "loss": 1.1443,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995930469655214e-05,
+      "loss": 1.1824,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995851838642376e-05,
+      "loss": 1.0905,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999577245538876e-05,
+      "loss": 1.0665,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995692319900333e-05,
+      "loss": 1.1177,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995611432183134e-05,
+      "loss": 1.1126,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999552979224324e-05,
+      "loss": 1.1559,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995447400086804e-05,
+      "loss": 1.1925,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999536425572002e-05,
+      "loss": 1.1038,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999528035914915e-05,
+      "loss": 1.077,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995195710380503e-05,
+      "loss": 1.1164,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995110309420456e-05,
+      "loss": 1.1447,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999502415627543e-05,
+      "loss": 1.1773,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994937250951907e-05,
+      "loss": 1.1105,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994849593456434e-05,
+      "loss": 1.0879,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994761183795603e-05,
+      "loss": 1.1312,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994672021976063e-05,
+      "loss": 1.1093,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994582108004535e-05,
+      "loss": 6.7607,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994491441887776e-05,
+      "loss": 1.13,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994400023632615e-05,
+      "loss": 1.094,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999430785324593e-05,
+      "loss": 1.0977,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994214930734657e-05,
+      "loss": 1.0976,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994121256105785e-05,
+      "loss": 1.1355,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999402682936637e-05,
+      "loss": 1.129,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993931650523516e-05,
+      "loss": 1.0901,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999383571958438e-05,
+      "loss": 1.16,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993739036556192e-05,
+      "loss": 1.1121,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999364160144622e-05,
+      "loss": 1.1088,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993543414261797e-05,
+      "loss": 1.1162,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993444475010318e-05,
+      "loss": 1.1077,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993344783699217e-05,
+      "loss": 1.1191,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993244340336013e-05,
+      "loss": 1.1314,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999314314492825e-05,
+      "loss": 1.1022,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999304119748355e-05,
+      "loss": 1.1032,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992938498009583e-05,
+      "loss": 1.0726,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999283504651408e-05,
+      "loss": 1.1156,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992730843004826e-05,
+      "loss": 1.0711,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999262588748966e-05,
+      "loss": 1.0996,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992520179976487e-05,
+      "loss": 1.102,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992413720473253e-05,
+      "loss": 1.1121,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992306508987974e-05,
+      "loss": 1.071,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992198545528723e-05,
+      "loss": 1.131,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992089830103616e-05,
+      "loss": 1.116,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999198036272084e-05,
+      "loss": 1.122,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991870143388633e-05,
+      "loss": 1.1522,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991759172115287e-05,
+      "loss": 1.1166,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999164744890915e-05,
+      "loss": 1.1489,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999153497377864e-05,
+      "loss": 1.0676,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999142174673222e-05,
+      "loss": 1.1728,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991307767778397e-05,
+      "loss": 1.1325,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999119303692576e-05,
+      "loss": 1.1068,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991077554182943e-05,
+      "loss": 1.1236,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990961319558635e-05,
+      "loss": 1.1846,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990844333061587e-05,
+      "loss": 1.1395,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999072659470059e-05,
+      "loss": 1.1456,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999060810448452e-05,
+      "loss": 6.7552,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990488862422284e-05,
+      "loss": 1.119,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999036886852286e-05,
+      "loss": 1.057,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990248122795276e-05,
+      "loss": 1.084,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990126625248623e-05,
+      "loss": 1.1146,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990004375892037e-05,
+      "loss": 1.1132,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989881374734726e-05,
+      "loss": 1.1246,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989757621785943e-05,
+      "loss": 1.1658,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989633117054996e-05,
+      "loss": 1.1137,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989507860551263e-05,
+      "loss": 1.0864,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989381852284165e-05,
+      "loss": 1.038,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989255092263188e-05,
+      "loss": 1.0835,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989127580497868e-05,
+      "loss": 1.1058,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988999316997803e-05,
+      "loss": 1.1584,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988870301772647e-05,
+      "loss": 1.1062,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988740534832107e-05,
+      "loss": 1.1412,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988610016185947e-05,
+      "loss": 1.0912,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988478745843993e-05,
+      "loss": 1.1146,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998834672381612e-05,
+      "loss": 1.1115,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988213950112264e-05,
+      "loss": 1.1211,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998808042474242e-05,
+      "loss": 1.1321,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987946147716635e-05,
+      "loss": 1.0896,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987811119045014e-05,
+      "loss": 1.0724,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987675338737715e-05,
+      "loss": 1.1129,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987538806804966e-05,
+      "loss": 1.0793,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998740152325703e-05,
+      "loss": 1.0961,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987263488104243e-05,
+      "loss": 1.1078,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987124701356998e-05,
+      "loss": 1.1325,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998698516302573e-05,
+      "loss": 1.132,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998684487312095e-05,
+      "loss": 1.1413,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998670383165321e-05,
+      "loss": 1.0835,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986562038633117e-05,
+      "loss": 1.1153,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986419494071354e-05,
+      "loss": 1.1084,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986276197978644e-05,
+      "loss": 1.1545,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986132150365772e-05,
+      "loss": 1.1384,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985987351243573e-05,
+      "loss": 1.1324,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985841800622948e-05,
+      "loss": 1.1083,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998569549851485e-05,
+      "loss": 1.0699,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985548444930295e-05,
+      "loss": 1.0713,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985400639880337e-05,
+      "loss": 1.1127,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985252083376108e-05,
+      "loss": 1.1386,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985102775428785e-05,
+      "loss": 1.0734,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984952716049603e-05,
+      "loss": 1.1105,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998480190524986e-05,
+      "loss": 1.104,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984650343040895e-05,
+      "loss": 1.0613,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984498029434127e-05,
+      "loss": 1.118,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998434496444101e-05,
+      "loss": 1.0873,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998419114807306e-05,
+      "loss": 1.1031,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984036580341865e-05,
+      "loss": 1.0984,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983881261259044e-05,
+      "loss": 1.1015,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983725190836294e-05,
+      "loss": 1.0877,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998356836908536e-05,
+      "loss": 1.0676,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983410796018033e-05,
+      "loss": 1.1061,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998325247164618e-05,
+      "loss": 1.0782,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998309339598172e-05,
+      "loss": 1.0498,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982933569036617e-05,
+      "loss": 1.1157,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99827729908229e-05,
+      "loss": 1.1184,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998261166135265e-05,
+      "loss": 1.0954,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998244958063802e-05,
+      "loss": 1.1608,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982286748691194e-05,
+      "loss": 1.112,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982123165524433e-05,
+      "loss": 1.1763,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981958831150047e-05,
+      "loss": 1.0969,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99817937455804e-05,
+      "loss": 1.0999,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998162790882792e-05,
+      "loss": 1.0666,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998146132090508e-05,
+      "loss": 1.0984,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981293981824426e-05,
+      "loss": 1.1209,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981125891598545e-05,
+      "loss": 1.1009,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998095705024009e-05,
+      "loss": 1.0893,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998078745776176e-05,
+      "loss": 1.0582,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980617114176327e-05,
+      "loss": 1.1,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980446019496606e-05,
+      "loss": 1.1153,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980274173735475e-05,
+      "loss": 1.0582,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980101576905858e-05,
+      "loss": 1.1002,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979928229020756e-05,
+      "loss": 1.1291,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979754130093207e-05,
+      "loss": 1.1311,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979579280136313e-05,
+      "loss": 1.117,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979403679163237e-05,
+      "loss": 5.9381,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979227327187188e-05,
+      "loss": 1.0509,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979050224221442e-05,
+      "loss": 1.1207,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997887237027932e-05,
+      "loss": 1.0742,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978693765374217e-05,
+      "loss": 1.0725,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978514409519566e-05,
+      "loss": 1.1019,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997833430272887e-05,
+      "loss": 1.0723,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978153445015677e-05,
+      "loss": 1.1454,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977971836393602e-05,
+      "loss": 1.0903,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977789476876307e-05,
+      "loss": 1.0988,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977606366477525e-05,
+      "loss": 1.0678,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977422505211022e-05,
+      "loss": 1.1023,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977237893090652e-05,
+      "loss": 1.1104,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977052530130292e-05,
+      "loss": 1.121,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99768664163439e-05,
+      "loss": 1.1007,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997667955174548e-05,
+      "loss": 1.0822,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997649193634909e-05,
+      "loss": 1.0993,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997630357016886e-05,
+      "loss": 1.0897,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997611445321896e-05,
+      "loss": 1.0863,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975924585513617e-05,
+      "loss": 1.1233,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975733967067124e-05,
+      "loss": 1.1062,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975542597893832e-05,
+      "loss": 1.1371,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997535047800813e-05,
+      "loss": 1.1019,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975157607424488e-05,
+      "loss": 1.0273,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997496398615741e-05,
+      "loss": 1.0552,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974769614221475e-05,
+      "loss": 1.1111,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974574491631308e-05,
+      "loss": 1.0734,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997437861840159e-05,
+      "loss": 1.1141,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974181994547063e-05,
+      "loss": 1.0413,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997398462008253e-05,
+      "loss": 1.1022,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973786495022836e-05,
+      "loss": 1.0988,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973587619382898e-05,
+      "loss": 1.0925,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973387993177673e-05,
+      "loss": 1.168,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973187616422195e-05,
+      "loss": 1.1225,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972986489131537e-05,
+      "loss": 1.0917,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972784611320835e-05,
+      "loss": 1.113,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972581983005287e-05,
+      "loss": 1.0811,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972378604200133e-05,
+      "loss": 1.1097,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972174474920684e-05,
+      "loss": 1.0011,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99719695951823e-05,
+      "loss": 1.063,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99717639650004e-05,
+      "loss": 1.0931,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997155758439046e-05,
+      "loss": 1.0909,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997135045336801e-05,
+      "loss": 1.086,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971142571948638e-05,
+      "loss": 1.0962,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970933940147983e-05,
+      "loss": 1.066,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970724557981755e-05,
+      "loss": 1.0697,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970514425465706e-05,
+      "loss": 1.075,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970303542615648e-05,
+      "loss": 1.123,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970091909447453e-05,
+      "loss": 5.7262,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996987952597705e-05,
+      "loss": 1.1092,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996966639222042e-05,
+      "loss": 1.0433,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99694525081936e-05,
+      "loss": 1.0897,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996923787391269e-05,
+      "loss": 1.0948,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996902248939384e-05,
+      "loss": 1.0541,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968806354653258e-05,
+      "loss": 1.0725,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996858946970721e-05,
+      "loss": 1.0477,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996837183457202e-05,
+      "loss": 1.0832,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968153449264064e-05,
+      "loss": 1.1123,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967934313799776e-05,
+      "loss": 5.7261,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996771442819565e-05,
+      "loss": 1.1589,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996749379246823e-05,
+      "loss": 1.1036,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967272406634122e-05,
+      "loss": 1.0285,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967050270709986e-05,
+      "loss": 5.5694,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996682738471254e-05,
+      "loss": 1.0576,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996660374865856e-05,
+      "loss": 1.1235,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966379362564865e-05,
+      "loss": 1.0428,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966154226448352e-05,
+      "loss": 1.1177,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996592834032596e-05,
+      "loss": 1.0968,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996570170421469e-05,
+      "loss": 1.0585,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996547431813159e-05,
+      "loss": 1.0467,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965246182093783e-05,
+      "loss": 1.0633,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996501729611843e-05,
+      "loss": 1.0589,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996478766022276e-05,
+      "loss": 1.1402,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964557274424053e-05,
+      "loss": 1.1173,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964326138739645e-05,
+      "loss": 1.0834,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996409425318693e-05,
+      "loss": 1.0891,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963861617783365e-05,
+      "loss": 1.1069,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963628232546448e-05,
+      "loss": 1.0815,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996339409749375e-05,
+      "loss": 1.0942,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996315921264288e-05,
+      "loss": 1.0613,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962923578011527e-05,
+      "loss": 5.3155,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996268719361742e-05,
+      "loss": 1.0893,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962450059478346e-05,
+      "loss": 1.1021,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962212175612152e-05,
+      "loss": 1.0367,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961973542036738e-05,
+      "loss": 1.0596,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961734158770063e-05,
+      "loss": 1.1116,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961494025830143e-05,
+      "loss": 1.1087,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996125314323505e-05,
+      "loss": 1.1151,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996101151100291e-05,
+      "loss": 1.1217,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960769129151907e-05,
+      "loss": 1.0917,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960525997700283e-05,
+      "loss": 1.132,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960282116666334e-05,
+      "loss": 1.1518,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960037486068417e-05,
+      "loss": 1.0764,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959792105924932e-05,
+      "loss": 1.0859,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959545976254353e-05,
+      "loss": 1.0641,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959299097075205e-05,
+      "loss": 1.1047,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959051468406058e-05,
+      "loss": 1.0865,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958803090265554e-05,
+      "loss": 1.0749,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958553962672386e-05,
+      "loss": 1.0796,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958304085645296e-05,
+      "loss": 1.0581,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958053459203092e-05,
+      "loss": 1.0871,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957802083364636e-05,
+      "loss": 1.1071,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957549958148844e-05,
+      "loss": 1.1209,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957297083574692e-05,
+      "loss": 1.1,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957043459661202e-05,
+      "loss": 1.0754,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956789086427473e-05,
+      "loss": 1.0756,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956533963892636e-05,
+      "loss": 1.1136,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956278092075904e-05,
+      "loss": 1.1237,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956021470996517e-05,
+      "loss": 1.051,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955764100673797e-05,
+      "loss": 1.1218,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995550598112711e-05,
+      "loss": 1.1133,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995524711237588e-05,
+      "loss": 1.0852,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995498749443959e-05,
+      "loss": 6.2748,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954727127337778e-05,
+      "loss": 1.0886,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995446601109004e-05,
+      "loss": 1.1452,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954204145716017e-05,
+      "loss": 1.12,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953941531235427e-05,
+      "loss": 1.1096,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953678167668022e-05,
+      "loss": 1.0489,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953414055033634e-05,
+      "loss": 1.108,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995314919335213e-05,
+      "loss": 1.1124,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952883582643444e-05,
+      "loss": 1.1339,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952617222927565e-05,
+      "loss": 1.088,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995235011422454e-05,
+      "loss": 1.0467,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952082256554468e-05,
+      "loss": 1.0779,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951813649937508e-05,
+      "loss": 1.1033,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995154429439387e-05,
+      "loss": 6.223,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951274189943834e-05,
+      "loss": 1.0616,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951003336607716e-05,
+      "loss": 1.1184,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950731734405907e-05,
+      "loss": 1.094,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995045938335884e-05,
+      "loss": 1.1092,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995018628348702e-05,
+      "loss": 1.0783,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949912434810987e-05,
+      "loss": 1.081,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949637837351358e-05,
+      "loss": 1.0384,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99493624911288e-05,
+      "loss": 1.0874,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949086396164025e-05,
+      "loss": 1.0786,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994880955247782e-05,
+      "loss": 1.0337,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994853196009101e-05,
+      "loss": 1.0708,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994825361902449e-05,
+      "loss": 1.0883,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994797452929921e-05,
+      "loss": 5.6613,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994769469093617e-05,
+      "loss": 1.0559,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947414103956425e-05,
+      "loss": 1.0564,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947132768381098e-05,
+      "loss": 1.0843,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994685068423136e-05,
+      "loss": 1.0931,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946567851528432e-05,
+      "loss": 1.0757,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946284270293606e-05,
+      "loss": 1.0775,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945999940548216e-05,
+      "loss": 1.1038,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994571486231367e-05,
+      "loss": 1.0676,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945429035611413e-05,
+      "loss": 1.092,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945142460462958e-05,
+      "loss": 1.1361,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944855136889872e-05,
+      "loss": 1.0133,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944567064913774e-05,
+      "loss": 1.0918,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944278244556347e-05,
+      "loss": 1.0996,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943988675839323e-05,
+      "loss": 1.0998,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943698358784496e-05,
+      "loss": 1.1303,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943407293413716e-05,
+      "loss": 1.0658,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994311547974888e-05,
+      "loss": 1.1154,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942822917811954e-05,
+      "loss": 1.0641,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942529607624953e-05,
+      "loss": 1.0686,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942235549209955e-05,
+      "loss": 1.0143,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941940742589085e-05,
+      "loss": 1.084,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941645187784523e-05,
+      "loss": 1.1012,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994134888481852e-05,
+      "loss": 1.1267,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941051833713375e-05,
+      "loss": 5.5405,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940754034491437e-05,
+      "loss": 1.0984,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940455487175116e-05,
+      "loss": 1.1267,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940156191786886e-05,
+      "loss": 1.0583,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939856148349267e-05,
+      "loss": 1.0937,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939555356884835e-05,
+      "loss": 5.4334,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939253817416232e-05,
+      "loss": 5.3205,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993895152996615e-05,
+      "loss": 1.097,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938648494557334e-05,
+      "loss": 1.1196,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938344711212592e-05,
+      "loss": 1.1172,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938040179954786e-05,
+      "loss": 1.0529,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937734900806827e-05,
+      "loss": 1.08,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99374288737917e-05,
+      "loss": 1.0531,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 1.0837,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936814576252096e-05,
+      "loss": 1.0602,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993650630577385e-05,
+      "loss": 1.106,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936197287520888e-05,
+      "loss": 1.1132,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993588752151647e-05,
+      "loss": 1.0711,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935577007783897e-05,
+      "loss": 1.1048,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935265746346544e-05,
+      "loss": 1.0935,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934953737227834e-05,
+      "loss": 1.0842,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993464098045125e-05,
+      "loss": 1.1234,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934327476040323e-05,
+      "loss": 5.4383,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993401322401865e-05,
+      "loss": 1.1001,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933698224409876e-05,
+      "loss": 1.1171,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993338247723771e-05,
+      "loss": 1.0665,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933065982525918e-05,
+      "loss": 1.0966,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932748740298306e-05,
+      "loss": 1.1104,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993243075057876e-05,
+      "loss": 0.9764,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932112013391203e-05,
+      "loss": 1.0451,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931792528759622e-05,
+      "loss": 1.101,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931472296708066e-05,
+      "loss": 1.0655,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931151317260628e-05,
+      "loss": 1.0721,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993082959044146e-05,
+      "loss": 1.06,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993050711627479e-05,
+      "loss": 1.0937,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930183894784868e-05,
+      "loss": 1.0849,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929859925996027e-05,
+      "loss": 1.0637,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929535209932646e-05,
+      "loss": 1.084,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929209746619157e-05,
+      "loss": 1.1375,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928883536080063e-05,
+      "loss": 1.0923,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99285565783399e-05,
+      "loss": 1.0581,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928228873423285e-05,
+      "loss": 1.0844,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927900421354875e-05,
+      "loss": 1.0677,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927571222159387e-05,
+      "loss": 1.0689,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927241275861595e-05,
+      "loss": 1.1177,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992691058248633e-05,
+      "loss": 1.1048,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992657914205848e-05,
+      "loss": 1.1093,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926246954602982e-05,
+      "loss": 1.0809,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925914020144842e-05,
+      "loss": 1.0715,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992558033870911e-05,
+      "loss": 1.0563,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99252459103209e-05,
+      "loss": 1.0767,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924910735005376e-05,
+      "loss": 1.0756,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924574812787766e-05,
+      "loss": 1.0704,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924238143693348e-05,
+      "loss": 1.0681,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992390072774746e-05,
+      "loss": 1.0368,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923562564975492e-05,
+      "loss": 1.0936,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923223655402895e-05,
+      "loss": 1.1021,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922883999055167e-05,
+      "loss": 1.0537,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992254359595788e-05,
+      "loss": 5.2371,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992220244613664e-05,
+      "loss": 1.0871,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992186054961713e-05,
+      "loss": 1.0619,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992151790642507e-05,
+      "loss": 1.0952,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921174516586252e-05,
+      "loss": 1.1223,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920830380126516e-05,
+      "loss": 1.0687,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920485497071764e-05,
+      "loss": 1.0437,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992013986744794e-05,
+      "loss": 1.0612,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991979349128107e-05,
+      "loss": 1.1148,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991944636859721e-05,
+      "loss": 1.0747,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991909849942248e-05,
+      "loss": 1.0895,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918749883783068e-05,
+      "loss": 5.3154,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918400521705203e-05,
+      "loss": 1.0989,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991805041321518e-05,
+      "loss": 1.0544,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991769955833934e-05,
+      "loss": 1.0946,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917347957104098e-05,
+      "loss": 1.0729,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916995609535906e-05,
+      "loss": 1.0617,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916642515661278e-05,
+      "loss": 1.0937,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991628867550679e-05,
+      "loss": 1.0979,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915934089099074e-05,
+      "loss": 1.0804,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915578756464806e-05,
+      "loss": 1.0893,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915222677630734e-05,
+      "loss": 1.1002,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991486585262365e-05,
+      "loss": 1.0647,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914508281470413e-05,
+      "loss": 1.0223,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914149964197923e-05,
+      "loss": 1.06,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991379090083315e-05,
+      "loss": 1.0667,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991343109140312e-05,
+      "loss": 1.0946,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913070535934905e-05,
+      "loss": 1.0211,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912709234455638e-05,
+      "loss": 1.0486,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912347186992512e-05,
+      "loss": 1.0693,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911984393572773e-05,
+      "loss": 1.0842,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991162085422372e-05,
+      "loss": 1.1189,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911256568972715e-05,
+      "loss": 1.0689,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910891537847166e-05,
+      "loss": 1.0959,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910525760874555e-05,
+      "loss": 1.0399,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910159238082397e-05,
+      "loss": 1.0623,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990979196949828e-05,
+      "loss": 1.0958,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909423955149844e-05,
+      "loss": 1.0614,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909055195064782e-05,
+      "loss": 1.0736,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908685689270845e-05,
+      "loss": 1.0492,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990831543779584e-05,
+      "loss": 1.0465,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907944440667637e-05,
+      "loss": 1.1165,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907572697914144e-05,
+      "loss": 1.0698,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907200209563345e-05,
+      "loss": 1.094,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906826975643264e-05,
+      "loss": 1.0805,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906452996182e-05,
+      "loss": 1.0906,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990607827120769e-05,
+      "loss": 1.0741,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905702800748533e-05,
+      "loss": 1.0839,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905326584832786e-05,
+      "loss": 1.0373,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904949623488766e-05,
+      "loss": 1.0724,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904571916744836e-05,
+      "loss": 1.0771,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904193464629423e-05,
+      "loss": 1.085,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903814267171006e-05,
+      "loss": 1.0575,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903434324398118e-05,
+      "loss": 1.0778,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903053636339364e-05,
+      "loss": 1.1046,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902672203023377e-05,
+      "loss": 1.0608,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902290024478873e-05,
+      "loss": 1.0956,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990190710073461e-05,
+      "loss": 1.083,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901523431819404e-05,
+      "loss": 1.0708,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990113901776213e-05,
+      "loss": 1.1009,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900753858591714e-05,
+      "loss": 1.1131,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900367954337144e-05,
+      "loss": 1.0524,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989998130502746e-05,
+      "loss": 1.0764,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989959391069176e-05,
+      "loss": 1.078,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899205771359198e-05,
+      "loss": 1.0683,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989881688705898e-05,
+      "loss": 1.1089,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989842725782038e-05,
+      "loss": 1.0346,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898036883672713e-05,
+      "loss": 1.1039,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897645764645353e-05,
+      "loss": 1.0809,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989725390076774e-05,
+      "loss": 1.0948,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896861292069367e-05,
+      "loss": 1.0909,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989646793857977e-05,
+      "loss": 1.0855,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896073840328563e-05,
+      "loss": 1.0845,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989567899734539e-05,
+      "loss": 1.0904,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989528340965998e-05,
+      "loss": 1.133,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989488707730209e-05,
+      "loss": 1.0931,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989449000030155e-05,
+      "loss": 1.0936,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894092178688246e-05,
+      "loss": 1.0819,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893693612492116e-05,
+      "loss": 1.0872,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989329430174315e-05,
+      "loss": 1.1073,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892894246471403e-05,
+      "loss": 1.1216,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892493446706973e-05,
+      "loss": 1.0818,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989209190248003e-05,
+      "loss": 1.0611,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891689613820794e-05,
+      "loss": 1.0565,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989128658075953e-05,
+      "loss": 1.1747,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989088280332658e-05,
+      "loss": 1.0697,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989047828155232e-05,
+      "loss": 1.0572,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98900730154672e-05,
+      "loss": 1.0384,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988966700510171e-05,
+      "loss": 1.0691,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889260250486413e-05,
+      "loss": 1.0593,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888852751651915e-05,
+      "loss": 1.0756,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888444508628887e-05,
+      "loss": 1.0887,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888035521448044e-05,
+      "loss": 1.1049,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887625790140168e-05,
+      "loss": 1.038,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887215314736097e-05,
+      "loss": 1.0288,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886804095266715e-05,
+      "loss": 1.0664,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886392131762973e-05,
+      "loss": 1.0411,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885979424255874e-05,
+      "loss": 5.3497,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988556597277647e-05,
+      "loss": 1.0352,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885151777355888e-05,
+      "loss": 1.1135,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884736838025285e-05,
+      "loss": 1.0671,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884321154815894e-05,
+      "loss": 1.0735,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883904727758994e-05,
+      "loss": 1.1016,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883487556885926e-05,
+      "loss": 5.0628,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883069642228087e-05,
+      "loss": 1.0962,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882650983816922e-05,
+      "loss": 1.1219,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882231581683938e-05,
+      "loss": 1.0895,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881811435860698e-05,
+      "loss": 1.0649,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988139054637882e-05,
+      "loss": 1.0454,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880968913269983e-05,
+      "loss": 4.5705,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880546536565912e-05,
+      "loss": 1.0636,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880123416298393e-05,
+      "loss": 1.0555,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987969955249927e-05,
+      "loss": 1.0602,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987927494520044e-05,
+      "loss": 1.0525,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987884959443386e-05,
+      "loss": 1.0104,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987842350023153e-05,
+      "loss": 1.0642,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987799666262553e-05,
+      "loss": 1.101,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987756908164797e-05,
+      "loss": 1.0444,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877140757331036e-05,
+      "loss": 1.0935,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876711689706954e-05,
+      "loss": 1.0831,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876281878808022e-05,
+      "loss": 1.0204,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875851324666577e-05,
+      "loss": 1.0975,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875420027315028e-05,
+      "loss": 1.0626,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874987986785828e-05,
+      "loss": 1.0786,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874555203111492e-05,
+      "loss": 1.0163,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874121676324588e-05,
+      "loss": 1.0692,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873687406457742e-05,
+      "loss": 1.0419,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873252393543634e-05,
+      "loss": 1.126,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872816637615006e-05,
+      "loss": 1.0862,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872380138704642e-05,
+      "loss": 1.1197,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871942896845397e-05,
+      "loss": 1.0544,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871504912070175e-05,
+      "loss": 1.1035,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871066184411938e-05,
+      "loss": 1.0719,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98706267139037e-05,
+      "loss": 1.0749,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987018650057853e-05,
+      "loss": 1.0673,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869745544469567e-05,
+      "loss": 1.049,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869303845609987e-05,
+      "loss": 1.0682,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986886140403303e-05,
+      "loss": 1.1386,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868418219772e-05,
+      "loss": 1.1008,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867974292860234e-05,
+      "loss": 1.0956,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867529623331158e-05,
+      "loss": 1.0532,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986708421121822e-05,
+      "loss": 1.0582,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986663805655495e-05,
+      "loss": 1.0666,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986619115937492e-05,
+      "loss": 1.111,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986574351971176e-05,
+      "loss": 1.0441,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986529513759916e-05,
+      "loss": 1.0385,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864846013070863e-05,
+      "loss": 1.0828,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864396146160663e-05,
+      "loss": 1.1028,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986394553690242e-05,
+      "loss": 1.1885,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863494185330044e-05,
+      "loss": 1.0609,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98630420914775e-05,
+      "loss": 1.0783,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986258925537881e-05,
+      "loss": 1.042,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862135677068058e-05,
+      "loss": 1.0267,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986168135657937e-05,
+      "loss": 1.0442,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986122629394694e-05,
+      "loss": 1.076,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860770489205013e-05,
+      "loss": 1.0154,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860313942387894e-05,
+      "loss": 1.0698,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985985665352994e-05,
+      "loss": 1.0493,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859398622665557e-05,
+      "loss": 1.0791,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858939849829224e-05,
+      "loss": 1.0137,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858480335055458e-05,
+      "loss": 1.073,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985802007837885e-05,
+      "loss": 1.1024,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857559079834023e-05,
+      "loss": 1.0874,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985709733945568e-05,
+      "loss": 1.0573,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856634857278565e-05,
+      "loss": 1.0982,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856171633337487e-05,
+      "loss": 1.0609,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98557076676673e-05,
+      "loss": 1.0918,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855242960302924e-05,
+      "loss": 1.123,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854777511279328e-05,
+      "loss": 1.0634,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985431132063154e-05,
+      "loss": 4.6648,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985384438839465e-05,
+      "loss": 1.0484,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853376714603784e-05,
+      "loss": 1.0878,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852908299294148e-05,
+      "loss": 1.146,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852439142500984e-05,
+      "loss": 1.0855,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851969244259607e-05,
+      "loss": 1.0378,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851498604605378e-05,
+      "loss": 1.0792,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985102722357371e-05,
+      "loss": 1.057,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985055510120008e-05,
+      "loss": 1.1113,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850082237520018e-05,
+      "loss": 1.0374,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849608632569107e-05,
+      "loss": 0.9808,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849134286382994e-05,
+      "loss": 1.0707,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984865919899737e-05,
+      "loss": 1.0306,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984818337044799e-05,
+      "loss": 1.0977,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984770680077066e-05,
+      "loss": 1.1128,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847229490001253e-05,
+      "loss": 1.0338,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984675143817568e-05,
+      "loss": 1.0101,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846272645329923e-05,
+      "loss": 1.0645,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984579311150001e-05,
+      "loss": 1.1416,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845312836722023e-05,
+      "loss": 1.0765,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844831821032117e-05,
+      "loss": 1.1312,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844350064466488e-05,
+      "loss": 1.0883,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843867567061387e-05,
+      "loss": 1.1095,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843384328853125e-05,
+      "loss": 1.0923,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842900349878068e-05,
+      "loss": 1.0901,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984241563017264e-05,
+      "loss": 1.0845,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841930169773317e-05,
+      "loss": 1.0927,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841443968716634e-05,
+      "loss": 1.0586,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984095702703918e-05,
+      "loss": 1.015,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840469344777595e-05,
+      "loss": 1.0895,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983998092196859e-05,
+      "loss": 1.1425,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839491758648912e-05,
+      "loss": 1.023,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839001854855378e-05,
+      "loss": 1.0945,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838511210624856e-05,
+      "loss": 4.5243,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838019825994266e-05,
+      "loss": 1.0928,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837527701000586e-05,
+      "loss": 1.0501,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983703483568086e-05,
+      "loss": 1.0746,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983654123007217e-05,
+      "loss": 1.1051,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836046884211666e-05,
+      "loss": 1.137,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983555179813655e-05,
+      "loss": 1.0464,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983505597188408e-05,
+      "loss": 1.0904,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834559405491568e-05,
+      "loss": 0.9787,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834062098996386e-05,
+      "loss": 1.0314,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833564052435955e-05,
+      "loss": 1.0815,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833065265847758e-05,
+      "loss": 1.0523,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983256573926933e-05,
+      "loss": 1.0562,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983206547273827e-05,
+      "loss": 1.0969,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831564466292213e-05,
+      "loss": 1.0582,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831062719968874e-05,
+      "loss": 1.1075,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830560233806006e-05,
+      "loss": 1.0092,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830057007841422e-05,
+      "loss": 1.0301,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829553042113e-05,
+      "loss": 1.0893,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829048336658663e-05,
+      "loss": 1.0736,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828542891516387e-05,
+      "loss": 1.0373,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828036706724215e-05,
+      "loss": 1.0564,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827529782320242e-05,
+      "loss": 1.1024,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827022118342607e-05,
+      "loss": 1.0475,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982651371482953e-05,
+      "loss": 1.0765,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982600457181925e-05,
+      "loss": 1.1116,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825494689350106e-05,
+      "loss": 1.0181,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824984067460452e-05,
+      "loss": 1.0832,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824472706188725e-05,
+      "loss": 1.0808,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98239606055734e-05,
+      "loss": 1.043,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823447765653023e-05,
+      "loss": 1.0823,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982293418646618e-05,
+      "loss": 1.1243,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822419868051528e-05,
+      "loss": 1.0749,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821904810447766e-05,
+      "loss": 1.0527,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982138901369366e-05,
+      "loss": 1.0696,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820872477828025e-05,
+      "loss": 1.0312,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982035520288973e-05,
+      "loss": 1.1059,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819837188917703e-05,
+      "loss": 1.043,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981931843595093e-05,
+      "loss": 1.0777,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981879894402845e-05,
+      "loss": 1.0376,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818278713189357e-05,
+      "loss": 1.0456,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98177577434728e-05,
+      "loss": 1.1436,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817236034917984e-05,
+      "loss": 1.059,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816713587564172e-05,
+      "loss": 1.0592,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981619040145068e-05,
+      "loss": 1.0579,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815666476616885e-05,
+      "loss": 1.0539,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815141813102208e-05,
+      "loss": 1.043,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814616410946134e-05,
+      "loss": 1.0671,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981409027018821e-05,
+      "loss": 1.0849,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981356339086802e-05,
+      "loss": 1.0562,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981303577302522e-05,
+      "loss": 4.6214,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812507416699518e-05,
+      "loss": 1.1117,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811978321930673e-05,
+      "loss": 1.091,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98114484887585e-05,
+      "loss": 1.0564,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810917917222875e-05,
+      "loss": 1.0771,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981038660736373e-05,
+      "loss": 1.1006,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809854559221037e-05,
+      "loss": 1.0892,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809321772834848e-05,
+      "loss": 1.0515,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980878824824525e-05,
+      "loss": 1.0622,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808253985492398e-05,
+      "loss": 1.0588,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807718984616495e-05,
+      "loss": 1.1285,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807183245657806e-05,
+      "loss": 1.085,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806646768656647e-05,
+      "loss": 1.0987,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806109553653387e-05,
+      "loss": 1.0497,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805571600688457e-05,
+      "loss": 1.0509,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805032909802343e-05,
+      "loss": 1.0868,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980449348103558e-05,
+      "loss": 1.0491,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980395331442877e-05,
+      "loss": 1.037,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803412410022555e-05,
+      "loss": 1.0417,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802870767857645e-05,
+      "loss": 1.1346,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98023283879748e-05,
+      "loss": 1.0415,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801785270414838e-05,
+      "loss": 1.0537,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801241415218636e-05,
+      "loss": 1.1024,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800696822427115e-05,
+      "loss": 1.0351,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980015149208126e-05,
+      "loss": 1.0491,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979960542422211e-05,
+      "loss": 1.1178,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799058618890762e-05,
+      "loss": 1.0307,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798511076128367e-05,
+      "loss": 1.0913,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979796279597612e-05,
+      "loss": 1.0903,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97974137784753e-05,
+      "loss": 4.9729,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796864023667207e-05,
+      "loss": 1.0534,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796313531593223e-05,
+      "loss": 1.0627,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795762302294767e-05,
+      "loss": 1.0711,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979521033581333e-05,
+      "loss": 1.0731,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794657632190448e-05,
+      "loss": 1.0952,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794104191467713e-05,
+      "loss": 1.0533,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793550013686775e-05,
+      "loss": 1.0438,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979299509888934e-05,
+      "loss": 1.0629,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792439447117163e-05,
+      "loss": 4.4676,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979188305841207e-05,
+      "loss": 1.0378,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791325932815924e-05,
+      "loss": 4.4783,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790768070370653e-05,
+      "loss": 1.084,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979020947111824e-05,
+      "loss": 1.0966,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789650135100726e-05,
+      "loss": 1.1127,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789090062360198e-05,
+      "loss": 1.0345,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978852925293881e-05,
+      "loss": 1.0499,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978796770687876e-05,
+      "loss": 1.1079,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978740542422231e-05,
+      "loss": 1.0388,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978684240501178e-05,
+      "loss": 1.0489,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786278649289528e-05,
+      "loss": 1.0438,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785714157097992e-05,
+      "loss": 1.0592,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785148928479644e-05,
+      "loss": 1.0567,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784582963477027e-05,
+      "loss": 1.0693,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978401626213273e-05,
+      "loss": 1.0413,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783448824489398e-05,
+      "loss": 1.092,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978288065058974e-05,
+      "loss": 1.0889,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782311740476506e-05,
+      "loss": 1.0482,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978174209419252e-05,
+      "loss": 1.0233,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978117171178064e-05,
+      "loss": 1.073,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780600593283796e-05,
+      "loss": 1.0583,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780028738744966e-05,
+      "loss": 1.0529,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779456148207187e-05,
+      "loss": 1.0478,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778882821713546e-05,
+      "loss": 1.1214,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778308759307194e-05,
+      "loss": 1.072,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777733961031326e-05,
+      "loss": 1.0185,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777158426929203e-05,
+      "loss": 1.0714,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776582157044134e-05,
+      "loss": 1.0333,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977600515141949e-05,
+      "loss": 1.1042,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775427410098697e-05,
+      "loss": 1.0744,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977484893312522e-05,
+      "loss": 4.2623,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774269720542607e-05,
+      "loss": 1.1797,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773689772394437e-05,
+      "loss": 1.0508,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773109088724357e-05,
+      "loss": 1.0836,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977252766957607e-05,
+      "loss": 1.0741,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771945514993324e-05,
+      "loss": 1.0455,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771362625019934e-05,
+      "loss": 1.0244,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770778999699764e-05,
+      "loss": 1.0963,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977019463907674e-05,
+      "loss": 1.0541,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976960954319483e-05,
+      "loss": 1.1006,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976902371209807e-05,
+      "loss": 1.0989,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768437145830545e-05,
+      "loss": 1.0525,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97678498444364e-05,
+      "loss": 1.0322,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976726180795983e-05,
+      "loss": 1.0589,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976667303644509e-05,
+      "loss": 1.0542,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766083529936487e-05,
+      "loss": 1.0783,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765493288478385e-05,
+      "loss": 1.0686,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97649023121152e-05,
+      "loss": 1.0491,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976431060089141e-05,
+      "loss": 1.0763,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763718154851542e-05,
+      "loss": 1.0684,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976312497404018e-05,
+      "loss": 0.9918,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976253105850197e-05,
+      "loss": 1.0292,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97619364082816e-05,
+      "loss": 1.0804,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761341023423824e-05,
+      "loss": 1.0813,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976074490397345e-05,
+      "loss": 1.0704,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760148049975332e-05,
+      "loss": 1.0405,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975955046147439e-05,
+      "loss": 1.0287,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97589521385156e-05,
+      "loss": 1.1373,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758353081143986e-05,
+      "loss": 1.1081,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757753289404626e-05,
+      "loss": 1.022,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757152763342663e-05,
+      "loss": 4.3866,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756551503003288e-05,
+      "loss": 1.0908,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755949508431753e-05,
+      "loss": 1.0654,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975534677967335e-05,
+      "loss": 1.0887,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975474331677345e-05,
+      "loss": 1.0749,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975413911977746e-05,
+      "loss": 1.0681,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975353418873085e-05,
+      "loss": 1.0513,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752928523679145e-05,
+      "loss": 1.0431,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752322124667923e-05,
+      "loss": 1.0322,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975171499174282e-05,
+      "loss": 1.0717,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975110712494953e-05,
+      "loss": 1.0531,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750498524333796e-05,
+      "loss": 4.2014,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974988918994141e-05,
+      "loss": 1.0473,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 1.0629,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748668320010184e-05,
+      "loss": 1.0327,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974805678456322e-05,
+      "loss": 1.0436,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747444515523368e-05,
+      "loss": 1.0136,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746831512936695e-05,
+      "loss": 1.0644,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746217776849343e-05,
+      "loss": 1.0554,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974560330730749e-05,
+      "loss": 4.1506,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974498810435739e-05,
+      "loss": 1.0522,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744372168045326e-05,
+      "loss": 1.0439,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743755498417658e-05,
+      "loss": 1.0946,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743138095520798e-05,
+      "loss": 1.1089,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97425199594012e-05,
+      "loss": 1.0198,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741901090105387e-05,
+      "loss": 1.0616,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741281487679932e-05,
+      "loss": 1.0493,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974066115217146e-05,
+      "loss": 1.0938,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974004008362666e-05,
+      "loss": 3.8865,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739418282092267e-05,
+      "loss": 1.0403,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738795747615072e-05,
+      "loss": 1.0808,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973817248024193e-05,
+      "loss": 1.0601,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737548480019744e-05,
+      "loss": 1.0928,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736923746995473e-05,
+      "loss": 1.0826,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973629828121613e-05,
+      "loss": 1.0827,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735672082728785e-05,
+      "loss": 1.0458,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973504515158056e-05,
+      "loss": 1.0434,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973441748781864e-05,
+      "loss": 1.0146,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973378909149026e-05,
+      "loss": 1.0597,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733159962642703e-05,
+      "loss": 4.0141,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973253010132332e-05,
+      "loss": 1.058,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731899507579517e-05,
+      "loss": 1.0539,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731268181458738e-05,
+      "loss": 1.0523,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97306361230085e-05,
+      "loss": 1.0269,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730003332276365e-05,
+      "loss": 1.0393,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729369809309962e-05,
+      "loss": 1.0943,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728735554156956e-05,
+      "loss": 1.0302,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972810056686509e-05,
+      "loss": 1.0224,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727464847482138e-05,
+      "loss": 1.0655,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972682839605595e-05,
+      "loss": 1.111,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726191212634414e-05,
+      "loss": 1.0382,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725553297265492e-05,
+      "loss": 1.0192,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724914649997183e-05,
+      "loss": 1.035,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972427527087755e-05,
+      "loss": 1.0375,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972363515995471e-05,
+      "loss": 1.0229,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972299431727684e-05,
+      "loss": 1.0627,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722352742892155e-05,
+      "loss": 1.0221,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972171043684895e-05,
+      "loss": 1.0003,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721067399195552e-05,
+      "loss": 1.0837,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972042362998036e-05,
+      "loss": 1.0654,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719779129251813e-05,
+      "loss": 1.0655,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971913389705842e-05,
+      "loss": 1.0567,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718487933448735e-05,
+      "loss": 1.0628,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717841238471377e-05,
+      "loss": 1.1069,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717193812175e-05,
+      "loss": 1.0714,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716545654608336e-05,
+      "loss": 1.0669,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971589676582016e-05,
+      "loss": 1.0699,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715247145859306e-05,
+      "loss": 1.0607,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971459679477466e-05,
+      "loss": 1.1148,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713945712615162e-05,
+      "loss": 1.1031,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971329389942981e-05,
+      "loss": 1.0258,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971264135526766e-05,
+      "loss": 3.9597,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711988080177815e-05,
+      "loss": 1.0466,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711334074209443e-05,
+      "loss": 1.0724,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710679337411753e-05,
+      "loss": 1.0812,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710023869834026e-05,
+      "loss": 1.0861,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709367671525587e-05,
+      "loss": 3.9432,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708710742535814e-05,
+      "loss": 1.0504,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708053082914148e-05,
+      "loss": 1.0448,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707394692710084e-05,
+      "loss": 1.0654,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706735571973165e-05,
+      "loss": 1.0794,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706075720752992e-05,
+      "loss": 1.042,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705415139099226e-05,
+      "loss": 1.0464,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970475382706158e-05,
+      "loss": 1.0329,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704091784689817e-05,
+      "loss": 1.0199,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703429012033764e-05,
+      "loss": 1.1257,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702765509143297e-05,
+      "loss": 1.1071,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970210127606834e-05,
+      "loss": 4.0752,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701436312858893e-05,
+      "loss": 1.0632,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970077061956499e-05,
+      "loss": 1.0724,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970010419623673e-05,
+      "loss": 1.0895,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699437042924266e-05,
+      "loss": 1.086,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698769159677803e-05,
+      "loss": 1.0351,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698100546547603e-05,
+      "loss": 1.0352,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969743120358398e-05,
+      "loss": 1.0951,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969676113083731e-05,
+      "loss": 1.0743,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696090328358023e-05,
+      "loss": 1.0615,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969541879619659e-05,
+      "loss": 1.0351,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694746534403558e-05,
+      "loss": 1.0982,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694073543029507e-05,
+      "loss": 1.0342,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693399822125094e-05,
+      "loss": 1.0804,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692725371741013e-05,
+      "loss": 1.0543,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692050191928024e-05,
+      "loss": 1.0882,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691374282736936e-05,
+      "loss": 1.0699,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690697644218612e-05,
+      "loss": 1.0354,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969002027642398e-05,
+      "loss": 1.0674,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968934217940401e-05,
+      "loss": 1.0212,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688663353209733e-05,
+      "loss": 1.0528,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968798379789223e-05,
+      "loss": 1.0053,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687303513502654e-05,
+      "loss": 1.0403,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686622500092185e-05,
+      "loss": 1.0644,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685940757712086e-05,
+      "loss": 1.0841,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685258286413652e-05,
+      "loss": 4.284,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684575086248245e-05,
+      "loss": 1.0351,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968389115726728e-05,
+      "loss": 1.0342,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683206499522232e-05,
+      "loss": 1.0624,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682521113064615e-05,
+      "loss": 1.0137,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681834997946017e-05,
+      "loss": 3.9802,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968114815421806e-05,
+      "loss": 1.0476,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680460581932448e-05,
+      "loss": 1.0229,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679772281140916e-05,
+      "loss": 1.0428,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679083251895262e-05,
+      "loss": 0.9871,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967839349424734e-05,
+      "loss": 1.0613,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677703008249055e-05,
+      "loss": 1.0727,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677011793952376e-05,
+      "loss": 1.0489,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676319851409314e-05,
+      "loss": 1.0601,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675627180671947e-05,
+      "loss": 1.0514,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674933781792403e-05,
+      "loss": 1.0062,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674239654822854e-05,
+      "loss": 1.0253,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967354479981555e-05,
+      "loss": 1.0254,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672849216822772e-05,
+      "loss": 1.0483,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967215290589687e-05,
+      "loss": 1.0334,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671455867090247e-05,
+      "loss": 1.0311,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670758100455356e-05,
+      "loss": 1.0767,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670059606044712e-05,
+      "loss": 1.0346,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669360383910872e-05,
+      "loss": 1.0834,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668660434106467e-05,
+      "loss": 1.0731,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667959756684164e-05,
+      "loss": 1.0793,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667258351696695e-05,
+      "loss": 1.0453,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966655621919684e-05,
+      "loss": 1.0687,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966585335923745e-05,
+      "loss": 4.1276,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665149771871407e-05,
+      "loss": 1.1176,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664445457151665e-05,
+      "loss": 1.0422,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966374041513123e-05,
+      "loss": 1.0265,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663034645863152e-05,
+      "loss": 1.0503,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662328149400555e-05,
+      "loss": 1.0298,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661620925796592e-05,
+      "loss": 3.8993,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96609129751045e-05,
+      "loss": 1.0639,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660204297377546e-05,
+      "loss": 1.1202,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659494892669067e-05,
+      "loss": 1.0612,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965878476103245e-05,
+      "loss": 1.0498,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658073902521134e-05,
+      "loss": 1.0208,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965736231718861e-05,
+      "loss": 1.0981,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965665000508844e-05,
+      "loss": 3.7314,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655936966274215e-05,
+      "loss": 1.0702,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655223200799606e-05,
+      "loss": 1.0713,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654508708718327e-05,
+      "loss": 3.772,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653793490084145e-05,
+      "loss": 1.0547,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653077544950878e-05,
+      "loss": 1.0314,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652360873372415e-05,
+      "loss": 1.0898,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965164347540268e-05,
+      "loss": 1.0521,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650925351095672e-05,
+      "loss": 1.0815,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650206500505423e-05,
+      "loss": 1.0591,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649486923686038e-05,
+      "loss": 1.0551,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964876662069166e-05,
+      "loss": 1.0543,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648045591576506e-05,
+      "loss": 1.0506,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964732383639483e-05,
+      "loss": 1.0848,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964660135520095e-05,
+      "loss": 1.0285,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645878148049237e-05,
+      "loss": 1.0647,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645154214994117e-05,
+      "loss": 1.0611,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644429556090066e-05,
+      "loss": 1.0533,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643704171391625e-05,
+      "loss": 1.071,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964297806095338e-05,
+      "loss": 1.0211,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964225122482997e-05,
+      "loss": 1.0414,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96415236630761e-05,
+      "loss": 1.0611,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964079537574652e-05,
+      "loss": 1.0459,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964006636289604e-05,
+      "loss": 1.032,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639336624579512e-05,
+      "loss": 1.033,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638606160851867e-05,
+      "loss": 1.0508,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963787497176807e-05,
+      "loss": 1.0488,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637143057383138e-05,
+      "loss": 1.0455,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963641041775217e-05,
+      "loss": 1.0313,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635677052930285e-05,
+      "loss": 1.0476,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963494296297268e-05,
+      "loss": 1.0662,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634208147934597e-05,
+      "loss": 1.0615,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633472607871336e-05,
+      "loss": 1.0192,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632736342838247e-05,
+      "loss": 1.0447,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631999352890746e-05,
+      "loss": 1.0357,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631261638084288e-05,
+      "loss": 1.0569,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630523198474386e-05,
+      "loss": 1.0005,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629784034116623e-05,
+      "loss": 1.0567,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629044145066618e-05,
+      "loss": 1.101,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628303531380053e-05,
+      "loss": 1.0156,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627562193112664e-05,
+      "loss": 1.0688,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626820130320235e-05,
+      "loss": 1.0208,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626077343058622e-05,
+      "loss": 1.072,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962533383138371e-05,
+      "loss": 1.0623,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624589595351464e-05,
+      "loss": 1.0842,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623844635017884e-05,
+      "loss": 1.0604,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962309895043903e-05,
+      "loss": 1.045,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962235254167103e-05,
+      "loss": 1.0583,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621605408770046e-05,
+      "loss": 1.0503,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620857551792305e-05,
+      "loss": 3.8478,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962010897079409e-05,
+      "loss": 1.0595,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619359665831734e-05,
+      "loss": 3.6483,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618609636961624e-05,
+      "loss": 1.0901,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961785888424021e-05,
+      "loss": 1.058,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617107407723983e-05,
+      "loss": 1.0799,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96163552074695e-05,
+      "loss": 1.0453,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615602283533364e-05,
+      "loss": 1.0475,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961484863597224e-05,
+      "loss": 1.0577,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614094264842848e-05,
+      "loss": 1.0291,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613339170201947e-05,
+      "loss": 1.0622,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961258335210637e-05,
+      "loss": 1.0019,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611826810612996e-05,
+      "loss": 1.0204,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961106954577876e-05,
+      "loss": 1.0405,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610311557660645e-05,
+      "loss": 1.0563,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609552846315695e-05,
+      "loss": 1.0495,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608793411801007e-05,
+      "loss": 1.0535,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608033254173736e-05,
+      "loss": 1.0893,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607272373491085e-05,
+      "loss": 1.0379,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960651076981032e-05,
+      "loss": 1.03,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605748443188745e-05,
+      "loss": 1.0735,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604985393683734e-05,
+      "loss": 1.0441,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604221621352716e-05,
+      "loss": 1.0492,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960345712625316e-05,
+      "loss": 1.0617,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602691908442602e-05,
+      "loss": 1.062,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601925967978635e-05,
+      "loss": 1.0513,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960115930491889e-05,
+      "loss": 1.008,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600391919321068e-05,
+      "loss": 1.0661,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959962381124292e-05,
+      "loss": 1.0837,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959885498074224e-05,
+      "loss": 1.0864,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95980854278769e-05,
+      "loss": 1.0413,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597315152704808e-05,
+      "loss": 1.0345,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959654415528393e-05,
+      "loss": 1.0341,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595772435672287e-05,
+      "loss": 1.0385,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594999993927958e-05,
+      "loss": 1.0281,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959422683010907e-05,
+      "loss": 1.1006,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593452944273807e-05,
+      "loss": 1.0796,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592678336480416e-05,
+      "loss": 1.0621,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959190300678718e-05,
+      "loss": 1.0046,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591126955252452e-05,
+      "loss": 1.0558,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959035018193463e-05,
+      "loss": 0.989,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958957268689218e-05,
+      "loss": 1.0763,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588794470183603e-05,
+      "loss": 1.0587,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588015531867465e-05,
+      "loss": 1.0409,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958723587200239e-05,
+      "loss": 1.034,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958645549064705e-05,
+      "loss": 1.0599,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958567438786017e-05,
+      "loss": 1.0374,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584892563700532e-05,
+      "loss": 1.0676,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584110018226975e-05,
+      "loss": 1.0564,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583326751498392e-05,
+      "loss": 1.0278,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582542763573723e-05,
+      "loss": 0.9828,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958175805451197e-05,
+      "loss": 1.107,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958097262437218e-05,
+      "loss": 1.0442,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580186473213473e-05,
+      "loss": 1.0713,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579399601095e-05,
+      "loss": 1.0814,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578612008075987e-05,
+      "loss": 1.0981,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577823694215695e-05,
+      "loss": 1.0677,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577034659573452e-05,
+      "loss": 1.0038,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957624490420864e-05,
+      "loss": 1.0564,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957545442818069e-05,
+      "loss": 1.0486,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957466323154909e-05,
+      "loss": 1.0205,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957387131437338e-05,
+      "loss": 1.0033,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957307867671316e-05,
+      "loss": 1.0302,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957228531862808e-05,
+      "loss": 1.0446,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571491240177835e-05,
+      "loss": 1.0271,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570696441422196e-05,
+      "loss": 1.0534,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956990092242097e-05,
+      "loss": 1.0088,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569104683234024e-05,
+      "loss": 1.0731,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956830772392128e-05,
+      "loss": 1.0516,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956751004454271e-05,
+      "loss": 1.0405,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566711645158352e-05,
+      "loss": 1.0446,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956591252582828e-05,
+      "loss": 1.0416,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956511268661264e-05,
+      "loss": 4.2108,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956431212757162e-05,
+      "loss": 1.0588,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563510848765466e-05,
+      "loss": 1.037,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956270885025448e-05,
+      "loss": 1.0394,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561906132099016e-05,
+      "loss": 1.0725,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956110269435948e-05,
+      "loss": 1.0479,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560298537096342e-05,
+      "loss": 1.0643,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559493660370114e-05,
+      "loss": 1.0682,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558688064241367e-05,
+      "loss": 1.0502,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955788174877073e-05,
+      "loss": 1.05,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557074714018876e-05,
+      "loss": 1.065,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556266960046548e-05,
+      "loss": 1.0391,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555458486914523e-05,
+      "loss": 1.0311,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955464929468365e-05,
+      "loss": 1.0476,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553839383414823e-05,
+      "loss": 1.0232,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553028753168995e-05,
+      "loss": 1.0588,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552217404007166e-05,
+      "loss": 4.1956,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551405335990395e-05,
+      "loss": 1.0557,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550592549179795e-05,
+      "loss": 1.0333,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549779043636535e-05,
+      "loss": 1.0488,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548964819421835e-05,
+      "loss": 1.0875,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548149876596966e-05,
+      "loss": 1.0645,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547334215223257e-05,
+      "loss": 1.0419,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546517835362097e-05,
+      "loss": 1.0301,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545700737074913e-05,
+      "loss": 1.0627,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544882920423206e-05,
+      "loss": 1.0172,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544064385468516e-05,
+      "loss": 1.0364,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954324513227244e-05,
+      "loss": 1.0817,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954242516089664e-05,
+      "loss": 1.0117,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954160447140281e-05,
+      "loss": 1.0119,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540783063852723e-05,
+      "loss": 1.0426,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539960938308188e-05,
+      "loss": 1.0519,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539138094831072e-05,
+      "loss": 1.0321,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538314533483313e-05,
+      "loss": 1.0295,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537490254326867e-05,
+      "loss": 1.0553,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953666525742378e-05,
+      "loss": 1.0668,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953583954283613e-05,
+      "loss": 1.0252,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535013110626068e-05,
+      "loss": 1.0051,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953418596085577e-05,
+      "loss": 1.0219,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95333580935875e-05,
+      "loss": 1.0466,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953252950888355e-05,
+      "loss": 1.0326,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531700206806274e-05,
+      "loss": 1.0511,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953087018741809e-05,
+      "loss": 1.0119,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530039450781452e-05,
+      "loss": 1.0008,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952920799695888e-05,
+      "loss": 1.0076,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528375826012946e-05,
+      "loss": 1.0547,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527542938006282e-05,
+      "loss": 1.0327,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526709333001556e-05,
+      "loss": 0.9947,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952587501106151e-05,
+      "loss": 1.0394,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525039972248923e-05,
+      "loss": 1.0754,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524204216626646e-05,
+      "loss": 1.0369,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523367744257564e-05,
+      "loss": 4.2469,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522530555204632e-05,
+      "loss": 4.1254,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521692649530856e-05,
+      "loss": 1.0755,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952085402729928e-05,
+      "loss": 1.0514,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952001468857303e-05,
+      "loss": 1.0494,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951917463341526e-05,
+      "loss": 1.0471,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518333861889193e-05,
+      "loss": 1.0696,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517492374058103e-05,
+      "loss": 1.0767,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516650169985315e-05,
+      "loss": 1.0027,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515807249734203e-05,
+      "loss": 1.0802,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514963613368213e-05,
+      "loss": 1.0057,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514119260950822e-05,
+      "loss": 1.034,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513274192545585e-05,
+      "loss": 1.0196,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512428408216083e-05,
+      "loss": 1.0975,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511581908025973e-05,
+      "loss": 1.0471,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510734692038958e-05,
+      "loss": 1.0406,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509886760318796e-05,
+      "loss": 1.0089,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509038112929302e-05,
+      "loss": 4.7326,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508188749934333e-05,
+      "loss": 1.0631,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507338671397812e-05,
+      "loss": 1.0179,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506487877383715e-05,
+      "loss": 1.0155,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505636367956066e-05,
+      "loss": 1.0601,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504784143178942e-05,
+      "loss": 1.0126,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503931203116484e-05,
+      "loss": 1.0676,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503077547832877e-05,
+      "loss": 1.0384,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502223177392363e-05,
+      "loss": 1.0548,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501368091859238e-05,
+      "loss": 1.0601,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500512291297853e-05,
+      "loss": 0.9977,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949965577577261e-05,
+      "loss": 1.0084,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498798545347965e-05,
+      "loss": 1.0059,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949794060008843e-05,
+      "loss": 1.0861,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497081940058576e-05,
+      "loss": 1.0154,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496222565323015e-05,
+      "loss": 1.0171,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949536247594642e-05,
+      "loss": 1.0271,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494501671993518e-05,
+      "loss": 1.0468,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949364015352909e-05,
+      "loss": 1.0439,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492777920617967e-05,
+      "loss": 1.0725,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491914973325042e-05,
+      "loss": 1.0224,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491051311715253e-05,
+      "loss": 0.9909,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949018693585359e-05,
+      "loss": 0.9547,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489321845805116e-05,
+      "loss": 1.0754,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488456041634916e-05,
+      "loss": 1.0503,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487589523408163e-05,
+      "loss": 1.0591,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486722291190054e-05,
+      "loss": 1.0166,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485854345045857e-05,
+      "loss": 1.0164,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484985685040894e-05,
+      "loss": 1.0521,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484116311240534e-05,
+      "loss": 1.1718,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483246223710196e-05,
+      "loss": 1.0888,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948237542251537e-05,
+      "loss": 1.0528,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481503907721575e-05,
+      "loss": 1.0772,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948063167939441e-05,
+      "loss": 1.0174,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479758737599504e-05,
+      "loss": 1.0103,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478885082402556e-05,
+      "loss": 1.0032,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478010713869315e-05,
+      "loss": 1.0571,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477135632065576e-05,
+      "loss": 1.0431,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94762598370572e-05,
+      "loss": 1.0649,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947538332891009e-05,
+      "loss": 1.0604,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947450610769021e-05,
+      "loss": 1.0636,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947362817346358e-05,
+      "loss": 1.017,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947274952629626e-05,
+      "loss": 1.0018,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471870166254377e-05,
+      "loss": 1.0842,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470990093404112e-05,
+      "loss": 1.0343,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470109307811688e-05,
+      "loss": 1.0494,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469227809543395e-05,
+      "loss": 1.0397,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468345598665565e-05,
+      "loss": 0.997,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946746267524459e-05,
+      "loss": 1.0501,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466579039346917e-05,
+      "loss": 1.1444,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465694691039046e-05,
+      "loss": 1.0827,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464809630387522e-05,
+      "loss": 1.0628,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946392385745896e-05,
+      "loss": 4.5338,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463037372320008e-05,
+      "loss": 1.0791,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946215017503739e-05,
+      "loss": 1.0235,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461262265677866e-05,
+      "loss": 1.0424,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460373644308253e-05,
+      "loss": 1.0199,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945948431099543e-05,
+      "loss": 1.0785,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458594265806322e-05,
+      "loss": 1.0315,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457703508807914e-05,
+      "loss": 1.0683,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456812040067233e-05,
+      "loss": 1.1117,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455919859651366e-05,
+      "loss": 1.0456,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945502696762746e-05,
+      "loss": 1.0067,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454133364062714e-05,
+      "loss": 0.9273,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453239049024365e-05,
+      "loss": 1.0271,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945234402257972e-05,
+      "loss": 4.9409,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451448284796136e-05,
+      "loss": 1.0417,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945055183574102e-05,
+      "loss": 1.0672,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944965467548184e-05,
+      "loss": 1.0135,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94487568040861e-05,
+      "loss": 1.0661,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447858221621382e-05,
+      "loss": 1.0377,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94469589281553e-05,
+      "loss": 1.0687,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446058923755537e-05,
+      "loss": 1.0207,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944515820848982e-05,
+      "loss": 0.9911,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444256782425935e-05,
+      "loss": 1.0156,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944335464563172e-05,
+      "loss": 1.0473,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442451798175055e-05,
+      "loss": 1.0628,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441548240123898e-05,
+      "loss": 1.0349,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944064397154624e-05,
+      "loss": 1.0808,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943973899251013e-05,
+      "loss": 1.044,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 1.0358,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437926903335034e-05,
+      "loss": 1.0361,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437019793332423e-05,
+      "loss": 4.517,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436111973144093e-05,
+      "loss": 1.0354,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435203442838374e-05,
+      "loss": 1.0407,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434294202483634e-05,
+      "loss": 1.0932,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94333842521483e-05,
+      "loss": 1.0386,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432473591900844e-05,
+      "loss": 1.0382,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431562221809806e-05,
+      "loss": 0.9996,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943065014194377e-05,
+      "loss": 1.0454,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942973735237137e-05,
+      "loss": 1.0522,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428823853161304e-05,
+      "loss": 1.1059,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427909644382312e-05,
+      "loss": 1.0503,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426994726103198e-05,
+      "loss": 0.9999,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942607909839281e-05,
+      "loss": 1.0776,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942516276132006e-05,
+      "loss": 1.1067,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424245714953903e-05,
+      "loss": 1.0872,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942332795936335e-05,
+      "loss": 1.0185,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422409494617473e-05,
+      "loss": 1.0311,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421490320785387e-05,
+      "loss": 1.0274,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942057043793626e-05,
+      "loss": 1.0398,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419649846139328e-05,
+      "loss": 4.7973,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418728545463866e-05,
+      "loss": 1.0366,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417806535979205e-05,
+      "loss": 1.1347,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941688381775473e-05,
+      "loss": 1.0277,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415960390859886e-05,
+      "loss": 1.0379,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415036255364162e-05,
+      "loss": 1.0389,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414111411337103e-05,
+      "loss": 1.1034,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413185858848313e-05,
+      "loss": 1.0532,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941225959796744e-05,
+      "loss": 1.0213,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411332628764194e-05,
+      "loss": 1.0039,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941040495130833e-05,
+      "loss": 1.0638,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940947656566967e-05,
+      "loss": 1.0333,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940854747191806e-05,
+      "loss": 1.0393,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407617670123442e-05,
+      "loss": 1.0995,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406687160355778e-05,
+      "loss": 1.0711,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940575594268509e-05,
+      "loss": 1.0048,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404824017181464e-05,
+      "loss": 1.0589,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940389138391503e-05,
+      "loss": 1.0296,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402958042955974e-05,
+      "loss": 1.0671,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402023994374533e-05,
+      "loss": 1.0287,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401089238241e-05,
+      "loss": 1.0296,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940015377462572e-05,
+      "loss": 1.03,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399217603599096e-05,
+      "loss": 1.0658,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398280725231573e-05,
+      "loss": 1.0302,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397343139593658e-05,
+      "loss": 1.0371,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939640484675591e-05,
+      "loss": 1.0787,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395465846788946e-05,
+      "loss": 1.0578,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939452613976342e-05,
+      "loss": 1.0241,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393585725750063e-05,
+      "loss": 1.0382,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392644604819632e-05,
+      "loss": 1.0396,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939170277704296e-05,
+      "loss": 1.059,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939076024249092e-05,
+      "loss": 1.0574,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938981700123445e-05,
+      "loss": 1.0405,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938887305334453e-05,
+      "loss": 1.0535,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387928398892194e-05,
+      "loss": 1.0377,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386983037948536e-05,
+      "loss": 1.0559,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386036970584695e-05,
+      "loss": 1.023,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938509019687187e-05,
+      "loss": 1.0163,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384142716881315e-05,
+      "loss": 1.0724,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383194530684327e-05,
+      "loss": 1.0209,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938224563835226e-05,
+      "loss": 1.0512,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381296039956536e-05,
+      "loss": 1.0119,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380345735568603e-05,
+      "loss": 1.0362,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379394725259984e-05,
+      "loss": 1.0095,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378443009102245e-05,
+      "loss": 1.0564,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377490587167008e-05,
+      "loss": 1.0554,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937653745952595e-05,
+      "loss": 1.0828,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375583626250796e-05,
+      "loss": 1.0741,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374629087413324e-05,
+      "loss": 1.0801,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373673843085377e-05,
+      "loss": 1.0315,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372717893338837e-05,
+      "loss": 0.9832,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371761238245645e-05,
+      "loss": 0.9972,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370803877877793e-05,
+      "loss": 1.0625,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936984581230733e-05,
+      "loss": 0.9965,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936888704160635e-05,
+      "loss": 1.0184,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367927565847013e-05,
+      "loss": 1.0446,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936696738510152e-05,
+      "loss": 5.3805,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366006499442132e-05,
+      "loss": 1.0415,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365044908941154e-05,
+      "loss": 1.0545,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364082613670963e-05,
+      "loss": 1.0853,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936311961370397e-05,
+      "loss": 1.0706,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362155909112642e-05,
+      "loss": 1.0188,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936119149996951e-05,
+      "loss": 1.0376,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360226386347146e-05,
+      "loss": 1.0237,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359260568318182e-05,
+      "loss": 1.0473,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358294045955303e-05,
+      "loss": 1.0294,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357326819331236e-05,
+      "loss": 1.0606,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356358888518783e-05,
+      "loss": 1.0289,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355390253590775e-05,
+      "loss": 1.0366,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354420914620115e-05,
+      "loss": 0.9973,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353450871679746e-05,
+      "loss": 1.0317,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352480124842668e-05,
+      "loss": 0.9647,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935150867418194e-05,
+      "loss": 1.0764,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350536519770664e-05,
+      "loss": 1.0482,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349563661682e-05,
+      "loss": 1.0437,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348590099989165e-05,
+      "loss": 1.0423,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347615834765422e-05,
+      "loss": 1.0828,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934664086608409e-05,
+      "loss": 1.0296,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345665194018535e-05,
+      "loss": 1.0049,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934468881864219e-05,
+      "loss": 4.4325,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343711740028528e-05,
+      "loss": 1.06,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934273395825108e-05,
+      "loss": 0.9758,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341755473383433e-05,
+      "loss": 1.0244,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340776285499218e-05,
+      "loss": 1.0388,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933979639467212e-05,
+      "loss": 0.9831,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338815800975893e-05,
+      "loss": 1.0086,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337834504484325e-05,
+      "loss": 1.0325,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336852505271263e-05,
+      "loss": 1.0224,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933586980341061e-05,
+      "loss": 1.0867,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334886398976318e-05,
+      "loss": 1.0312,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333902292042395e-05,
+      "loss": 1.0445,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93329174826829e-05,
+      "loss": 1.007,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933193197097194e-05,
+      "loss": 1.0481,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330945756983687e-05,
+      "loss": 1.039,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329958840792355e-05,
+      "loss": 6.852,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328971222472216e-05,
+      "loss": 1.0473,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327982902097596e-05,
+      "loss": 1.0275,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326993879742867e-05,
+      "loss": 1.0577,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326004155482457e-05,
+      "loss": 1.0655,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325013729390853e-05,
+      "loss": 0.9983,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324022601542586e-05,
+      "loss": 1.0277,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932303077201225e-05,
+      "loss": 1.0405,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322038240874476e-05,
+      "loss": 1.0418,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932104500820396e-05,
+      "loss": 1.0819,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320051074075455e-05,
+      "loss": 0.9942,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319056438563752e-05,
+      "loss": 1.0312,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318061101743704e-05,
+      "loss": 1.0426,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317065063690223e-05,
+      "loss": 1.1092,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316068324478253e-05,
+      "loss": 1.0602,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315070884182815e-05,
+      "loss": 1.0714,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314072742878963e-05,
+      "loss": 1.0342,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313073900641818e-05,
+      "loss": 1.0182,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931207435754655e-05,
+      "loss": 1.0234,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311074113668377e-05,
+      "loss": 1.0443,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931007316908257e-05,
+      "loss": 0.9896,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309071523864455e-05,
+      "loss": 1.066,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930806917808942e-05,
+      "loss": 1.0205,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307066131832886e-05,
+      "loss": 4.2549,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306062385170344e-05,
+      "loss": 1.0209,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930505793817733e-05,
+      "loss": 1.0311,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304052790929434e-05,
+      "loss": 1.08,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303046943502296e-05,
+      "loss": 1.033,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302040395971615e-05,
+      "loss": 4.6341,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301033148413136e-05,
+      "loss": 1.0299,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300025200902666e-05,
+      "loss": 1.053,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299016553516048e-05,
+      "loss": 1.0224,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298007206329196e-05,
+      "loss": 1.0677,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929699715941807e-05,
+      "loss": 1.0519,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295986412858673e-05,
+      "loss": 1.0256,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294974966727074e-05,
+      "loss": 1.0737,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929396282109939e-05,
+      "loss": 1.0797,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929294997605179e-05,
+      "loss": 1.0196,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291936431660498e-05,
+      "loss": 1.0095,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290922188001784e-05,
+      "loss": 1.0494,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928990724515198e-05,
+      "loss": 1.0946,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288891603187462e-05,
+      "loss": 0.9526,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287875262184663e-05,
+      "loss": 1.0509,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286858222220068e-05,
+      "loss": 1.0056,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928584048337022e-05,
+      "loss": 1.015,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92848220457117e-05,
+      "loss": 1.0337,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928380290932116e-05,
+      "loss": 1.008,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282783074275287e-05,
+      "loss": 5.3995,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281762540650835e-05,
+      "loss": 1.0426,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92807413085246e-05,
+      "loss": 1.0484,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279719377973443e-05,
+      "loss": 1.0302,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927869674907426e-05,
+      "loss": 1.0356,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277673421904014e-05,
+      "loss": 5.2084,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276649396539716e-05,
+      "loss": 1.0214,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927562467305843e-05,
+      "loss": 1.0407,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927459925153727e-05,
+      "loss": 1.0957,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273573132053403e-05,
+      "loss": 1.0934,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272546314684054e-05,
+      "loss": 1.0394,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271518799506494e-05,
+      "loss": 0.9944,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927049058659805e-05,
+      "loss": 1.0585,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269461676036097e-05,
+      "loss": 1.0585,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926843206789807e-05,
+      "loss": 1.0196,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926740176226145e-05,
+      "loss": 1.0786,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266370759203776e-05,
+      "loss": 1.0655,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265339058802633e-05,
+      "loss": 1.0399,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264306661135667e-05,
+      "loss": 1.061,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926327356628057e-05,
+      "loss": 1.0479,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926223977431508e-05,
+      "loss": 1.0473,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261205285317002e-05,
+      "loss": 1.043,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260170099364185e-05,
+      "loss": 1.0009,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259134216534537e-05,
+      "loss": 1.0326,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258097636906007e-05,
+      "loss": 1.0637,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925706036055661e-05,
+      "loss": 1.0348,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256022387564394e-05,
+      "loss": 1.0061,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254983718007486e-05,
+      "loss": 1.0591,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253944351964046e-05,
+      "loss": 1.0471,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252904289512288e-05,
+      "loss": 1.052,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925186353073049e-05,
+      "loss": 0.9689,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925082207569697e-05,
+      "loss": 1.0157,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92497799244901e-05,
+      "loss": 1.0395,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248737077188316e-05,
+      "loss": 1.0486,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247693533870087e-05,
+      "loss": 10.7618,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246649294613957e-05,
+      "loss": 1.0154,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245604359498498e-05,
+      "loss": 1.087,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244558728602353e-05,
+      "loss": 1.0594,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243512402004218e-05,
+      "loss": 1.0285,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242465379782823e-05,
+      "loss": 1.0142,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241417662016965e-05,
+      "loss": 1.0185,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240369248785494e-05,
+      "loss": 0.9944,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923932014016731e-05,
+      "loss": 1.0602,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923827033624136e-05,
+      "loss": 1.0592,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237219837086647e-05,
+      "loss": 1.0824,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923616864278223e-05,
+      "loss": 1.0656,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235116753407213e-05,
+      "loss": 1.0422,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234064169040757e-05,
+      "loss": 1.0661,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233010889762078e-05,
+      "loss": 1.052,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231956915650434e-05,
+      "loss": 1.0293,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923090224678515e-05,
+      "loss": 1.0477,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229846883245594e-05,
+      "loss": 1.1565,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922879082511118e-05,
+      "loss": 1.0966,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227734072461394e-05,
+      "loss": 7.2301,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226676625375753e-05,
+      "loss": 1.0602,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225618483933836e-05,
+      "loss": 1.0811,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922455964821528e-05,
+      "loss": 1.0362,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223500118299762e-05,
+      "loss": 1.0662,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922243989426702e-05,
+      "loss": 1.0289,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922137897619684e-05,
+      "loss": 1.0343,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220317364169062e-05,
+      "loss": 1.0691,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219255058263582e-05,
+      "loss": 1.1091,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218192058560338e-05,
+      "loss": 1.1071,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217128365139328e-05,
+      "loss": 1.1001,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216063978080602e-05,
+      "loss": 1.0247,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921499889746426e-05,
+      "loss": 1.0731,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921393312337046e-05,
+      "loss": 1.0368,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212866655879397e-05,
+      "loss": 1.0911,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211799495071335e-05,
+      "loss": 1.0313,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210731641026583e-05,
+      "loss": 1.0497,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209663093825503e-05,
+      "loss": 1.0952,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208593853548506e-05,
+      "loss": 1.0583,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207523920276057e-05,
+      "loss": 1.0689,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206453294088682e-05,
+      "loss": 1.0899,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205381975066943e-05,
+      "loss": 1.0303,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204309963291465e-05,
+      "loss": 1.0941,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203237258842925e-05,
+      "loss": 1.0528,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202163861802047e-05,
+      "loss": 1.0614,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920108977224961e-05,
+      "loss": 0.998,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200014990266446e-05,
+      "loss": 1.0283,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919893951593344e-05,
+      "loss": 1.0634,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197863349331522e-05,
+      "loss": 1.0607,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196786490541683e-05,
+      "loss": 1.0457,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919570893964496e-05,
+      "loss": 1.0047,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919463069672245e-05,
+      "loss": 1.0387,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193551761855288e-05,
+      "loss": 1.0624,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192472135124672e-05,
+      "loss": 1.0297,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191391816611857e-05,
+      "loss": 0.9719,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190310806398133e-05,
+      "loss": 1.0407,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189229104564857e-05,
+      "loss": 1.0266,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188146711193433e-05,
+      "loss": 1.0079,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187063626365313e-05,
+      "loss": 1.0549,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185979850162007e-05,
+      "loss": 1.0398,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184895382665077e-05,
+      "loss": 5.72,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183810223956133e-05,
+      "loss": 1.0165,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918272437411684e-05,
+      "loss": 1.0042,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918163783322891e-05,
+      "loss": 0.9981,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180550601374115e-05,
+      "loss": 1.0008,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179462678634274e-05,
+      "loss": 1.0202,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917837406509126e-05,
+      "loss": 1.0174,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177284760826995e-05,
+      "loss": 1.0409,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917619476592346e-05,
+      "loss": 1.0184,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917510408046267e-05,
+      "loss": 1.0284,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917401270452672e-05,
+      "loss": 1.067,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172920638197735e-05,
+      "loss": 0.9874,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171827881557897e-05,
+      "loss": 1.0565,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170734434689444e-05,
+      "loss": 0.9877,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169640297674662e-05,
+      "loss": 1.1135,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168545470595897e-05,
+      "loss": 1.0374,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167449953535532e-05,
+      "loss": 1.0072,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166353746576013e-05,
+      "loss": 1.0443,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916525684979984e-05,
+      "loss": 1.0807,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164159263289553e-05,
+      "loss": 1.0258,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163060987127755e-05,
+      "loss": 1.0114,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91619620213971e-05,
+      "loss": 1.064,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160862366180282e-05,
+      "loss": 1.0672,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915976202156007e-05,
+      "loss": 1.0366,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158660987619258e-05,
+      "loss": 1.0037,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915755926444071e-05,
+      "loss": 0.988,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156456852107337e-05,
+      "loss": 1.0631,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91553537507021e-05,
+      "loss": 0.9713,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154249960308013e-05,
+      "loss": 1.0239,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915314548100814e-05,
+      "loss": 0.9981,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152040312885604e-05,
+      "loss": 1.0556,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915093445602357e-05,
+      "loss": 1.0334,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149827910505264e-05,
+      "loss": 1.0372,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148720676413955e-05,
+      "loss": 1.049,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147612753832972e-05,
+      "loss": 1.0432,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914650414284569e-05,
+      "loss": 1.0793,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914539484353554e-05,
+      "loss": 1.0018,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144284855985996e-05,
+      "loss": 1.0436,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91431741802806e-05,
+      "loss": 1.0595,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142062816502932e-05,
+      "loss": 1.0608,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140950764736624e-05,
+      "loss": 1.0573,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913983802506537e-05,
+      "loss": 1.0763,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138724597572904e-05,
+      "loss": 1.0234,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137610482343022e-05,
+      "loss": 1.0361,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136495679459566e-05,
+      "loss": 1.0583,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913538018900643e-05,
+      "loss": 1.0283,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913426401106756e-05,
+      "loss": 1.0241,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913314714572695e-05,
+      "loss": 1.0334,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132029593068663e-05,
+      "loss": 1.046,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130911353176788e-05,
+      "loss": 1.027,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129792426135485e-05,
+      "loss": 1.056,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128672812028957e-05,
+      "loss": 1.0581,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912755251094146e-05,
+      "loss": 1.0646,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126431522957307e-05,
+      "loss": 1.0003,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125309848160852e-05,
+      "loss": 1.0216,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124187486636512e-05,
+      "loss": 1.0764,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123064438468748e-05,
+      "loss": 0.9892,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121940703742074e-05,
+      "loss": 1.0106,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120816282541062e-05,
+      "loss": 1.0701,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911969117495033e-05,
+      "loss": 0.9491,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911856538105454e-05,
+      "loss": 1.0478,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117438900938424e-05,
+      "loss": 1.0286,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911631173468675e-05,
+      "loss": 1.0227,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115183882384348e-05,
+      "loss": 0.9909,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114055344116087e-05,
+      "loss": 1.0479,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112926119966905e-05,
+      "loss": 1.0379,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911179621002177e-05,
+      "loss": 1.0316,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911066561436573e-05,
+      "loss": 1.0753,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910953433308385e-05,
+      "loss": 1.0019,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108402366261282e-05,
+      "loss": 1.019,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91072697139832e-05,
+      "loss": 1.0121,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910613637633485e-05,
+      "loss": 1.0493,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105002353401515e-05,
+      "loss": 1.0405,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103867645268542e-05,
+      "loss": 1.0565,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910273225202132e-05,
+      "loss": 1.0573,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101596173745295e-05,
+      "loss": 6.672,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100459410525965e-05,
+      "loss": 1.013,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099321962448873e-05,
+      "loss": 1.0754,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909818382959962e-05,
+      "loss": 1.0158,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097045012063857e-05,
+      "loss": 1.0651,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095905509927287e-05,
+      "loss": 1.0464,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909476532327566e-05,
+      "loss": 1.0652,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093624452194785e-05,
+      "loss": 0.9918,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092482896770517e-05,
+      "loss": 0.9958,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091340657088764e-05,
+      "loss": 1.0613,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909019773323548e-05,
+      "loss": 1.0585,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089054125296692e-05,
+      "loss": 1.018,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087909833358446e-05,
+      "loss": 1.0754,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086764857506862e-05,
+      "loss": 1.013,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085619197828107e-05,
+      "loss": 1.036,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90844728544084e-05,
+      "loss": 1.0147,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083325827334e-05,
+      "loss": 0.996,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082178116691238e-05,
+      "loss": 1.0547,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908102972256648e-05,
+      "loss": 1.0907,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079880645046146e-05,
+      "loss": 1.0342,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078730884216714e-05,
+      "loss": 1.0018,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907758044016471e-05,
+      "loss": 1.0058,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076429312976707e-05,
+      "loss": 1.0353,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075277502739337e-05,
+      "loss": 1.044,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907412500953928e-05,
+      "loss": 8.9886,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907297183346327e-05,
+      "loss": 1.0488,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907181797459808e-05,
+      "loss": 0.9887,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070663433030553e-05,
+      "loss": 1.0321,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906950820884757e-05,
+      "loss": 1.0119,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906835230213607e-05,
+      "loss": 1.0629,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067195712983043e-05,
+      "loss": 1.0436,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906603844147552e-05,
+      "loss": 1.0312,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90648804877006e-05,
+      "loss": 1.0246,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906372185174542e-05,
+      "loss": 1.0218,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906256253369718e-05,
+      "loss": 1.0196,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906140253364312e-05,
+      "loss": 1.0276,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060241851670534e-05,
+      "loss": 1.0342,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059080487866773e-05,
+      "loss": 1.0277,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057918442319238e-05,
+      "loss": 1.0546,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056755715115372e-05,
+      "loss": 1.0355,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055592306342685e-05,
+      "loss": 1.0185,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905442821608872e-05,
+      "loss": 1.0105,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053263444441096e-05,
+      "loss": 1.0286,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905209799148745e-05,
+      "loss": 1.012,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90509318573155e-05,
+      "loss": 7.3578,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049765042013e-05,
+      "loss": 1.0376,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904859754566776e-05,
+      "loss": 1.0565,
+      "step": 1958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904742936836764e-05,
+      "loss": 1.2417,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904626051020055e-05,
+      "loss": 1.0259,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045090971254457e-05,
+      "loss": 1.0767,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904392075161737e-05,
+      "loss": 1.0543,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904274985137736e-05,
+      "loss": 1.0143,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041578270622535e-05,
+      "loss": 0.9986,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904040600944107e-05,
+      "loss": 1.0092,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039233067921183e-05,
+      "loss": 1.0511,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038059446151143e-05,
+      "loss": 1.0305,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903688514421927e-05,
+      "loss": 1.0119,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035710162213937e-05,
+      "loss": 1.0057,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034534500223566e-05,
+      "loss": 1.0117,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033358158336636e-05,
+      "loss": 1.0223,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032181136641673e-05,
+      "loss": 1.0283,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903100343522725e-05,
+      "loss": 1.0796,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029825054181998e-05,
+      "loss": 1.0164,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028645993594593e-05,
+      "loss": 1.0456,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027466253553767e-05,
+      "loss": 1.0095,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026285834148302e-05,
+      "loss": 1.0201,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025104735467035e-05,
+      "loss": 1.0039,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023922957598847e-05,
+      "loss": 1.0365,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022740500632673e-05,
+      "loss": 1.0279,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021557364657496e-05,
+      "loss": 1.0159,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020373549762358e-05,
+      "loss": 1.0249,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901918905603634e-05,
+      "loss": 1.0333,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018003883568593e-05,
+      "loss": 1.0328,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016818032448297e-05,
+      "loss": 1.1004,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90156315027647e-05,
+      "loss": 1.0354,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901444429460709e-05,
+      "loss": 0.997,
+      "step": 1987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013256408064817e-05,
+      "loss": 1.0104,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012067843227272e-05,
+      "loss": 1.0658,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010878600183895e-05,
+      "loss": 1.0257,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.0026,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008498079837708e-05,
+      "loss": 0.9911,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007306802714042e-05,
+      "loss": 1.0759,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900611484774284e-05,
+      "loss": 0.9885,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004922215013806e-05,
+      "loss": 1.0142,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003728904616696e-05,
+      "loss": 1.0112,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002534916641308e-05,
+      "loss": 1.0062,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001340251177493e-05,
+      "loss": 1.0794,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000144908315163e-05,
+      "loss": 1.0144,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998948888144267e-05,
+      "loss": 1.0924,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899775219075482e-05,
+      "loss": 0.9573,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996554816236874e-05,
+      "loss": 1.0184,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899535676468054e-05,
+      "loss": 1.0313,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994158036175974e-05,
+      "loss": 1.0566,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899295863081339e-05,
+      "loss": 1.0227,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899175854868305e-05,
+      "loss": 1.0468,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990557789875265e-05,
+      "loss": 1.0377,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89893563544804e-05,
+      "loss": 1.0385,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988154242588867e-05,
+      "loss": 1.0077,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986951454291134e-05,
+      "loss": 1.0205,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985747989677716e-05,
+      "loss": 1.0621,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984543848839183e-05,
+      "loss": 1.0516,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983339031866144e-05,
+      "loss": 1.0192,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982133538849275e-05,
+      "loss": 1.0654,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980927369879295e-05,
+      "loss": 1.0839,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979720525046978e-05,
+      "loss": 1.0253,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897851300444314e-05,
+      "loss": 1.0293,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897730480815865e-05,
+      "loss": 1.0624,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897609593628444e-05,
+      "loss": 1.0674,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974886388911483e-05,
+      "loss": 1.0181,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973676166130793e-05,
+      "loss": 1.0128,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897246526803346e-05,
+      "loss": 1.0032,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971253694710606e-05,
+      "loss": 1.0515,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89700414462534e-05,
+      "loss": 1.0118,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896882852275308e-05,
+      "loss": 0.9996,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967614924300922e-05,
+      "loss": 0.9968,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896640065098825e-05,
+      "loss": 1.0118,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965185702906458e-05,
+      "loss": 1.0233,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963970080146964e-05,
+      "loss": 1.0121,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962753782801255e-05,
+      "loss": 1.0339,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961536810960867e-05,
+      "loss": 1.0099,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960319164717375e-05,
+      "loss": 1.0345,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959100844162423e-05,
+      "loss": 1.0743,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895788184938769e-05,
+      "loss": 1.0011,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956662180484913e-05,
+      "loss": 0.9773,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955441837545884e-05,
+      "loss": 5.1108,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954220820662427e-05,
+      "loss": 1.0138,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952999129926445e-05,
+      "loss": 1.0275,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895177676542987e-05,
+      "loss": 0.9986,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950553727264687e-05,
+      "loss": 1.0841,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949330015522943e-05,
+      "loss": 1.0655,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948105630296728e-05,
+      "loss": 1.0357,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894688057167818e-05,
+      "loss": 1.0379,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945654839759494e-05,
+      "loss": 1.0213,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894442843463291e-05,
+      "loss": 1.0514,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943201356390727e-05,
+      "loss": 1.0499,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894197360512528e-05,
+      "loss": 1.0765,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940745180928975e-05,
+      "loss": 1.0479,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939516083894247e-05,
+      "loss": 1.0031,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89382863141136e-05,
+      "loss": 1.0182,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937055871679575e-05,
+      "loss": 1.0576,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935824756684772e-05,
+      "loss": 1.0039,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893459296922184e-05,
+      "loss": 1.0248,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933360509383478e-05,
+      "loss": 1.0319,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893212737726243e-05,
+      "loss": 1.02,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89308935729515e-05,
+      "loss": 1.0608,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929659096543535e-05,
+      "loss": 0.995,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892842394813144e-05,
+      "loss": 1.0164,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927188127808166e-05,
+      "loss": 1.0397,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892595163566671e-05,
+      "loss": 1.067,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924714471800133e-05,
+      "loss": 1.051,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923476636301528e-05,
+      "loss": 1.0683,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892223812926406e-05,
+      "loss": 1.0618,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920998950780926e-05,
+      "loss": 0.999,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919759100945378e-05,
+      "loss": 1.0672,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891851857985073e-05,
+      "loss": 1.0075,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891727738759033e-05,
+      "loss": 1.0001,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916035524257587e-05,
+      "loss": 1.0408,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891479298994596e-05,
+      "loss": 1.0565,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913549784748953e-05,
+      "loss": 1.0151,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912305908760127e-05,
+      "loss": 1.0066,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891106136207309e-05,
+      "loss": 1.0385,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909816144781495e-05,
+      "loss": 1.0572,
+      "step": 2073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908570256979058e-05,
+      "loss": 1.0117,
+      "step": 2074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907323698759536e-05,
+      "loss": 1.0329,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906076470216737e-05,
+      "loss": 1.0211,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904828571444526e-05,
+      "loss": 1.0126,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903580002536808e-05,
+      "loss": 1.0015,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902330763587553e-05,
+      "loss": 1.0345,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901080854690765e-05,
+      "loss": 1.0472,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899830275940512e-05,
+      "loss": 0.9994,
+      "step": 2081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88985790274309e-05,
+      "loss": 1.0282,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88973271092561e-05,
+      "loss": 1.0106,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896074521510317e-05,
+      "loss": 1.0258,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894821264287823e-05,
+      "loss": 1.0199,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893567337682926e-05,
+      "loss": 1.0259,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892312741789993e-05,
+      "loss": 1.0166,
+      "step": 2087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891057476703443e-05,
+      "loss": 0.9876,
+      "step": 2088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889801542517734e-05,
+      "loss": 0.9958,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888544939327386e-05,
+      "loss": 1.0216,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887287667226964e-05,
+      "loss": 1.0146,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888602972631109e-05,
+      "loss": 1.0185,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884771116674418e-05,
+      "loss": 0.9698,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883511838411673e-05,
+      "loss": 1.0651,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882251891617624e-05,
+      "loss": 1.0682,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880991276387088e-05,
+      "loss": 1.0667,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879729992814928e-05,
+      "loss": 1.0542,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878468040996067e-05,
+      "loss": 1.051,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887720542102547e-05,
+      "loss": 1.0013,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887594213299816e-05,
+      "loss": 1.0231,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874678177009205e-05,
+      "loss": 1.019,
+      "step": 2101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873413553153725e-05,
+      "loss": 0.9979,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872148261526887e-05,
+      "loss": 1.0248,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887088230222391e-05,
+      "loss": 1.0256,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869615675340068e-05,
+      "loss": 1.0199,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868348380970678e-05,
+      "loss": 1.0653,
+      "step": 2106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867080419211113e-05,
+      "loss": 1.0365,
+      "step": 2107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865811790156794e-05,
+      "loss": 1.0201,
+      "step": 2108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864542493903187e-05,
+      "loss": 1.0213,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863272530545825e-05,
+      "loss": 1.0373,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862001900180266e-05,
+      "loss": 1.0467,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886073060290214e-05,
+      "loss": 1.0233,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859458638807115e-05,
+      "loss": 1.0732,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858186007990916e-05,
+      "loss": 1.0195,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885691271054931e-05,
+      "loss": 1.013,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855638746578123e-05,
+      "loss": 1.0049,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885436411617323e-05,
+      "loss": 1.0408,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853088819430554e-05,
+      "loss": 1.0399,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851812856446062e-05,
+      "loss": 0.9933,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885053622731578e-05,
+      "loss": 1.0068,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884925893213578e-05,
+      "loss": 0.9895,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847980971002187e-05,
+      "loss": 1.0245,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846702344011175e-05,
+      "loss": 1.0748,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845423051258965e-05,
+      "loss": 1.0243,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884414309284183e-05,
+      "loss": 1.0185,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88428624688561e-05,
+      "loss": 1.0228,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884158117939814e-05,
+      "loss": 1.0274,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840299224564376e-05,
+      "loss": 1.0463,
+      "step": 2128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839016604451287e-05,
+      "loss": 1.0568,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837733319155392e-05,
+      "loss": 0.9557,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883644936877327e-05,
+      "loss": 0.9961,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883516475340154e-05,
+      "loss": 0.9945,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883387947313688e-05,
+      "loss": 1.0338,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832593528076007e-05,
+      "loss": 1.0504,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831306918315703e-05,
+      "loss": 0.9752,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883001964395279e-05,
+      "loss": 1.0108,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828731705084143e-05,
+      "loss": 1.0397,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8827443101806682e-05,
+      "loss": 1.0187,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826153834217384e-05,
+      "loss": 1.0232,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824863902413275e-05,
+      "loss": 0.9965,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823573306491425e-05,
+      "loss": 1.0353,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822282046548962e-05,
+      "loss": 1.0323,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820990122683057e-05,
+      "loss": 1.014,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819697534990943e-05,
+      "loss": 0.9834,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881840428356988e-05,
+      "loss": 1.0031,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817110368517202e-05,
+      "loss": 1.0486,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815815789930277e-05,
+      "loss": 1.0406,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814520547906534e-05,
+      "loss": 0.9807,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813224642543443e-05,
+      "loss": 0.9967,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881192807393853e-05,
+      "loss": 0.9786,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881063084218937e-05,
+      "loss": 1.0297,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809332947393586e-05,
+      "loss": 0.9758,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808034389648847e-05,
+      "loss": 0.9973,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880673516905288e-05,
+      "loss": 1.0576,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805435285703458e-05,
+      "loss": 1.0694,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804134739698404e-05,
+      "loss": 1.0,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880283353113559e-05,
+      "loss": 1.0115,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880153166011294e-05,
+      "loss": 1.011,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800229126728427e-05,
+      "loss": 1.0026,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798925931080068e-05,
+      "loss": 1.0068,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797622073265946e-05,
+      "loss": 1.0207,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796317553384176e-05,
+      "loss": 1.0304,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879501237153293e-05,
+      "loss": 1.0714,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793706527810433e-05,
+      "loss": 1.0085,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792400022314954e-05,
+      "loss": 0.9972,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791092855144814e-05,
+      "loss": 1.057,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789785026398388e-05,
+      "loss": 0.9997,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878847653617409e-05,
+      "loss": 0.9377,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787167384570402e-05,
+      "loss": 0.9995,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785857571685834e-05,
+      "loss": 0.9994,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784547097618963e-05,
+      "loss": 1.0644,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783235962468406e-05,
+      "loss": 1.033,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878192416633283e-05,
+      "loss": 1.0128,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878061170931096e-05,
+      "loss": 1.0234,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779298591501565e-05,
+      "loss": 1.0268,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877798481300346e-05,
+      "loss": 0.978,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877667037391552e-05,
+      "loss": 1.0853,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775355274336655e-05,
+      "loss": 4.8319,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877403951436584e-05,
+      "loss": 1.0089,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877272309410209e-05,
+      "loss": 0.9798,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771406013644475e-05,
+      "loss": 0.9882,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770088273092108e-05,
+      "loss": 1.0142,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876876987254416e-05,
+      "loss": 1.0228,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767450812099848e-05,
+      "loss": 1.0639,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766131091858434e-05,
+      "loss": 0.9802,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764810711919236e-05,
+      "loss": 1.0562,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763489672381622e-05,
+      "loss": 4.7808,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762167973345005e-05,
+      "loss": 1.0109,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876084561490885e-05,
+      "loss": 1.0164,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759522597172674e-05,
+      "loss": 1.0344,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758198920236036e-05,
+      "loss": 0.9973,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875687458419855e-05,
+      "loss": 1.05,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755549589159886e-05,
+      "loss": 1.0219,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875422393521975e-05,
+      "loss": 0.9669,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875289762247791e-05,
+      "loss": 1.0584,
+      "step": 2195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875157065103417e-05,
+      "loss": 0.9827,
+      "step": 2196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750243020988402e-05,
+      "loss": 1.0064,
+      "step": 2197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874891473244051e-05,
+      "loss": 1.0493,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747585785490456e-05,
+      "loss": 1.0104,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746256180238254e-05,
+      "loss": 1.0182,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744925916783958e-05,
+      "loss": 1.0715,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743594995227682e-05,
+      "loss": 1.0001,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874226341566958e-05,
+      "loss": 0.9963,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740931178209864e-05,
+      "loss": 0.9872,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739598282948792e-05,
+      "loss": 0.993,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738264729986673e-05,
+      "loss": 1.007,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736930519423858e-05,
+      "loss": 1.0342,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873559565136076e-05,
+      "loss": 1.017,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734260125897832e-05,
+      "loss": 1.0017,
+      "step": 2209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873292394313558e-05,
+      "loss": 0.9992,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873158710317456e-05,
+      "loss": 1.024,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730249606115367e-05,
+      "loss": 0.989,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728911452058668e-05,
+      "loss": 0.9851,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727572641105157e-05,
+      "loss": 1.0559,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726233173355593e-05,
+      "loss": 1.0846,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724893048910778e-05,
+      "loss": 0.9786,
+      "step": 2216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723552267871557e-05,
+      "loss": 1.0507,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872221083033883e-05,
+      "loss": 0.9901,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720868736413556e-05,
+      "loss": 0.9985,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719525986196735e-05,
+      "loss": 1.0236,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718182579789404e-05,
+      "loss": 4.7383,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871683851729267e-05,
+      "loss": 1.0563,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715493798807687e-05,
+      "loss": 0.9965,
+      "step": 2223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871414842443564e-05,
+      "loss": 1.0025,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712802394277777e-05,
+      "loss": 1.0216,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87114557084354e-05,
+      "loss": 0.994,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710108367009853e-05,
+      "loss": 0.9854,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870876037010253e-05,
+      "loss": 1.0097,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707411717814873e-05,
+      "loss": 1.0233,
+      "step": 2229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706062410248376e-05,
+      "loss": 1.0398,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870471244750458e-05,
+      "loss": 1.0101,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870336182968508e-05,
+      "loss": 1.0636,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702010556891514e-05,
+      "loss": 1.016,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700658629225578e-05,
+      "loss": 1.0206,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699306046789004e-05,
+      "loss": 1.0496,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697952809683587e-05,
+      "loss": 0.9606,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696598918011165e-05,
+      "loss": 0.9876,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869524437187362e-05,
+      "loss": 1.0381,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693889171372896e-05,
+      "loss": 1.0466,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692533316610976e-05,
+      "loss": 1.0384,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869117680768989e-05,
+      "loss": 0.9819,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689819644711732e-05,
+      "loss": 0.9638,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688461827778627e-05,
+      "loss": 1.0074,
+      "step": 2243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687103356992767e-05,
+      "loss": 1.0504,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685744232456375e-05,
+      "loss": 0.9972,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684384454271738e-05,
+      "loss": 0.9905,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683024022541186e-05,
+      "loss": 1.0057,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681662937367094e-05,
+      "loss": 0.978,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680301198851895e-05,
+      "loss": 1.0192,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678938807098067e-05,
+      "loss": 1.0323,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867757576220814e-05,
+      "loss": 0.973,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676212064284683e-05,
+      "loss": 1.0421,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674847713430327e-05,
+      "loss": 1.0015,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673482709747742e-05,
+      "loss": 4.2491,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672117053339657e-05,
+      "loss": 1.033,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670750744308844e-05,
+      "loss": 4.1634,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669383782758123e-05,
+      "loss": 1.0049,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668016168790365e-05,
+      "loss": 1.0003,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666647902508493e-05,
+      "loss": 0.9929,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866527898401547e-05,
+      "loss": 1.0035,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663909413414326e-05,
+      "loss": 1.0611,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662539190808117e-05,
+      "loss": 1.0247,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866116831629996e-05,
+      "loss": 1.0105,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659796789993026e-05,
+      "loss": 1.0086,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865842461199053e-05,
+      "loss": 1.0402,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865705178239573e-05,
+      "loss": 0.9804,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655678301311945e-05,
+      "loss": 1.0223,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654304168842533e-05,
+      "loss": 0.9918,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652929385090907e-05,
+      "loss": 1.0148,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651553950160527e-05,
+      "loss": 1.0213,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86501778641549e-05,
+      "loss": 1.0482,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864880112717758e-05,
+      "loss": 1.0146,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647423739332176e-05,
+      "loss": 1.0301,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646045700722348e-05,
+      "loss": 1.0082,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86446670114518e-05,
+      "loss": 1.0245,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643287671624278e-05,
+      "loss": 1.0076,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641907681343597e-05,
+      "loss": 1.0053,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86405270407136e-05,
+      "loss": 1.0417,
+      "step": 2278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863914574983819e-05,
+      "loss": 1.0022,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863776380882131e-05,
+      "loss": 0.9784,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863638121776697e-05,
+      "loss": 1.0238,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863499797677921e-05,
+      "loss": 1.0322,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633614085962127e-05,
+      "loss": 1.0633,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863222954541987e-05,
+      "loss": 1.064,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630844355256624e-05,
+      "loss": 1.0697,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629458515576643e-05,
+      "loss": 1.0416,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628072026484215e-05,
+      "loss": 1.0346,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626684888083676e-05,
+      "loss": 1.0044,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625297100479417e-05,
+      "loss": 0.987,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862390866377588e-05,
+      "loss": 1.011,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622519578077552e-05,
+      "loss": 1.0223,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621129843488964e-05,
+      "loss": 1.0041,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619739460114705e-05,
+      "loss": 1.028,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618348428059412e-05,
+      "loss": 1.0163,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861695674742776e-05,
+      "loss": 1.0569,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615564418324484e-05,
+      "loss": 0.9822,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614171440854363e-05,
+      "loss": 1.0656,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861277781512223e-05,
+      "loss": 1.0277,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861138354123296e-05,
+      "loss": 4.1849,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609988619291473e-05,
+      "loss": 1.0303,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608593049402754e-05,
+      "loss": 0.9652,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860719683167182e-05,
+      "loss": 1.0281,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605799966203752e-05,
+      "loss": 1.025,
+      "step": 2303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604402453103666e-05,
+      "loss": 1.0092,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860300429247673e-05,
+      "loss": 1.0156,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860160548442817e-05,
+      "loss": 0.9592,
+      "step": 2306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600206029063247e-05,
+      "loss": 0.9963,
+      "step": 2307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598805926487283e-05,
+      "loss": 0.9695,
+      "step": 2308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597405176805635e-05,
+      "loss": 1.0278,
+      "step": 2309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596003780123725e-05,
+      "loss": 1.01,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594601736547012e-05,
+      "loss": 1.0286,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593199046181013e-05,
+      "loss": 1.0208,
+      "step": 2312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591795709131274e-05,
+      "loss": 1.0054,
+      "step": 2313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590391725503417e-05,
+      "loss": 1.0379,
+      "step": 2314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858898709540309e-05,
+      "loss": 1.0015,
+      "step": 2315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858758181893601e-05,
+      "loss": 1.0229,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858617589620792e-05,
+      "loss": 1.0215,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858476932732463e-05,
+      "loss": 0.9677,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858336211239199e-05,
+      "loss": 1.0254,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85819542515159e-05,
+      "loss": 1.0305,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580545744802307e-05,
+      "loss": 0.9889,
+      "step": 2321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579136592357213e-05,
+      "loss": 1.0107,
+      "step": 2322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857772679428666e-05,
+      "loss": 1.0044,
+      "step": 2323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576316350696744e-05,
+      "loss": 1.0364,
+      "step": 2324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857490526169361e-05,
+      "loss": 0.9993,
+      "step": 2325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573493527383447e-05,
+      "loss": 1.022,
+      "step": 2326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572081147872497e-05,
+      "loss": 1.0314,
+      "step": 2327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570668123267052e-05,
+      "loss": 1.0631,
+      "step": 2328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569254453673445e-05,
+      "loss": 0.9948,
+      "step": 2329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567840139198065e-05,
+      "loss": 1.0101,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566425179947346e-05,
+      "loss": 1.0731,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856500957602777e-05,
+      "loss": 1.0046,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856359332754587e-05,
+      "loss": 0.9552,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562176434608225e-05,
+      "loss": 1.0776,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560758897321465e-05,
+      "loss": 0.9904,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559340715792266e-05,
+      "loss": 0.9929,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557921890127356e-05,
+      "loss": 1.0075,
+      "step": 2337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556502420433507e-05,
+      "loss": 1.0143,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855508230681754e-05,
+      "loss": 0.9819,
+      "step": 2339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553661549386327e-05,
+      "loss": 0.9989,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552240148246793e-05,
+      "loss": 1.0171,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85508181035059e-05,
+      "loss": 1.0158,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549395415270664e-05,
+      "loss": 0.994,
+      "step": 2343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547972083648155e-05,
+      "loss": 1.0652,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546548108745482e-05,
+      "loss": 1.0397,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545123490669804e-05,
+      "loss": 1.0535,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854369822952834e-05,
+      "loss": 0.9801,
+      "step": 2347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854227232542834e-05,
+      "loss": 3.927,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540845778477116e-05,
+      "loss": 1.0651,
+      "step": 2349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539418588782017e-05,
+      "loss": 1.0315,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537990756450458e-05,
+      "loss": 4.0181,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536562281589877e-05,
+      "loss": 0.996,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535133164307785e-05,
+      "loss": 1.0249,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533703404711726e-05,
+      "loss": 0.9688,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85322730029093e-05,
+      "loss": 1.0292,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530841959008146e-05,
+      "loss": 0.9786,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852941027311596e-05,
+      "loss": 1.0141,
+      "step": 2357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527977945340492e-05,
+      "loss": 0.9742,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852654497578952e-05,
+      "loss": 0.9995,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525111364570892e-05,
+      "loss": 1.0337,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852367711179249e-05,
+      "loss": 0.9932,
+      "step": 2361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852224221756225e-05,
+      "loss": 1.0357,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520806681988158e-05,
+      "loss": 1.0124,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851937050517824e-05,
+      "loss": 1.044,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517933687240583e-05,
+      "loss": 1.0263,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516496228283308e-05,
+      "loss": 1.0278,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515058128414596e-05,
+      "loss": 1.0208,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851361938774267e-05,
+      "loss": 1.0195,
+      "step": 2368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512180006375804e-05,
+      "loss": 0.9651,
+      "step": 2369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510739984422317e-05,
+      "loss": 1.0168,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509299321990583e-05,
+      "loss": 1.0055,
+      "step": 2371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507858019189014e-05,
+      "loss": 1.0016,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506416076126073e-05,
+      "loss": 1.0121,
+      "step": 2373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504973492910288e-05,
+      "loss": 1.024,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503530269650207e-05,
+      "loss": 1.0523,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502086406454444e-05,
+      "loss": 1.0139,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850064190343166e-05,
+      "loss": 1.0214,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849919676069056e-05,
+      "loss": 1.0182,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497750978339897e-05,
+      "loss": 1.0317,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496304556488475e-05,
+      "loss": 0.9937,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494857495245145e-05,
+      "loss": 1.0582,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493409794718806e-05,
+      "loss": 1.03,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491961455018407e-05,
+      "loss": 1.0246,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849051247625294e-05,
+      "loss": 0.9997,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848906285853145e-05,
+      "loss": 3.7764,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848761260196303e-05,
+      "loss": 0.9728,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486161706656812e-05,
+      "loss": 0.9899,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484710172721995e-05,
+      "loss": 0.9809,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483258000267806e-05,
+      "loss": 1.0302,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481805189403533e-05,
+      "loss": 0.9912,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480351740238508e-05,
+      "loss": 1.0425,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478897652882105e-05,
+      "loss": 0.9905,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477442927443755e-05,
+      "loss": 1.0226,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847598756403294e-05,
+      "loss": 1.0245,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474531562759175e-05,
+      "loss": 1.0247,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473074923732035e-05,
+      "loss": 1.0313,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847161764706114e-05,
+      "loss": 1.0246,
+      "step": 2397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470159732856158e-05,
+      "loss": 1.0208,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468701181226803e-05,
+      "loss": 1.0084,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 1.0036,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465782166134087e-05,
+      "loss": 1.0568,
+      "step": 2401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464321702890392e-05,
+      "loss": 1.0,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462860602661668e-05,
+      "loss": 0.9916,
+      "step": 2403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461398865557873e-05,
+      "loss": 1.0164,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459936491689007e-05,
+      "loss": 1.0364,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845847348116512e-05,
+      "loss": 1.0314,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845700983409632e-05,
+      "loss": 1.0158,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455545550592744e-05,
+      "loss": 0.9468,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454080630764594e-05,
+      "loss": 0.9622,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452615074722104e-05,
+      "loss": 1.012,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451148882575576e-05,
+      "loss": 1.0289,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844968205443534e-05,
+      "loss": 0.9846,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844821459041179e-05,
+      "loss": 1.0611,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446746490615354e-05,
+      "loss": 1.0354,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445277755156518e-05,
+      "loss": 0.9967,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443808384145813e-05,
+      "loss": 0.9912,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442338377693813e-05,
+      "loss": 1.0205,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440867735911143e-05,
+      "loss": 1.0381,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439396458908482e-05,
+      "loss": 1.0546,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843792454679655e-05,
+      "loss": 1.0375,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436451999686114e-05,
+      "loss": 0.9903,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843497881768799e-05,
+      "loss": 1.0278,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843350500091305e-05,
+      "loss": 1.0011,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432030549472195e-05,
+      "loss": 0.97,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430555463476394e-05,
+      "loss": 0.9833,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429079743036653e-05,
+      "loss": 1.05,
+      "step": 2426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427603388264027e-05,
+      "loss": 1.0246,
+      "step": 2427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842612639926962e-05,
+      "loss": 0.9984,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424648776164578e-05,
+      "loss": 1.0032,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842317051906011e-05,
+      "loss": 1.0212,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842169162806746e-05,
+      "loss": 0.9861,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420212103297917e-05,
+      "loss": 1.0105,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418731944862825e-05,
+      "loss": 0.9991,
+      "step": 2433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417251152873577e-05,
+      "loss": 0.9683,
+      "step": 2434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415769727441608e-05,
+      "loss": 0.991,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414287668678402e-05,
+      "loss": 1.0425,
+      "step": 2436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412804976695492e-05,
+      "loss": 0.9655,
+      "step": 2437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411321651604464e-05,
+      "loss": 0.9964,
+      "step": 2438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409837693516935e-05,
+      "loss": 0.971,
+      "step": 2439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408353102544593e-05,
+      "loss": 0.9873,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406867878799154e-05,
+      "loss": 1.0143,
+      "step": 2441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405382022392392e-05,
+      "loss": 1.0083,
+      "step": 2442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403895533436124e-05,
+      "loss": 1.0145,
+      "step": 2443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402408412042213e-05,
+      "loss": 1.02,
+      "step": 2444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840092065832258e-05,
+      "loss": 0.9586,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839943227238918e-05,
+      "loss": 0.9763,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397943254354026e-05,
+      "loss": 0.9915,
+      "step": 2447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396453604329172e-05,
+      "loss": 0.9946,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839496332242672e-05,
+      "loss": 1.0172,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839347240875883e-05,
+      "loss": 1.0116,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391980863437693e-05,
+      "loss": 1.0133,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839048868657556e-05,
+      "loss": 1.0062,
+      "step": 2452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388995878284722e-05,
+      "loss": 1.0052,
+      "step": 2453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387502438677525e-05,
+      "loss": 1.0235,
+      "step": 2454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838600836786635e-05,
+      "loss": 1.0461,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384513665963645e-05,
+      "loss": 1.012,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383018333081882e-05,
+      "loss": 0.994,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381522369333606e-05,
+      "loss": 1.0674,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380025774831386e-05,
+      "loss": 3.8274,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378528549687856e-05,
+      "loss": 1.0285,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377030694015682e-05,
+      "loss": 0.9915,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375532207927593e-05,
+      "loss": 0.989,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837403309153635e-05,
+      "loss": 0.9825,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837253334495478e-05,
+      "loss": 1.0099,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371032968295734e-05,
+      "loss": 1.0261,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369531961672137e-05,
+      "loss": 1.0223,
+      "step": 2466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368030325196934e-05,
+      "loss": 1.0055,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366528058983144e-05,
+      "loss": 1.0043,
+      "step": 2468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365025163143814e-05,
+      "loss": 1.0038,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363521637792047e-05,
+      "loss": 1.017,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362017483040987e-05,
+      "loss": 1.0334,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360512699003834e-05,
+      "loss": 0.9975,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835900728579383e-05,
+      "loss": 0.963,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835750124352426e-05,
+      "loss": 1.0391,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355994572308472e-05,
+      "loss": 1.0295,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835448727225984e-05,
+      "loss": 1.0179,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352979343491807e-05,
+      "loss": 1.0383,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351470786117844e-05,
+      "loss": 1.0275,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349961600251482e-05,
+      "loss": 1.0334,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348451786006295e-05,
+      "loss": 1.0392,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346941343495907e-05,
+      "loss": 0.9874,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834543027283398e-05,
+      "loss": 0.9511,
+      "step": 2482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834391857413423e-05,
+      "loss": 1.003,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834240624751043e-05,
+      "loss": 0.9977,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834089329307638e-05,
+      "loss": 0.958,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833937971094595e-05,
+      "loss": 1.0299,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833786550123303e-05,
+      "loss": 0.9505,
+      "step": 2487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336350664051587e-05,
+      "loss": 1.0265,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334835199515607e-05,
+      "loss": 3.8913,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333319107739147e-05,
+      "loss": 0.9839,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331802388836294e-05,
+      "loss": 1.0276,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330285042921198e-05,
+      "loss": 0.9946,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328767070108036e-05,
+      "loss": 1.014,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327248470511052e-05,
+      "loss": 0.9488,
+      "step": 2494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325729244244524e-05,
+      "loss": 1.0448,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324209391422783e-05,
+      "loss": 1.0175,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832268891216021e-05,
+      "loss": 0.9939,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321167806571223e-05,
+      "loss": 1.0481,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319646074770294e-05,
+      "loss": 0.948,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831812371687195e-05,
+      "loss": 0.9926,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316600732990742e-05,
+      "loss": 0.9738,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315077123241295e-05,
+      "loss": 1.0131,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313552887738263e-05,
+      "loss": 1.0339,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831202802659635e-05,
+      "loss": 0.9876,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831050253993032e-05,
+      "loss": 1.0325,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308976427854964e-05,
+      "loss": 1.0027,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307449690485132e-05,
+      "loss": 0.9953,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305922327935725e-05,
+      "loss": 1.0398,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304394340321678e-05,
+      "loss": 3.6648,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302865727757987e-05,
+      "loss": 1.0075,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830133649035968e-05,
+      "loss": 1.0645,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299806628241844e-05,
+      "loss": 1.0337,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298276141519614e-05,
+      "loss": 1.0131,
+      "step": 2513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829674503030816e-05,
+      "loss": 1.0058,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829521329472271e-05,
+      "loss": 1.0385,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293680934878534e-05,
+      "loss": 1.0338,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829214795089095e-05,
+      "loss": 0.9537,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290614342875324e-05,
+      "loss": 1.0043,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828908011094707e-05,
+      "loss": 0.962,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287545255221642e-05,
+      "loss": 1.0256,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286009775814548e-05,
+      "loss": 1.0002,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284473672841344e-05,
+      "loss": 0.9844,
+      "step": 2522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828293694641763e-05,
+      "loss": 3.5464,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281399596659046e-05,
+      "loss": 1.0096,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279861623681293e-05,
+      "loss": 3.5373,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278323027600108e-05,
+      "loss": 1.0091,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827678380853128e-05,
+      "loss": 1.0274,
+      "step": 2527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827524396659064e-05,
+      "loss": 1.0393,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273703501894075e-05,
+      "loss": 1.0467,
+      "step": 2529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827216241455751e-05,
+      "loss": 1.0128,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270620704696916e-05,
+      "loss": 0.9841,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269078372428323e-05,
+      "loss": 1.0124,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267535417867797e-05,
+      "loss": 0.9868,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826599184113145e-05,
+      "loss": 3.6738,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264447642335448e-05,
+      "loss": 0.9973,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262902821595995e-05,
+      "loss": 1.0238,
+      "step": 2536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261357379029355e-05,
+      "loss": 0.994,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259811314751823e-05,
+      "loss": 1.0279,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258264628879753e-05,
+      "loss": 1.0362,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825671732152954e-05,
+      "loss": 1.021,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255169392817627e-05,
+      "loss": 0.9674,
+      "step": 2541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253620842860506e-05,
+      "loss": 3.6445,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825207167177471e-05,
+      "loss": 1.0125,
+      "step": 2543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250521879676823e-05,
+      "loss": 1.0519,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248971466683473e-05,
+      "loss": 0.9648,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247420432911345e-05,
+      "loss": 0.9763,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245868778477157e-05,
+      "loss": 0.9977,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244316503497676e-05,
+      "loss": 1.0299,
+      "step": 2548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242763608089723e-05,
+      "loss": 1.0181,
+      "step": 2549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241210092370164e-05,
+      "loss": 1.0015,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239655956455907e-05,
+      "loss": 1.0587,
+      "step": 2551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238101200463903e-05,
+      "loss": 0.9899,
+      "step": 2552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236545824511168e-05,
+      "loss": 1.0311,
+      "step": 2553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823498982871474e-05,
+      "loss": 1.0343,
+      "step": 2554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823343321319172e-05,
+      "loss": 0.9943,
+      "step": 2555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231875978059253e-05,
+      "loss": 0.9792,
+      "step": 2556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823031812343453e-05,
+      "loss": 1.0195,
+      "step": 2557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822875964943479e-05,
+      "loss": 0.959,
+      "step": 2558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822720055617731e-05,
+      "loss": 1.0295,
+      "step": 2559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225640843779422e-05,
+      "loss": 1.0293,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8224080512358506e-05,
+      "loss": 1.0055,
+      "step": 2561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822251956203198e-05,
+      "loss": 1.0348,
+      "step": 2562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822095799291732e-05,
+      "loss": 1.0236,
+      "step": 2563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219395805132037e-05,
+      "loss": 1.0276,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217832998793694e-05,
+      "loss": 1.0334,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216269574019906e-05,
+      "loss": 0.9776,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214705530928322e-05,
+      "loss": 0.9713,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821314086963665e-05,
+      "loss": 1.0193,
+      "step": 2568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821157559026264e-05,
+      "loss": 0.9926,
+      "step": 2569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821000969292408e-05,
+      "loss": 1.0374,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208443177738816e-05,
+      "loss": 1.027,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820687604482474e-05,
+      "loss": 1.0239,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820530829429978e-05,
+      "loss": 0.958,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203739926281926e-05,
+      "loss": 0.9706,
+      "step": 2574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82021709408892e-05,
+      "loss": 1.0107,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820060133823968e-05,
+      "loss": 1.0469,
+      "step": 2576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199031118451484e-05,
+      "loss": 1.0295,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819746028164278e-05,
+      "loss": 0.9503,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195888827931782e-05,
+      "loss": 1.0641,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819431675743675e-05,
+      "loss": 1.0007,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192744070275992e-05,
+      "loss": 0.967,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191170766567857e-05,
+      "loss": 1.0371,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189596846430753e-05,
+      "loss": 1.02,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188022309983114e-05,
+      "loss": 1.0085,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186447157343442e-05,
+      "loss": 0.9601,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184871388630277e-05,
+      "loss": 0.9794,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183295003962192e-05,
+      "loss": 0.9691,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181718003457827e-05,
+      "loss": 1.0359,
+      "step": 2588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818014038723586e-05,
+      "loss": 1.0317,
+      "step": 2589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178562155415013e-05,
+      "loss": 0.9943,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817698330811405e-05,
+      "loss": 1.0027,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81754038454518e-05,
+      "loss": 1.0008,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817382376754712e-05,
+      "loss": 0.9406,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172243074518916e-05,
+      "loss": 1.048,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170661766486147e-05,
+      "loss": 1.0294,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169079843567815e-05,
+      "loss": 0.9748,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167497305882967e-05,
+      "loss": 0.9665,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165914153550702e-05,
+      "loss": 1.0036,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816433038669015e-05,
+      "loss": 1.0338,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816274600542051e-05,
+      "loss": 0.9895,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161161009861007e-05,
+      "loss": 0.9962,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159575400130924e-05,
+      "loss": 0.9854,
+      "step": 2602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815798917634958e-05,
+      "loss": 1.0089,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156402338636356e-05,
+      "loss": 1.0114,
+      "step": 2604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154814887110666e-05,
+      "loss": 1.012,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815322682189197e-05,
+      "loss": 0.9948,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151638143099787e-05,
+      "loss": 0.9993,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150048850853664e-05,
+      "loss": 0.9854,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148458945273213e-05,
+      "loss": 1.0114,
+      "step": 2609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146868426478077e-05,
+      "loss": 1.0388,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145277294587953e-05,
+      "loss": 1.0319,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814368554972258e-05,
+      "loss": 3.5185,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142093192001747e-05,
+      "loss": 0.9987,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140500221545286e-05,
+      "loss": 0.9899,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138906638473076e-05,
+      "loss": 0.9876,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137312442905047e-05,
+      "loss": 0.966,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813571763496117e-05,
+      "loss": 1.0205,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134122214761453e-05,
+      "loss": 1.0035,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813252618242597e-05,
+      "loss": 1.0118,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813092953807483e-05,
+      "loss": 0.9992,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8129332281828187e-05,
+      "loss": 1.0433,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127734413806243e-05,
+      "loss": 1.0218,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812613593412924e-05,
+      "loss": 0.9993,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124536842917483e-05,
+      "loss": 1.0038,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812293714029131e-05,
+      "loss": 1.0125,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81213368263711e-05,
+      "loss": 1.03,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811973590127729e-05,
+      "loss": 1.0123,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811813436513036e-05,
+      "loss": 0.9869,
+      "step": 2628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116532218050827e-05,
+      "loss": 1.0547,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811492946015927e-05,
+      "loss": 1.0051,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113326091576296e-05,
+      "loss": 0.991,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811172211242257e-05,
+      "loss": 0.981,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811011752281881e-05,
+      "loss": 0.9879,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108512322885755e-05,
+      "loss": 1.0341,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106906512744213e-05,
+      "loss": 0.9989,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105300092515027e-05,
+      "loss": 0.9927,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810369306231909e-05,
+      "loss": 1.0358,
+      "step": 2637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810208542227734e-05,
+      "loss": 0.966,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100477172510756e-05,
+      "loss": 1.046,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098868313140376e-05,
+      "loss": 1.045,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809725884428726e-05,
+      "loss": 1.0487,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809564876607255e-05,
+      "loss": 1.0425,
+      "step": 2642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094038078617394e-05,
+      "loss": 0.9807,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809242678204302e-05,
+      "loss": 0.9963,
+      "step": 2644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090814876470674e-05,
+      "loss": 0.9918,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089202362021667e-05,
+      "loss": 0.9883,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087589238817347e-05,
+      "loss": 0.9531,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085975506979113e-05,
+      "loss": 1.0241,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084361166628405e-05,
+      "loss": 1.029,
+      "step": 2649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082746217886708e-05,
+      "loss": 0.974,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081130660875557e-05,
+      "loss": 0.998,
+      "step": 2651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079514495716536e-05,
+      "loss": 3.5322,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8077897722531264e-05,
+      "loss": 0.9841,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076280341441414e-05,
+      "loss": 0.9751,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074662352568704e-05,
+      "loss": 1.0436,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8073043756034892e-05,
+      "loss": 0.9967,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807142455196179e-05,
+      "loss": 1.018,
+      "step": 2657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806980474047125e-05,
+      "loss": 1.025,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806818432168517e-05,
+      "loss": 1.0423,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80665632957255e-05,
+      "loss": 0.9799,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064941662714227e-05,
+      "loss": 1.0131,
+      "step": 2661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063319422773388e-05,
+      "loss": 1.0267,
+      "step": 2662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061696576025063e-05,
+      "loss": 0.9873,
+      "step": 2663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060073122591383e-05,
+      "loss": 0.9722,
+      "step": 2664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058449062594522e-05,
+      "loss": 0.9865,
+      "step": 2665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056824396156695e-05,
+      "loss": 0.9861,
+      "step": 2666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055199123400168e-05,
+      "loss": 1.0161,
+      "step": 2667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053573244447256e-05,
+      "loss": 1.0032,
+      "step": 2668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051946759420313e-05,
+      "loss": 0.9426,
+      "step": 2669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050319668441733e-05,
+      "loss": 0.9949,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048691971633975e-05,
+      "loss": 0.9866,
+      "step": 2671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047063669119525e-05,
+      "loss": 1.0262,
+      "step": 2672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045434761020924e-05,
+      "loss": 1.034,
+      "step": 2673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043805247460752e-05,
+      "loss": 3.5682,
+      "step": 2674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804217512856164e-05,
+      "loss": 0.9837,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040544404446267e-05,
+      "loss": 3.5378,
+      "step": 2676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038913075237352e-05,
+      "loss": 1.0587,
+      "step": 2677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037281141057656e-05,
+      "loss": 1.0413,
+      "step": 2678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8035648602029997e-05,
+      "loss": 1.0325,
+      "step": 2679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034015458277233e-05,
+      "loss": 1.0326,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803238170992226e-05,
+      "loss": 0.9942,
+      "step": 2681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803074735708803e-05,
+      "loss": 1.0056,
+      "step": 2682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029112399897535e-05,
+      "loss": 1.0101,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027476838473816e-05,
+      "loss": 0.9568,
+      "step": 2684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025840672939964e-05,
+      "loss": 1.0676,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024203903419095e-05,
+      "loss": 1.0091,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022566530034393e-05,
+      "loss": 1.0027,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802092855290908e-05,
+      "loss": 0.984,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801928997216642e-05,
+      "loss": 1.0292,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017650787929728e-05,
+      "loss": 0.9988,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016011000322355e-05,
+      "loss": 0.9345,
+      "step": 2691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801437060946771e-05,
+      "loss": 0.9936,
+      "step": 2692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012729615489238e-05,
+      "loss": 1.0082,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011088018510433e-05,
+      "loss": 1.0205,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009445818654833e-05,
+      "loss": 1.0236,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007803016046025e-05,
+      "loss": 1.0182,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006159610807634e-05,
+      "loss": 1.046,
+      "step": 2697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004515603063336e-05,
+      "loss": 0.9795,
+      "step": 2698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002870992936857e-05,
+      "loss": 1.0024,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800122578055196e-05,
+      "loss": 0.9578,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799957996603245e-05,
+      "loss": 0.9977,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799793354950219e-05,
+      "loss": 1.0002,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996286531085078e-05,
+      "loss": 0.9877,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994638910905065e-05,
+      "loss": 1.0001,
+      "step": 2704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992990689086136e-05,
+      "loss": 1.0226,
+      "step": 2705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991341865752335e-05,
+      "loss": 1.0415,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989692441027744e-05,
+      "loss": 1.0185,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988042415036487e-05,
+      "loss": 1.0436,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986391787902743e-05,
+      "loss": 1.0182,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984740559750726e-05,
+      "loss": 0.9981,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79830887307047e-05,
+      "loss": 1.0149,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798143630088897e-05,
+      "loss": 0.9887,
+      "step": 2712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79797832704279e-05,
+      "loss": 1.0355,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978129639445883e-05,
+      "loss": 0.9825,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976475408067363e-05,
+      "loss": 0.9687,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797482057641683e-05,
+      "loss": 1.0389,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7973165144618824e-05,
+      "loss": 0.9686,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797150911279792e-05,
+      "loss": 0.9747,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969852481078743e-05,
+      "loss": 1.0577,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968195249585964e-05,
+      "loss": 0.9765,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79665374184443e-05,
+      "loss": 0.9784,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964878987778508e-05,
+      "loss": 3.9141,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79632199577134e-05,
+      "loss": 1.0235,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796156032837382e-05,
+      "loss": 0.9454,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959900099884667e-05,
+      "loss": 0.977,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958239272370885e-05,
+      "loss": 1.0292,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956577845957453e-05,
+      "loss": 1.0303,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954915820769407e-05,
+      "loss": 1.0195,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953253196931824e-05,
+      "loss": 0.9812,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951589974569823e-05,
+      "loss": 0.9928,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794992615380857e-05,
+      "loss": 0.9976,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948261734773274e-05,
+      "loss": 1.028,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79465967175892e-05,
+      "loss": 1.0277,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794493110238164e-05,
+      "loss": 0.9612,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943264889275944e-05,
+      "loss": 0.9975,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941598078397506e-05,
+      "loss": 1.0365,
+      "step": 2736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793993066987176e-05,
+      "loss": 1.0122,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938262663824183e-05,
+      "loss": 1.0423,
+      "step": 2738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793659406038031e-05,
+      "loss": 0.9905,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934924859665704e-05,
+      "loss": 0.9918,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793325506180599e-05,
+      "loss": 0.9842,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793158466692682e-05,
+      "loss": 0.9829,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929913675153904e-05,
+      "loss": 0.9786,
+      "step": 2743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928242086612995e-05,
+      "loss": 1.036,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926569901429888e-05,
+      "loss": 1.0176,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924897119730418e-05,
+      "loss": 1.0303,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7923223741640477e-05,
+      "loss": 0.9969,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921549767285997e-05,
+      "loss": 0.9798,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791987519679295e-05,
+      "loss": 0.973,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791820003028735e-05,
+      "loss": 1.0161,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791652426789528e-05,
+      "loss": 1.025,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791484790974283e-05,
+      "loss": 0.986,
+      "step": 2752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913170955956168e-05,
+      "loss": 0.9716,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911493406661487e-05,
+      "loss": 0.9724,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909815261985034e-05,
+      "loss": 1.0213,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908136522053098e-05,
+      "loss": 0.9501,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906457186992018e-05,
+      "loss": 0.997,
+      "step": 2757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790477725692816e-05,
+      "loss": 0.9944,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790309673198796e-05,
+      "loss": 1.0029,
+      "step": 2759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901415612297883e-05,
+      "loss": 1.0407,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789973389798444e-05,
+      "loss": 0.982,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789805158917419e-05,
+      "loss": 0.9921,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896368685993738e-05,
+      "loss": 0.9738,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894685188569726e-05,
+      "loss": 0.9947,
+      "step": 2764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789300109702885e-05,
+      "loss": 1.0041,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891316411497845e-05,
+      "loss": 1.0554,
+      "step": 2766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889631132103496e-05,
+      "loss": 1.0641,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887945258972624e-05,
+      "loss": 1.0273,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886258792232105e-05,
+      "loss": 0.999,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788457173200885e-05,
+      "loss": 1.0262,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882884078429824e-05,
+      "loss": 0.9721,
+      "step": 2771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881195831622027e-05,
+      "loss": 1.0312,
+      "step": 2772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787950699171251e-05,
+      "loss": 1.079,
+      "step": 2773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787781755882837e-05,
+      "loss": 1.0356,
+      "step": 2774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876127533096744e-05,
+      "loss": 0.9997,
+      "step": 2775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874436914644813e-05,
+      "loss": 0.9489,
+      "step": 2776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787274570359981e-05,
+      "loss": 0.9989,
+      "step": 2777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871053900089e-05,
+      "loss": 1.0057,
+      "step": 2778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786936150423971e-05,
+      "loss": 1.0113,
+      "step": 2779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867668516179294e-05,
+      "loss": 0.9597,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865974936035158e-05,
+      "loss": 0.9177,
+      "step": 2781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786428076393476e-05,
+      "loss": 1.0053,
+      "step": 2782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786258600000559e-05,
+      "loss": 1.0092,
+      "step": 2783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786089064437519e-05,
+      "loss": 1.0116,
+      "step": 2784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859194697171142e-05,
+      "loss": 0.9954,
+      "step": 2785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785749815852108e-05,
+      "loss": 1.0034,
+      "step": 2786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785580102855267e-05,
+      "loss": 1.0616,
+      "step": 2787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854103307393634e-05,
+      "loss": 1.0129,
+      "step": 2788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785240499517174e-05,
+      "loss": 1.0317,
+      "step": 2789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850706092014783e-05,
+      "loss": 1.0033,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849006598050626e-05,
+      "loss": 0.9971,
+      "step": 2791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847306513407158e-05,
+      "loss": 3.7862,
+      "step": 2792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784560583821232e-05,
+      "loss": 0.9889,
+      "step": 2793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78439045725941e-05,
+      "loss": 0.9782,
+      "step": 2794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842202716680525e-05,
+      "loss": 1.0504,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784050027059967e-05,
+      "loss": 0.9959,
+      "step": 2796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7838797234479653e-05,
+      "loss": 1.0794,
+      "step": 2797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837093608448635e-05,
+      "loss": 1.0054,
+      "step": 2798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835389392634823e-05,
+      "loss": 1.004,
+      "step": 2799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783368458716647e-05,
+      "loss": 0.9677,
+      "step": 2800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831979192171868e-05,
+      "loss": 1.0061,
+      "step": 2801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783027320777936e-05,
+      "loss": 1.014,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782856663411733e-05,
+      "loss": 1.0174,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782685947131421e-05,
+      "loss": 1.0201,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825151719498465e-05,
+      "loss": 1.0113,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7823443378798616e-05,
+      "loss": 1.0244,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821734449343232e-05,
+      "loss": 1.0377,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820024931260905e-05,
+      "loss": 1.0371,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.9782,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78166041297301e-05,
+      "loss": 0.9794,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814892846539052e-05,
+      "loss": 0.9878,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781318097523593e-05,
+      "loss": 0.9866,
+      "step": 2812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811468515949574e-05,
+      "loss": 3.5126,
+      "step": 2813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809755468808844e-05,
+      "loss": 0.9867,
+      "step": 2814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780804183394266e-05,
+      "loss": 1.0033,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806327611479983e-05,
+      "loss": 3.5207,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804612801549817e-05,
+      "loss": 0.9974,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780289740428121e-05,
+      "loss": 1.0247,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801181419803257e-05,
+      "loss": 1.0499,
+      "step": 2819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779946484824509e-05,
+      "loss": 0.9908,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797747689735896e-05,
+      "loss": 1.0291,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796029944404897e-05,
+      "loss": 1.0207,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779431161238136e-05,
+      "loss": 0.981,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792592693794605e-05,
+      "loss": 1.0183,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790873188773983e-05,
+      "loss": 1.0157,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789153097448898e-05,
+      "loss": 1.0022,
+      "step": 2826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77874324199488e-05,
+      "loss": 1.0321,
+      "step": 2827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785711156403175e-05,
+      "loss": 0.9965,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778398930694156e-05,
+      "loss": 0.9935,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782266871693527e-05,
+      "loss": 0.9884,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780543850788706e-05,
+      "loss": 1.0064,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777882024435676e-05,
+      "loss": 0.9652,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77770960525274e-05,
+      "loss": 0.9804,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775371275430378e-05,
+      "loss": 1.0123,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77736459131955e-05,
+      "loss": 0.9906,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77719199659526e-05,
+      "loss": 0.9669,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770193433831565e-05,
+      "loss": 0.9764,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7768466316962332e-05,
+      "loss": 1.009,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766738615474876e-05,
+      "loss": 1.0284,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776501032949921e-05,
+      "loss": 0.9734,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763281459165398e-05,
+      "loss": 1.0101,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761552004603548e-05,
+      "loss": 1.0252,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775982196594381e-05,
+      "loss": 0.9939,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758091343316383e-05,
+      "loss": 0.9991,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756360136851497e-05,
+      "loss": 0.9663,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775462834667944e-05,
+      "loss": 0.9921,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752895972930538e-05,
+      "loss": 1.0152,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751163015735157e-05,
+      "loss": 0.9882,
+      "step": 2848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749429475223718e-05,
+      "loss": 3.4891,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747695351526675e-05,
+      "loss": 1.0625,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774596064477453e-05,
+      "loss": 1.0158,
+      "step": 2851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774422535509783e-05,
+      "loss": 1.0002,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742489482627164e-05,
+      "loss": 1.0007,
+      "step": 2853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740753027493168e-05,
+      "loss": 1.0267,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739015989826515e-05,
+      "loss": 1.0254,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737278369757932e-05,
+      "loss": 0.9479,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735540167418175e-05,
+      "loss": 0.9998,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773380138293807e-05,
+      "loss": 1.0657,
+      "step": 2858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732062016448447e-05,
+      "loss": 0.9924,
+      "step": 2859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730322068080218e-05,
+      "loss": 0.9421,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772858153796432e-05,
+      "loss": 0.9675,
+      "step": 2861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726840426231737e-05,
+      "loss": 0.9864,
+      "step": 2862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725098733013495e-05,
+      "loss": 0.9955,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772335645844067e-05,
+      "loss": 0.9898,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721613602644367e-05,
+      "loss": 1.0076,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771987016575576e-05,
+      "loss": 0.9927,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771812614790604e-05,
+      "loss": 0.9661,
+      "step": 2867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716381549226463e-05,
+      "loss": 1.0193,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771463636984831e-05,
+      "loss": 0.9564,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771289060990292e-05,
+      "loss": 0.9868,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771114426952167e-05,
+      "loss": 1.0583,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770939734883598e-05,
+      "loss": 0.9798,
+      "step": 2872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707649847977318e-05,
+      "loss": 1.0212,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705901767077193e-05,
+      "loss": 1.0105,
+      "step": 2874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770415310626715e-05,
+      "loss": 1.0158,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702403865678795e-05,
+      "loss": 1.0272,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770065404544376e-05,
+      "loss": 1.0037,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698903645693733e-05,
+      "loss": 0.9968,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769715266656044e-05,
+      "loss": 1.0063,
+      "step": 2879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695401108175648e-05,
+      "loss": 0.9915,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693648970671176e-05,
+      "loss": 0.9865,
+      "step": 2881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691896254178878e-05,
+      "loss": 1.0191,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690142958830658e-05,
+      "loss": 0.9569,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688389084758455e-05,
+      "loss": 0.9621,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768663463209426e-05,
+      "loss": 1.0255,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684879600970113e-05,
+      "loss": 0.9784,
+      "step": 2886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768312399151808e-05,
+      "loss": 1.0091,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681367803870282e-05,
+      "loss": 0.9716,
+      "step": 2888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767961103815888e-05,
+      "loss": 0.9902,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677853694516083e-05,
+      "loss": 0.9922,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767609577307414e-05,
+      "loss": 0.9884,
+      "step": 2891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674337273965345e-05,
+      "loss": 1.0059,
+      "step": 2892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767257819732203e-05,
+      "loss": 0.9818,
+      "step": 2893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767081854327658e-05,
+      "loss": 0.9668,
+      "step": 2894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669058311961413e-05,
+      "loss": 1.0053,
+      "step": 2895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667297503509e-05,
+      "loss": 1.0186,
+      "step": 2896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766553611805185e-05,
+      "loss": 1.0495,
+      "step": 2897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766377415572252e-05,
+      "loss": 0.9657,
+      "step": 2898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662011616653597e-05,
+      "loss": 1.0123,
+      "step": 2899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660248500977727e-05,
+      "loss": 1.0064,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658484808827597e-05,
+      "loss": 1.0257,
+      "step": 2901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765672054033593e-05,
+      "loss": 0.9963,
+      "step": 2902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654955695635498e-05,
+      "loss": 0.9664,
+      "step": 2903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653190274859117e-05,
+      "loss": 0.9862,
+      "step": 2904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651424278139644e-05,
+      "loss": 1.0019,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649657705609975e-05,
+      "loss": 3.664,
+      "step": 2906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647890557403055e-05,
+      "loss": 1.0038,
+      "step": 2907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646122833651874e-05,
+      "loss": 1.0,
+      "step": 2908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7644354534489466e-05,
+      "loss": 0.9346,
+      "step": 2909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642585660048898e-05,
+      "loss": 0.985,
+      "step": 2910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764081621046329e-05,
+      "loss": 1.0414,
+      "step": 2911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76390461858658e-05,
+      "loss": 1.0061,
+      "step": 2912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637275586389633e-05,
+      "loss": 1.0099,
+      "step": 2913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763550441216804e-05,
+      "loss": 0.9527,
+      "step": 2914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633732663334308e-05,
+      "loss": 1.038,
+      "step": 2915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631960340021767e-05,
+      "loss": 1.0398,
+      "step": 2916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76301874423638e-05,
+      "loss": 1.0177,
+      "step": 2917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762841397049382e-05,
+      "loss": 1.0041,
+      "step": 2918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76266399245453e-05,
+      "loss": 1.0127,
+      "step": 2919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624865304651743e-05,
+      "loss": 1.0355,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762309011094669e-05,
+      "loss": 1.0181,
+      "step": 2921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762131434356374e-05,
+      "loss": 1.0047,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619538002636533e-05,
+      "loss": 1.0221,
+      "step": 2923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617761088298742e-05,
+      "loss": 1.0115,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615983600684093e-05,
+      "loss": 1.0134,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614205539926345e-05,
+      "loss": 0.9722,
+      "step": 2926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612426906159315e-05,
+      "loss": 0.9519,
+      "step": 2927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761064769951685e-05,
+      "loss": 1.0097,
+      "step": 2928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608867920132842e-05,
+      "loss": 1.0178,
+      "step": 2929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607087568141238e-05,
+      "loss": 0.9611,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760530664367601e-05,
+      "loss": 0.9874,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760352514687118e-05,
+      "loss": 0.9887,
+      "step": 2932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601743077860826e-05,
+      "loss": 1.019,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599960436779053e-05,
+      "loss": 1.017,
+      "step": 2934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759817722376001e-05,
+      "loss": 0.9829,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75963934389379e-05,
+      "loss": 0.9989,
+      "step": 2936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594609082446955e-05,
+      "loss": 1.0126,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592824154421463e-05,
+      "loss": 0.9888,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759103865499575e-05,
+      "loss": 3.6991,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589252584304177e-05,
+      "loss": 0.9772,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587465942481165e-05,
+      "loss": 0.9561,
+      "step": 2941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758567872966116e-05,
+      "loss": 1.0284,
+      "step": 2942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583890945978663e-05,
+      "loss": 0.9392,
+      "step": 2943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582102591568215e-05,
+      "loss": 0.9662,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580313666564396e-05,
+      "loss": 0.9619,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578524171101835e-05,
+      "loss": 1.002,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75767341053152e-05,
+      "loss": 0.9872,
+      "step": 2947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574943469339202e-05,
+      "loss": 1.0311,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573152263308596e-05,
+      "loss": 0.9824,
+      "step": 2949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571360487358183e-05,
+      "loss": 0.9969,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569568141622802e-05,
+      "loss": 0.9959,
+      "step": 2951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567775226237333e-05,
+      "loss": 0.9675,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565981741336706e-05,
+      "loss": 0.9836,
+      "step": 2953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564187687055886e-05,
+      "loss": 1.0242,
+      "step": 2954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756239306352989e-05,
+      "loss": 0.9609,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560597870893775e-05,
+      "loss": 0.9543,
+      "step": 2956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755880210928263e-05,
+      "loss": 1.0226,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557005778831604e-05,
+      "loss": 1.0289,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555208879675875e-05,
+      "loss": 1.0183,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553411411950673e-05,
+      "loss": 0.9974,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551613375791265e-05,
+      "loss": 1.0217,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754981477133296e-05,
+      "loss": 1.0328,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548015598711113e-05,
+      "loss": 0.9974,
+      "step": 2963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754621585806113e-05,
+      "loss": 3.5355,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754441554951844e-05,
+      "loss": 0.9773,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542614673218536e-05,
+      "loss": 1.0194,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754081322929693e-05,
+      "loss": 1.0281,
+      "step": 2967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539011217889203e-05,
+      "loss": 1.0373,
+      "step": 2968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753720863913096e-05,
+      "loss": 0.957,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535405493157854e-05,
+      "loss": 0.9837,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533601780105588e-05,
+      "loss": 0.9843,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531797500109888e-05,
+      "loss": 0.9698,
+      "step": 2972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752999265330655e-05,
+      "loss": 1.0057,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752818723983139e-05,
+      "loss": 0.9956,
+      "step": 2974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526381259820272e-05,
+      "loss": 1.01,
+      "step": 2975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524574713409116e-05,
+      "loss": 1.0094,
+      "step": 2976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522767600733863e-05,
+      "loss": 1.0211,
+      "step": 2977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752095992193052e-05,
+      "loss": 0.9959,
+      "step": 2978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519151677135112e-05,
+      "loss": 1.0151,
+      "step": 2979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517342866483728e-05,
+      "loss": 1.0023,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515533490112483e-05,
+      "loss": 1.0002,
+      "step": 2981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513723548157546e-05,
+      "loss": 0.9989,
+      "step": 2982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751191304075513e-05,
+      "loss": 0.9875,
+      "step": 2983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510101968041476e-05,
+      "loss": 1.0304,
+      "step": 2984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508290330152883e-05,
+      "loss": 1.0095,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506478127225683e-05,
+      "loss": 1.0183,
+      "step": 2986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504665359396255e-05,
+      "loss": 0.9962,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750285202680102e-05,
+      "loss": 0.9732,
+      "step": 2988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501038129576442e-05,
+      "loss": 1.0229,
+      "step": 2989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499223667859024e-05,
+      "loss": 1.0051,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497408641785315e-05,
+      "loss": 0.9752,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495593051491905e-05,
+      "loss": 0.9979,
+      "step": 2992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493776897115427e-05,
+      "loss": 0.9683,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491960178792557e-05,
+      "loss": 1.0307,
+      "step": 2994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749014289666001e-05,
+      "loss": 1.0071,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488325050854545e-05,
+      "loss": 0.9487,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748650664151297e-05,
+      "loss": 1.0447,
+      "step": 2997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484687668772133e-05,
+      "loss": 0.9733,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482868132768905e-05,
+      "loss": 0.9803,
+      "step": 2999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481048033640234e-05,
+      "loss": 0.9934,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747922737152308e-05,
+      "loss": 0.9717,
+      "step": 3001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477406146554467e-05,
+      "loss": 1.0074,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7475584358871443e-05,
+      "loss": 1.0143,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473762008611115e-05,
+      "loss": 0.9838,
+      "step": 3004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471939095910616e-05,
+      "loss": 1.0004,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470115620907136e-05,
+      "loss": 0.9825,
+      "step": 3006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468291583737898e-05,
+      "loss": 0.9949,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746646698454017e-05,
+      "loss": 1.0328,
+      "step": 3008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746464182345127e-05,
+      "loss": 0.9703,
+      "step": 3009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462816100608543e-05,
+      "loss": 0.9758,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746098981614939e-05,
+      "loss": 1.0069,
+      "step": 3011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745916297021124e-05,
+      "loss": 0.9918,
+      "step": 3012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745733556293158e-05,
+      "loss": 0.9663,
+      "step": 3013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745550759444793e-05,
+      "loss": 0.999,
+      "step": 3014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745367906489786e-05,
+      "loss": 1.0341,
+      "step": 3015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451849974418963e-05,
+      "loss": 1.0245,
+      "step": 3016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450020323148902e-05,
+      "loss": 1.0076,
+      "step": 3017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744819011122536e-05,
+      "loss": 0.9499,
+      "step": 3018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744635933878607e-05,
+      "loss": 1.017,
+      "step": 3019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444528005968813e-05,
+      "loss": 1.0117,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74426961129114e-05,
+      "loss": 0.9613,
+      "step": 3021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440863659751696e-05,
+      "loss": 1.0146,
+      "step": 3022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74390306466276e-05,
+      "loss": 0.9865,
+      "step": 3023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743719707367706e-05,
+      "loss": 1.0033,
+      "step": 3024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743536294103805e-05,
+      "loss": 1.022,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433528248848612e-05,
+      "loss": 1.007,
+      "step": 3026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743169299724681e-05,
+      "loss": 1.0631,
+      "step": 3027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429857186370755e-05,
+      "loss": 0.9985,
+      "step": 3028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428020816358606e-05,
+      "loss": 0.9923,
+      "step": 3029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426183887348557e-05,
+      "loss": 1.0189,
+      "step": 3030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742434639947885e-05,
+      "loss": 3.4644,
+      "step": 3031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422508352887756e-05,
+      "loss": 1.0328,
+      "step": 3032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742066974771361e-05,
+      "loss": 0.9855,
+      "step": 3033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741883058409477e-05,
+      "loss": 1.0125,
+      "step": 3034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741699086216964e-05,
+      "loss": 1.0056,
+      "step": 3035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741515058207668e-05,
+      "loss": 0.9924,
+      "step": 3036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413309743954367e-05,
+      "loss": 3.5351,
+      "step": 3037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411468347941243e-05,
+      "loss": 1.0249,
+      "step": 3038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740962639417588e-05,
+      "loss": 0.9605,
+      "step": 3039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407783882796897e-05,
+      "loss": 0.9403,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405940813942946e-05,
+      "loss": 1.0259,
+      "step": 3041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404097187752737e-05,
+      "loss": 0.981,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402253004365007e-05,
+      "loss": 0.9369,
+      "step": 3043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740040826391854e-05,
+      "loss": 1.0071,
+      "step": 3044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398562966552165e-05,
+      "loss": 0.9979,
+      "step": 3045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739671711240475e-05,
+      "loss": 0.9345,
+      "step": 3046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394870701615208e-05,
+      "loss": 1.0241,
+      "step": 3047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739302373432248e-05,
+      "loss": 3.3709,
+      "step": 3048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391176210665576e-05,
+      "loss": 0.9788,
+      "step": 3049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738932813078352e-05,
+      "loss": 1.0307,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387479494815392e-05,
+      "loss": 1.0193,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385630302900317e-05,
+      "loss": 0.9756,
+      "step": 3052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7383780555177452e-05,
+      "loss": 0.9836,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381930251786e-05,
+      "loss": 1.0287,
+      "step": 3054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380079392865213e-05,
+      "loss": 1.0234,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7378227978554365e-05,
+      "loss": 1.0132,
+      "step": 3056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376376008992796e-05,
+      "loss": 1.0298,
+      "step": 3057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737452348431987e-05,
+      "loss": 3.2245,
+      "step": 3058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372670404675003e-05,
+      "loss": 1.014,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737081677019765e-05,
+      "loss": 0.9892,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73689625810273e-05,
+      "loss": 1.0182,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73671078373035e-05,
+      "loss": 1.0583,
+      "step": 3062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365252539165826e-05,
+      "loss": 0.9811,
+      "step": 3063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363396686753893e-05,
+      "loss": 1.0352,
+      "step": 3064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361540280207368e-05,
+      "loss": 0.9678,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359683319665958e-05,
+      "loss": 1.0327,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357825805269405e-05,
+      "loss": 0.9606,
+      "step": 3067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73559677371575e-05,
+      "loss": 0.9874,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735410911547007e-05,
+      "loss": 1.0331,
+      "step": 3069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735224994034699e-05,
+      "loss": 0.9921,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350390211928167e-05,
+      "loss": 0.9909,
+      "step": 3071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734852993035356e-05,
+      "loss": 0.992,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346669095763163e-05,
+      "loss": 1.0357,
+      "step": 3073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344807708297017e-05,
+      "loss": 1.0272,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734294576809519e-05,
+      "loss": 0.9762,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734108327529782e-05,
+      "loss": 0.9738,
+      "step": 3076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339220230045058e-05,
+      "loss": 0.9783,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733735663247711e-05,
+      "loss": 1.02,
+      "step": 3078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733549248273422e-05,
+      "loss": 0.9812,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733362778095668e-05,
+      "loss": 0.9775,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331762527284818e-05,
+      "loss": 0.9928,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329896721858998e-05,
+      "loss": 1.0238,
+      "step": 3082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328030364819638e-05,
+      "loss": 1.0006,
+      "step": 3083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7326163456307188e-05,
+      "loss": 0.9467,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7324295996462145e-05,
+      "loss": 0.9587,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322427985425045e-05,
+      "loss": 1.0063,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320559423336465e-05,
+      "loss": 1.0331,
+      "step": 3087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318690310337022e-05,
+      "loss": 0.9917,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316820646567382e-05,
+      "loss": 3.3516,
+      "step": 3089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314950432168244e-05,
+      "loss": 1.0038,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731307966728035e-05,
+      "loss": 0.969,
+      "step": 3091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311208352044487e-05,
+      "loss": 3.3683,
+      "step": 3092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309336486601477e-05,
+      "loss": 1.0165,
+      "step": 3093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73074640710922e-05,
+      "loss": 0.9666,
+      "step": 3094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730559110565755e-05,
+      "loss": 1.0368,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303717590438483e-05,
+      "loss": 1.0481,
+      "step": 3096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301843525575997e-05,
+      "loss": 1.0411,
+      "step": 3097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729996891121112e-05,
+      "loss": 0.9838,
+      "step": 3098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298093747484923e-05,
+      "loss": 0.9909,
+      "step": 3099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729621803453853e-05,
+      "loss": 1.0351,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294341772513096e-05,
+      "loss": 0.9942,
+      "step": 3101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292464961549815e-05,
+      "loss": 0.9627,
+      "step": 3102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729058760178993e-05,
+      "loss": 0.9993,
+      "step": 3103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288709693374727e-05,
+      "loss": 3.3896,
+      "step": 3104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286831236445517e-05,
+      "loss": 0.9772,
+      "step": 3105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284952231143677e-05,
+      "loss": 1.0297,
+      "step": 3106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283072677610604e-05,
+      "loss": 0.9807,
+      "step": 3107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281192575987745e-05,
+      "loss": 0.9742,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727931192641659e-05,
+      "loss": 3.533,
+      "step": 3109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727743072903867e-05,
+      "loss": 0.9859,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275548983995548e-05,
+      "loss": 1.0177,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273666691428838e-05,
+      "loss": 1.0173,
+      "step": 3112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271783851480195e-05,
+      "loss": 0.9935,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726990046429131e-05,
+      "loss": 1.0109,
+      "step": 3114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268016530003923e-05,
+      "loss": 1.0299,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266132048759802e-05,
+      "loss": 1.0256,
+      "step": 3116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264247020700772e-05,
+      "loss": 0.9976,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262361445968684e-05,
+      "loss": 1.009,
+      "step": 3118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260475324705445e-05,
+      "loss": 0.9797,
+      "step": 3119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725858865705299e-05,
+      "loss": 0.9791,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256701443153303e-05,
+      "loss": 0.9741,
+      "step": 3121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254813683148405e-05,
+      "loss": 0.9587,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725292537718036e-05,
+      "loss": 0.9471,
+      "step": 3123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251036525391277e-05,
+      "loss": 1.0006,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72491471279233e-05,
+      "loss": 0.9657,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247257184918615e-05,
+      "loss": 0.9576,
+      "step": 3126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245366696519448e-05,
+      "loss": 0.9917,
+      "step": 3127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243475662868073e-05,
+      "loss": 3.5416,
+      "step": 3128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7241584084106798e-05,
+      "loss": 0.9637,
+      "step": 3129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239691960377976e-05,
+      "loss": 1.0212,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237799291823996e-05,
+      "loss": 3.4264,
+      "step": 3131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235906078587295e-05,
+      "loss": 1.0181,
+      "step": 3132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234012320810345e-05,
+      "loss": 0.9747,
+      "step": 3133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232118018635663e-05,
+      "loss": 1.0356,
+      "step": 3134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230223172205802e-05,
+      "loss": 0.9871,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228327781663362e-05,
+      "loss": 1.0372,
+      "step": 3136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226431847150983e-05,
+      "loss": 0.9984,
+      "step": 3137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722453536881134e-05,
+      "loss": 0.9516,
+      "step": 3138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222638346787155e-05,
+      "loss": 0.972,
+      "step": 3139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220740781221186e-05,
+      "loss": 0.9956,
+      "step": 3140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721884267225624e-05,
+      "loss": 0.9938,
+      "step": 3141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216944020035156e-05,
+      "loss": 0.995,
+      "step": 3142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215044824700823e-05,
+      "loss": 0.9974,
+      "step": 3143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721314508639616e-05,
+      "loss": 3.4277,
+      "step": 3144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211244805264132e-05,
+      "loss": 0.9475,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209343981447747e-05,
+      "loss": 0.9552,
+      "step": 3146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207442615090053e-05,
+      "loss": 0.9987,
+      "step": 3147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205540706334137e-05,
+      "loss": 1.0325,
+      "step": 3148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203638255323126e-05,
+      "loss": 1.0001,
+      "step": 3149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720173526220019e-05,
+      "loss": 3.247,
+      "step": 3150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199831727108548e-05,
+      "loss": 1.0444,
+      "step": 3151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197927650191436e-05,
+      "loss": 0.9567,
+      "step": 3152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196023031592156e-05,
+      "loss": 1.0369,
+      "step": 3153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719411787145404e-05,
+      "loss": 1.0541,
+      "step": 3154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192212169920458e-05,
+      "loss": 0.9863,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190305927134824e-05,
+      "loss": 0.9816,
+      "step": 3156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188399143240597e-05,
+      "loss": 0.9824,
+      "step": 3157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718649181838127e-05,
+      "loss": 1.0034,
+      "step": 3158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718458395270038e-05,
+      "loss": 0.9795,
+      "step": 3159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71826755463415e-05,
+      "loss": 0.9474,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180766599448256e-05,
+      "loss": 0.9798,
+      "step": 3161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7178857112164298e-05,
+      "loss": 0.9796,
+      "step": 3162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717694708463333e-05,
+      "loss": 0.9544,
+      "step": 3163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175036516999088e-05,
+      "loss": 1.0083,
+      "step": 3164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173125409405354e-05,
+      "loss": 0.9737,
+      "step": 3165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717121376199595e-05,
+      "loss": 0.9722,
+      "step": 3166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169301574914737e-05,
+      "loss": 0.9526,
+      "step": 3167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167388848305623e-05,
+      "loss": 1.0315,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165475582312537e-05,
+      "loss": 1.0583,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163561777079474e-05,
+      "loss": 1.0302,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161647432750455e-05,
+      "loss": 0.9912,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159732549469543e-05,
+      "loss": 1.0257,
+      "step": 3172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157817127380843e-05,
+      "loss": 1.0058,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155901166628503e-05,
+      "loss": 0.9736,
+      "step": 3174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715398466735671e-05,
+      "loss": 0.9861,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152067629709685e-05,
+      "loss": 1.0076,
+      "step": 3176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150150053831702e-05,
+      "loss": 1.0189,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148231939867065e-05,
+      "loss": 1.0143,
+      "step": 3178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714631328796012e-05,
+      "loss": 1.0147,
+      "step": 3179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144394098255258e-05,
+      "loss": 0.9941,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142474370896913e-05,
+      "loss": 0.9591,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714055410602955e-05,
+      "loss": 1.0122,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138633303797676e-05,
+      "loss": 1.0321,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713671196434585e-05,
+      "loss": 1.0415,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713479008781865e-05,
+      "loss": 1.0314,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132867674360724e-05,
+      "loss": 0.9663,
+      "step": 3186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713094472411673e-05,
+      "loss": 0.9828,
+      "step": 3187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129021237231386e-05,
+      "loss": 1.0023,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127097213849445e-05,
+      "loss": 0.9739,
+      "step": 3189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71251726541157e-05,
+      "loss": 1.049,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123247558174984e-05,
+      "loss": 0.9919,
+      "step": 3191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121321926172166e-05,
+      "loss": 1.0008,
+      "step": 3192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119395758252165e-05,
+      "loss": 1.0292,
+      "step": 3193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117469054559935e-05,
+      "loss": 0.9672,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711554181524047e-05,
+      "loss": 0.9914,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113614040438806e-05,
+      "loss": 0.9711,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111685730300016e-05,
+      "loss": 3.3878,
+      "step": 3197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109756884969215e-05,
+      "loss": 0.9992,
+      "step": 3198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107827504591563e-05,
+      "loss": 0.9953,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105897589312253e-05,
+      "loss": 1.0077,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710396713927652e-05,
+      "loss": 1.0106,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710203615462964e-05,
+      "loss": 1.006,
+      "step": 3202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710010463551694e-05,
+      "loss": 0.9971,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098172582083764e-05,
+      "loss": 3.2758,
+      "step": 3204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096239994475516e-05,
+      "loss": 1.0092,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094306872837633e-05,
+      "loss": 0.9299,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709237321731559e-05,
+      "loss": 1.0157,
+      "step": 3207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709043902805491e-05,
+      "loss": 0.9701,
+      "step": 3208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088504305201145e-05,
+      "loss": 1.0174,
+      "step": 3209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708656904889989e-05,
+      "loss": 0.949,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084633259296798e-05,
+      "loss": 1.0027,
+      "step": 3211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082696936537534e-05,
+      "loss": 0.971,
+      "step": 3212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080760080767822e-05,
+      "loss": 0.9956,
+      "step": 3213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078822692133417e-05,
+      "loss": 0.9935,
+      "step": 3214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076884770780123e-05,
+      "loss": 1.029,
+      "step": 3215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074946316853776e-05,
+      "loss": 0.9903,
+      "step": 3216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707300733050025e-05,
+      "loss": 1.0028,
+      "step": 3217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 1.0283,
+      "step": 3218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069127761095403e-05,
+      "loss": 0.9747,
+      "step": 3219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067187178336034e-05,
+      "loss": 1.0079,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065246063733408e-05,
+      "loss": 1.0274,
+      "step": 3221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063304417433602e-05,
+      "loss": 3.3587,
+      "step": 3222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061362239582736e-05,
+      "loss": 0.9921,
+      "step": 3223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059419530326973e-05,
+      "loss": 0.9914,
+      "step": 3224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057476289812504e-05,
+      "loss": 1.0007,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705553251818558e-05,
+      "loss": 1.0077,
+      "step": 3226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053588215592466e-05,
+      "loss": 0.987,
+      "step": 3227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051643382179493e-05,
+      "loss": 0.99,
+      "step": 3228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704969801809301e-05,
+      "loss": 3.1924,
+      "step": 3229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047752123479428e-05,
+      "loss": 1.0364,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045805698485173e-05,
+      "loss": 0.9971,
+      "step": 3231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043858743256737e-05,
+      "loss": 0.9532,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041911257940625e-05,
+      "loss": 1.0026,
+      "step": 3233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039963242683405e-05,
+      "loss": 1.0231,
+      "step": 3234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703801469763167e-05,
+      "loss": 1.0176,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036065622932067e-05,
+      "loss": 0.9727,
+      "step": 3236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703411601873126e-05,
+      "loss": 0.9914,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032165885175983e-05,
+      "loss": 1.0238,
+      "step": 3238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703021522241298e-05,
+      "loss": 0.984,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702826403058906e-05,
+      "loss": 1.0258,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702631230985105e-05,
+      "loss": 1.0301,
+      "step": 3241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024360060345832e-05,
+      "loss": 0.9603,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702240728222033e-05,
+      "loss": 1.0025,
+      "step": 3243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020453975621492e-05,
+      "loss": 0.9829,
+      "step": 3244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018500140696316e-05,
+      "loss": 1.0224,
+      "step": 3245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701654577759184e-05,
+      "loss": 1.0049,
+      "step": 3246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701459088645514e-05,
+      "loss": 0.9777,
+      "step": 3247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012635467433332e-05,
+      "loss": 0.9793,
+      "step": 3248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701067952067357e-05,
+      "loss": 0.9749,
+      "step": 3249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700872304632305e-05,
+      "loss": 1.0147,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700676604452901e-05,
+      "loss": 1.0219,
+      "step": 3251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004808515438727e-05,
+      "loss": 1.0071,
+      "step": 3252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002850459199508e-05,
+      "loss": 0.9594,
+      "step": 3253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000891875958706e-05,
+      "loss": 1.0042,
+      "step": 3254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699893276586372e-05,
+      "loss": 0.9956,
+      "step": 3255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996973129061983e-05,
+      "loss": 1.0237,
+      "step": 3256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699501296570097e-05,
+      "loss": 0.9614,
+      "step": 3257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699305227592819e-05,
+      "loss": 1.0284,
+      "step": 3258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991091059891195e-05,
+      "loss": 0.9421,
+      "step": 3259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698912931773758e-05,
+      "loss": 1.0015,
+      "step": 3260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698716704961497e-05,
+      "loss": 0.9924,
+      "step": 3261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985204255671043e-05,
+      "loss": 1.0155,
+      "step": 3262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983240936053508e-05,
+      "loss": 0.9503,
+      "step": 3263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981277090910115e-05,
+      "loss": 0.9706,
+      "step": 3264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979312720388655e-05,
+      "loss": 1.0154,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977347824636952e-05,
+      "loss": 0.9782,
+      "step": 3266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697538240380288e-05,
+      "loss": 0.9594,
+      "step": 3267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973416458034347e-05,
+      "loss": 0.9938,
+      "step": 3268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971449987479296e-05,
+      "loss": 1.0192,
+      "step": 3269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696948299228572e-05,
+      "loss": 0.9796,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967515472601644e-05,
+      "loss": 1.0073,
+      "step": 3271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965547428575135e-05,
+      "loss": 1.0123,
+      "step": 3272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6963578860354297e-05,
+      "loss": 1.0542,
+      "step": 3273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961609768087273e-05,
+      "loss": 1.0188,
+      "step": 3274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959640151922253e-05,
+      "loss": 0.9799,
+      "step": 3275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957670012007458e-05,
+      "loss": 1.0153,
+      "step": 3276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955699348491152e-05,
+      "loss": 0.9881,
+      "step": 3277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953728161521638e-05,
+      "loss": 0.9619,
+      "step": 3278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951756451247257e-05,
+      "loss": 1.0103,
+      "step": 3279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949784217816394e-05,
+      "loss": 0.9797,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947811461377468e-05,
+      "loss": 0.9947,
+      "step": 3281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945838182078938e-05,
+      "loss": 3.3027,
+      "step": 3282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943864380069305e-05,
+      "loss": 1.0555,
+      "step": 3283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941890055497105e-05,
+      "loss": 1.0387,
+      "step": 3284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693991520851092e-05,
+      "loss": 0.9625,
+      "step": 3285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937939839259373e-05,
+      "loss": 0.9859,
+      "step": 3286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693596394789111e-05,
+      "loss": 1.0088,
+      "step": 3287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693398753455483e-05,
+      "loss": 0.9763,
+      "step": 3288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932010599399274e-05,
+      "loss": 1.0125,
+      "step": 3289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693003314257322e-05,
+      "loss": 1.0427,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928055164225467e-05,
+      "loss": 0.9963,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926076664504877e-05,
+      "loss": 0.9301,
+      "step": 3292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924097643560347e-05,
+      "loss": 1.0085,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69221181015408e-05,
+      "loss": 0.9909,
+      "step": 3294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920138038595214e-05,
+      "loss": 0.9629,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918157454872596e-05,
+      "loss": 0.9566,
+      "step": 3296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916176350521997e-05,
+      "loss": 0.9027,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914194725692505e-05,
+      "loss": 3.3197,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912212580533248e-05,
+      "loss": 0.9607,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691022991519339e-05,
+      "loss": 1.0247,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908246729822142e-05,
+      "loss": 0.9736,
+      "step": 3301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906263024568746e-05,
+      "loss": 1.0275,
+      "step": 3302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904278799582487e-05,
+      "loss": 0.9901,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902294055012685e-05,
+      "loss": 0.9767,
+      "step": 3304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690030879100871e-05,
+      "loss": 0.9771,
+      "step": 3305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689832300771996e-05,
+      "loss": 0.9738,
+      "step": 3306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896336705295873e-05,
+      "loss": 1.0227,
+      "step": 3307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894349883885934e-05,
+      "loss": 0.9846,
+      "step": 3308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892362543639654e-05,
+      "loss": 1.0514,
+      "step": 3309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68903746847066e-05,
+      "loss": 0.9829,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888386307236364e-05,
+      "loss": 1.0035,
+      "step": 3311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688639741137858e-05,
+      "loss": 0.9599,
+      "step": 3312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688440799728293e-05,
+      "loss": 1.0068,
+      "step": 3313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882418065099122e-05,
+      "loss": 1.0229,
+      "step": 3314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880427614976907e-05,
+      "loss": 0.9737,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878436647066086e-05,
+      "loss": 1.0402,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687644516151648e-05,
+      "loss": 0.9573,
+      "step": 3317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874453158477964e-05,
+      "loss": 0.9893,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872460638100445e-05,
+      "loss": 1.0177,
+      "step": 3319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870467600533875e-05,
+      "loss": 1.0061,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868474045928236e-05,
+      "loss": 0.9884,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866479974433554e-05,
+      "loss": 0.9725,
+      "step": 3322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864485386199895e-05,
+      "loss": 0.9803,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862490281377358e-05,
+      "loss": 0.9754,
+      "step": 3324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860494660116092e-05,
+      "loss": 0.9755,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858498522566273e-05,
+      "loss": 0.9633,
+      "step": 3326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856501868878127e-05,
+      "loss": 1.0108,
+      "step": 3327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854504699201903e-05,
+      "loss": 0.9707,
+      "step": 3328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852507013687908e-05,
+      "loss": 1.004,
+      "step": 3329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685050881248647e-05,
+      "loss": 0.9917,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684851009574797e-05,
+      "loss": 0.9958,
+      "step": 3331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846510863622826e-05,
+      "loss": 0.9823,
+      "step": 3332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684451111626148e-05,
+      "loss": 1.0063,
+      "step": 3333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842510853814433e-05,
+      "loss": 0.9959,
+      "step": 3334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840510076432208e-05,
+      "loss": 0.9952,
+      "step": 3335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683850878426538e-05,
+      "loss": 0.9886,
+      "step": 3336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683650697746455e-05,
+      "loss": 0.9751,
+      "step": 3337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834504656180372e-05,
+      "loss": 1.0015,
+      "step": 3338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832501820563528e-05,
+      "loss": 0.9434,
+      "step": 3339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830498470764746e-05,
+      "loss": 1.0042,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682849460693478e-05,
+      "loss": 0.9947,
+      "step": 3341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826490229224437e-05,
+      "loss": 0.9967,
+      "step": 3342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6824485337784556e-05,
+      "loss": 1.0107,
+      "step": 3343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822479932766017e-05,
+      "loss": 1.0217,
+      "step": 3344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820474014319737e-05,
+      "loss": 0.9727,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681846758259667e-05,
+      "loss": 0.9652,
+      "step": 3346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816460637747814e-05,
+      "loss": 1.0296,
+      "step": 3347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68144531799242e-05,
+      "loss": 1.0288,
+      "step": 3348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812445209276896e-05,
+      "loss": 0.9491,
+      "step": 3349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810436725957017e-05,
+      "loss": 0.9843,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808427730115716e-05,
+      "loss": 0.9392,
+      "step": 3351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806418221904172e-05,
+      "loss": 3.548,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804408201473617e-05,
+      "loss": 1.0241,
+      "step": 3353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680239766897531e-05,
+      "loss": 0.9549,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800386624560565e-05,
+      "loss": 1.0106,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798375068380712e-05,
+      "loss": 0.9384,
+      "step": 3356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796363000587136e-05,
+      "loss": 0.994,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794350421331257e-05,
+      "loss": 0.983,
+      "step": 3358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792337330764533e-05,
+      "loss": 1.0123,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790323729038456e-05,
+      "loss": 1.0101,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678830961630456e-05,
+      "loss": 1.012,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678629499271442e-05,
+      "loss": 0.952,
+      "step": 3362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678427985841965e-05,
+      "loss": 0.9826,
+      "step": 3363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782264213571892e-05,
+      "loss": 1.0066,
+      "step": 3364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678024805832284e-05,
+      "loss": 0.994,
+      "step": 3365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778231392824217e-05,
+      "loss": 1.0001,
+      "step": 3366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677621421722779e-05,
+      "loss": 0.9947,
+      "step": 3367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677419653168536e-05,
+      "loss": 0.9968,
+      "step": 3368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772178336348775e-05,
+      "loss": 1.126,
+      "step": 3369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770159631369908e-05,
+      "loss": 0.9921,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768140416900678e-05,
+      "loss": 0.9759,
+      "step": 3371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676612069309304e-05,
+      "loss": 1.0143,
+      "step": 3372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764100460098994e-05,
+      "loss": 0.9448,
+      "step": 3373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676207971807057e-05,
+      "loss": 1.0132,
+      "step": 3374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760058467159844e-05,
+      "loss": 0.9361,
+      "step": 3375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758036707518916e-05,
+      "loss": 1.024,
+      "step": 3376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756014439299945e-05,
+      "loss": 0.9588,
+      "step": 3377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675399166265511e-05,
+      "loss": 0.9618,
+      "step": 3378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675196837773664e-05,
+      "loss": 3.447,
+      "step": 3379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749944584696795e-05,
+      "loss": 0.9847,
+      "step": 3380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674792028368788e-05,
+      "loss": 0.9521,
+      "step": 3381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745895474862227e-05,
+      "loss": 1.0131,
+      "step": 3382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743870158372222e-05,
+      "loss": 0.9983,
+      "step": 3383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741844334370278e-05,
+      "loss": 0.9903,
+      "step": 3384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739818003008842e-05,
+      "loss": 1.0114,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737791164440418e-05,
+      "loss": 0.9675,
+      "step": 3386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673576381881753e-05,
+      "loss": 0.979,
+      "step": 3387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733735966292745e-05,
+      "loss": 0.9726,
+      "step": 3388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673170760701867e-05,
+      "loss": 0.9589,
+      "step": 3389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729678741147957e-05,
+      "loss": 3.3549,
+      "step": 3390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672764936883328e-05,
+      "loss": 3.4441,
+      "step": 3391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725619490227365e-05,
+      "loss": 0.9937,
+      "step": 3392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723589105482968e-05,
+      "loss": 0.9613,
+      "step": 3393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672155821475289e-05,
+      "loss": 0.9926,
+      "step": 3394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671952681818996e-05,
+      "loss": 0.9749,
+      "step": 3395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671749491594706e-05,
+      "loss": 0.99,
+      "step": 3396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715462508177094e-05,
+      "loss": 1.0498,
+      "step": 3397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713429595033016e-05,
+      "loss": 0.9756,
+      "step": 3398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711396176667806e-05,
+      "loss": 0.9866,
+      "step": 3399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709362253234502e-05,
+      "loss": 0.9472,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670732782488616e-05,
+      "loss": 1.0021,
+      "step": 3401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705292891775876e-05,
+      "loss": 3.3209,
+      "step": 3402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67032574540568e-05,
+      "loss": 0.9918,
+      "step": 3403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701221511882108e-05,
+      "loss": 1.0146,
+      "step": 3404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699185065405008e-05,
+      "loss": 0.9749,
+      "step": 3405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669714811477876e-05,
+      "loss": 0.991,
+      "step": 3406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695110660156652e-05,
+      "loss": 0.9693,
+      "step": 3407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693072701692015e-05,
+      "loss": 0.9949,
+      "step": 3408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691034239538217e-05,
+      "loss": 0.9973,
+      "step": 3409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688995273848664e-05,
+      "loss": 0.9705,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668695580477679e-05,
+      "loss": 0.9881,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6684915832476088e-05,
+      "loss": 0.9791,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668287535710007e-05,
+      "loss": 0.9755,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680834378802293e-05,
+      "loss": 1.0168,
+      "step": 3414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667879289773635e-05,
+      "loss": 0.9879,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6676750914055882e-05,
+      "loss": 0.987,
+      "step": 3416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674708427914547e-05,
+      "loss": 0.9708,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667266543946606e-05,
+      "loss": 1.0106,
+      "step": 3418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670621948864166e-05,
+      "loss": 0.9667,
+      "step": 3419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666857795626265e-05,
+      "loss": 0.9732,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666533461815327e-05,
+      "loss": 0.9946,
+      "step": 3421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666448846567606e-05,
+      "loss": 1.0241,
+      "step": 3422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666244296799875e-05,
+      "loss": 0.9662,
+      "step": 3423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660396968937322e-05,
+      "loss": 1.0049,
+      "step": 3424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665835046864576e-05,
+      "loss": 0.9686,
+      "step": 3425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656303467278065e-05,
+      "loss": 0.9503,
+      "step": 3426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654255964988283e-05,
+      "loss": 0.966,
+      "step": 3427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652207961930505e-05,
+      "loss": 0.9927,
+      "step": 3428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650159458258858e-05,
+      "loss": 1.0056,
+      "step": 3429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648110454127494e-05,
+      "loss": 1.0155,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646060949690613e-05,
+      "loss": 0.9957,
+      "step": 3431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644010945102455e-05,
+      "loss": 0.9517,
+      "step": 3432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664196044051729e-05,
+      "loss": 0.9998,
+      "step": 3433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639909436089435e-05,
+      "loss": 0.9986,
+      "step": 3434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637857931973233e-05,
+      "loss": 0.9588,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663580592832307e-05,
+      "loss": 3.5202,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633753425293377e-05,
+      "loss": 0.9711,
+      "step": 3437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631700423038607e-05,
+      "loss": 0.9836,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629646921713267e-05,
+      "loss": 1.025,
+      "step": 3439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627592921471888e-05,
+      "loss": 1.0084,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662553842246905e-05,
+      "loss": 0.9594,
+      "step": 3441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623483424859357e-05,
+      "loss": 1.0014,
+      "step": 3442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621427928797468e-05,
+      "loss": 1.0462,
+      "step": 3443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619371934438063e-05,
+      "loss": 0.9899,
+      "step": 3444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617315441935868e-05,
+      "loss": 1.0113,
+      "step": 3445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661525845144565e-05,
+      "loss": 1.0489,
+      "step": 3446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66132009631222e-05,
+      "loss": 0.9675,
+      "step": 3447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661114297712036e-05,
+      "loss": 1.0099,
+      "step": 3448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609084493595002e-05,
+      "loss": 0.9539,
+      "step": 3449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660702551270104e-05,
+      "loss": 0.9435,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604966034593423e-05,
+      "loss": 0.9781,
+      "step": 3451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602906059427137e-05,
+      "loss": 0.9925,
+      "step": 3452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600845587357205e-05,
+      "loss": 0.999,
+      "step": 3453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598784618538692e-05,
+      "loss": 0.9755,
+      "step": 3454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659672315312669e-05,
+      "loss": 0.9587,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594661191276342e-05,
+      "loss": 0.9553,
+      "step": 3456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659259873314282e-05,
+      "loss": 0.9909,
+      "step": 3457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590535778881332e-05,
+      "loss": 1.027,
+      "step": 3458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588472328647126e-05,
+      "loss": 1.0146,
+      "step": 3459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6586408382595495e-05,
+      "loss": 1.0365,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584343940881752e-05,
+      "loss": 0.9875,
+      "step": 3461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582279003661264e-05,
+      "loss": 1.0363,
+      "step": 3462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580213571089427e-05,
+      "loss": 0.9597,
+      "step": 3463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657814764332167e-05,
+      "loss": 1.0452,
+      "step": 3464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576081220513474e-05,
+      "loss": 0.9302,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574014302820343e-05,
+      "loss": 1.0273,
+      "step": 3466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6571946890397828e-05,
+      "loss": 1.0383,
+      "step": 3467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569878983401503e-05,
+      "loss": 3.4335,
+      "step": 3468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567810581987002e-05,
+      "loss": 0.9881,
+      "step": 3469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565741686309972e-05,
+      "loss": 1.0215,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563672296526115e-05,
+      "loss": 1.0364,
+      "step": 3471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656160241279116e-05,
+      "loss": 1.0092,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559532035260878e-05,
+      "loss": 1.0036,
+      "step": 3473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557461164091077e-05,
+      "loss": 0.9648,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65553897994376e-05,
+      "loss": 0.9914,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6553317941456327e-05,
+      "loss": 1.0188,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655124559030318e-05,
+      "loss": 0.9693,
+      "step": 3477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654917274613411e-05,
+      "loss": 0.9624,
+      "step": 3478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654709940910511e-05,
+      "loss": 0.994,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545025579372213e-05,
+      "loss": 0.9947,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542951257091484e-05,
+      "loss": 0.9723,
+      "step": 3481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540876442419024e-05,
+      "loss": 0.9974,
+      "step": 3482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538801135510978e-05,
+      "loss": 0.9767,
+      "step": 3483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653672533652352e-05,
+      "loss": 3.3939,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653464904561287e-05,
+      "loss": 0.9781,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532572262935275e-05,
+      "loss": 0.9671,
+      "step": 3486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530494988647024e-05,
+      "loss": 0.9887,
+      "step": 3487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528417222904444e-05,
+      "loss": 0.9522,
+      "step": 3488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526338965863903e-05,
+      "loss": 3.1405,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524260217681793e-05,
+      "loss": 0.9461,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522180978514556e-05,
+      "loss": 0.9624,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652010124851866e-05,
+      "loss": 0.9794,
+      "step": 3492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518021027850624e-05,
+      "loss": 1.0085,
+      "step": 3493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651594031666699e-05,
+      "loss": 0.9729,
+      "step": 3494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651385911512434e-05,
+      "loss": 0.9882,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511777423379306e-05,
+      "loss": 1.0039,
+      "step": 3496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509695241588533e-05,
+      "loss": 1.0119,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507612569908724e-05,
+      "loss": 0.9396,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650552940849661e-05,
+      "loss": 1.0351,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650344575750896e-05,
+      "loss": 0.9904,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501361617102582e-05,
+      "loss": 0.9953,
+      "step": 3501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649927698743431e-05,
+      "loss": 0.9935,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497191868661033e-05,
+      "loss": 0.9685,
+      "step": 3503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495106260939664e-05,
+      "loss": 1.038,
+      "step": 3504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493020164427155e-05,
+      "loss": 1.0046,
+      "step": 3505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490933579280498e-05,
+      "loss": 1.0055,
+      "step": 3506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488846505656714e-05,
+      "loss": 0.9577,
+      "step": 3507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486758943712872e-05,
+      "loss": 0.9677,
+      "step": 3508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648467089360607e-05,
+      "loss": 0.9926,
+      "step": 3509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482582355493448e-05,
+      "loss": 0.9691,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480493329532178e-05,
+      "loss": 1.005,
+      "step": 3511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478403815879468e-05,
+      "loss": 0.9898,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647631381469256e-05,
+      "loss": 0.8998,
+      "step": 3513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474223326128747e-05,
+      "loss": 0.9845,
+      "step": 3514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647213235034535e-05,
+      "loss": 3.2763,
+      "step": 3515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647004088749972e-05,
+      "loss": 0.9466,
+      "step": 3516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646794893774925e-05,
+      "loss": 1.0002,
+      "step": 3517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465856501251378e-05,
+      "loss": 1.0144,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463763578163563e-05,
+      "loss": 0.9888,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646167016864331e-05,
+      "loss": 0.9281,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459576272848162e-05,
+      "loss": 1.0027,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457481890935696e-05,
+      "loss": 1.0187,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6455387023063524e-05,
+      "loss": 0.991,
+      "step": 3523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453291669389294e-05,
+      "loss": 0.9463,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645119583007069e-05,
+      "loss": 0.9774,
+      "step": 3525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449099505265448e-05,
+      "loss": 0.9378,
+      "step": 3526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447002695131313e-05,
+      "loss": 0.9867,
+      "step": 3527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444905399826086e-05,
+      "loss": 1.0107,
+      "step": 3528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442807619507602e-05,
+      "loss": 1.0267,
+      "step": 3529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440709354333727e-05,
+      "loss": 0.9858,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438610604462372e-05,
+      "loss": 0.9853,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643651137005147e-05,
+      "loss": 0.9796,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434411651259008e-05,
+      "loss": 1.0088,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432311448242994e-05,
+      "loss": 0.9786,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430210761161487e-05,
+      "loss": 1.0056,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428109590172568e-05,
+      "loss": 0.9191,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642600793543436e-05,
+      "loss": 0.9925,
+      "step": 3537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423905797105035e-05,
+      "loss": 1.0276,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642180317534278e-05,
+      "loss": 0.9945,
+      "step": 3539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641970007030583e-05,
+      "loss": 1.0043,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417596482152457e-05,
+      "loss": 1.0036,
+      "step": 3541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415492411040963e-05,
+      "loss": 1.0121,
+      "step": 3542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413387857129697e-05,
+      "loss": 0.9797,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411282820577033e-05,
+      "loss": 0.9701,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409177301541387e-05,
+      "loss": 1.0075,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640707130018121e-05,
+      "loss": 0.9594,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404964816654993e-05,
+      "loss": 0.9781,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402857851121258e-05,
+      "loss": 1.0449,
+      "step": 3548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640075040373856e-05,
+      "loss": 1.0021,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398642474665504e-05,
+      "loss": 1.0102,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639653406406072e-05,
+      "loss": 0.9919,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6394425172082877e-05,
+      "loss": 0.948,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392315798890678e-05,
+      "loss": 0.9962,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390205944642864e-05,
+      "loss": 0.9988,
+      "step": 3554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388095609498217e-05,
+      "loss": 0.9444,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638598479361555e-05,
+      "loss": 0.9749,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383873497153713e-05,
+      "loss": 1.0111,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381761720271586e-05,
+      "loss": 1.0247,
+      "step": 3558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379649463128104e-05,
+      "loss": 0.9623,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377536725882213e-05,
+      "loss": 1.0215,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375423508692913e-05,
+      "loss": 0.9567,
+      "step": 3561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373309811719235e-05,
+      "loss": 0.9596,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6371195635120245e-05,
+      "loss": 0.9505,
+      "step": 3563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369080979055047e-05,
+      "loss": 0.963,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636696584368278e-05,
+      "loss": 0.9952,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636485022916262e-05,
+      "loss": 0.9422,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636273413565377e-05,
+      "loss": 1.038,
+      "step": 3567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636061756331549e-05,
+      "loss": 0.923,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358500512307057e-05,
+      "loss": 0.9886,
+      "step": 3569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356382982787787e-05,
+      "loss": 0.9555,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635426497491704e-05,
+      "loss": 3.3049,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635214648885421e-05,
+      "loss": 0.9908,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350027524758722e-05,
+      "loss": 1.0051,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347908082790032e-05,
+      "loss": 0.9629,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345788163107645e-05,
+      "loss": 1.0129,
+      "step": 3575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343667765871098e-05,
+      "loss": 0.9712,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634154689123996e-05,
+      "loss": 1.0105,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633942553937384e-05,
+      "loss": 1.0351,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337303710432376e-05,
+      "loss": 0.9717,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335181404575252e-05,
+      "loss": 0.9957,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333058621962185e-05,
+      "loss": 0.9907,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330935362752914e-05,
+      "loss": 0.9686,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328811627107238e-05,
+      "loss": 0.9876,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632668741518497e-05,
+      "loss": 0.9976,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632456272714598e-05,
+      "loss": 0.9963,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322437563150152e-05,
+      "loss": 0.9847,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320311923357416e-05,
+      "loss": 0.9795,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631818580792774e-05,
+      "loss": 1.0147,
+      "step": 3588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316059217021126e-05,
+      "loss": 0.9873,
+      "step": 3589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313932150797615e-05,
+      "loss": 0.9913,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311804609417267e-05,
+      "loss": 0.9833,
+      "step": 3591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309676593040206e-05,
+      "loss": 1.0207,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630754810182657e-05,
+      "loss": 3.3671,
+      "step": 3593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305419135936537e-05,
+      "loss": 0.979,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303289695530327e-05,
+      "loss": 0.9858,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630115978076819e-05,
+      "loss": 0.9754,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629902939181041e-05,
+      "loss": 1.012,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629689852881732e-05,
+      "loss": 0.953,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629476719194927e-05,
+      "loss": 0.9855,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629263538136665e-05,
+      "loss": 0.9958,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6290503097229903e-05,
+      "loss": 0.9979,
+      "step": 3601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288370339699486e-05,
+      "loss": 1.001,
+      "step": 3602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62862371089359e-05,
+      "loss": 0.9533,
+      "step": 3603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284103405099687e-05,
+      "loss": 1.023,
+      "step": 3604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628196922835142e-05,
+      "loss": 1.0074,
+      "step": 3605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62798345788517e-05,
+      "loss": 1.0088,
+      "step": 3606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6277699456761174e-05,
+      "loss": 1.0131,
+      "step": 3607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275563862240522e-05,
+      "loss": 0.9531,
+      "step": 3608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627342779545046e-05,
+      "loss": 0.9428,
+      "step": 3609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271291256551734e-05,
+      "loss": 0.972,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626915424570514e-05,
+      "loss": 0.9841,
+      "step": 3611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267016763071486e-05,
+      "loss": 0.969,
+      "step": 3612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264878808811636e-05,
+      "loss": 0.9868,
+      "step": 3613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262740383086485e-05,
+      "loss": 3.1882,
+      "step": 3614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260601486056952e-05,
+      "loss": 0.9824,
+      "step": 3615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258462117884008e-05,
+      "loss": 0.9539,
+      "step": 3616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256322278728652e-05,
+      "loss": 0.9503,
+      "step": 3617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254181968751916e-05,
+      "loss": 1.0034,
+      "step": 3618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252041188114866e-05,
+      "loss": 0.9826,
+      "step": 3619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249899936978613e-05,
+      "loss": 1.0111,
+      "step": 3620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247758215504293e-05,
+      "loss": 0.9607,
+      "step": 3621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245616023853084e-05,
+      "loss": 0.9926,
+      "step": 3622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62434733621862e-05,
+      "loss": 0.9776,
+      "step": 3623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624133023066488e-05,
+      "loss": 3.2942,
+      "step": 3624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239186629450414e-05,
+      "loss": 0.9591,
+      "step": 3625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237042558704113e-05,
+      "loss": 0.9868,
+      "step": 3626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.0052,
+      "step": 3627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232753009261468e-05,
+      "loss": 1.0529,
+      "step": 3628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623060753088793e-05,
+      "loss": 0.9195,
+      "step": 3629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228461583628187e-05,
+      "loss": 3.2645,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226315167643723e-05,
+      "loss": 0.9898,
+      "step": 3631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224168283096075e-05,
+      "loss": 0.9861,
+      "step": 3632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62220209301468e-05,
+      "loss": 0.9767,
+      "step": 3633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6219873108957514e-05,
+      "loss": 3.2257,
+      "step": 3634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217724819689834e-05,
+      "loss": 0.9588,
+      "step": 3635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215576062505436e-05,
+      "loss": 1.0055,
+      "step": 3636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213426837566027e-05,
+      "loss": 0.9827,
+      "step": 3637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211277145033347e-05,
+      "loss": 0.995,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209126985069175e-05,
+      "loss": 0.992,
+      "step": 3639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206976357835316e-05,
+      "loss": 0.9923,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204825263493616e-05,
+      "loss": 0.9911,
+      "step": 3641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620267370220596e-05,
+      "loss": 1.0146,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200521674134264e-05,
+      "loss": 0.9868,
+      "step": 3643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619836917944048e-05,
+      "loss": 0.9754,
+      "step": 3644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619621621828659e-05,
+      "loss": 0.9915,
+      "step": 3645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194062790834623e-05,
+      "loss": 0.9849,
+      "step": 3646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619190889724663e-05,
+      "loss": 1.0383,
+      "step": 3647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189754537684707e-05,
+      "loss": 0.9809,
+      "step": 3648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187599712310973e-05,
+      "loss": 1.0155,
+      "step": 3649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61854444212876e-05,
+      "loss": 1.034,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183288664776782e-05,
+      "loss": 0.9668,
+      "step": 3651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618113244294075e-05,
+      "loss": 0.9881,
+      "step": 3652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178975755941767e-05,
+      "loss": 0.9825,
+      "step": 3653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617681860394214e-05,
+      "loss": 0.9818,
+      "step": 3654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174660987104207e-05,
+      "loss": 1.0208,
+      "step": 3655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172502905590338e-05,
+      "loss": 0.9902,
+      "step": 3656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617034435956294e-05,
+      "loss": 1.0162,
+      "step": 3657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168185349184455e-05,
+      "loss": 1.0238,
+      "step": 3658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616602587461736e-05,
+      "loss": 0.9648,
+      "step": 3659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616386593602417e-05,
+      "loss": 0.9178,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161705533567428e-05,
+      "loss": 0.955,
+      "step": 3661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159544667409718e-05,
+      "loss": 3.2138,
+      "step": 3662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615738333771365e-05,
+      "loss": 1.0102,
+      "step": 3663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155221544641885e-05,
+      "loss": 0.9505,
+      "step": 3664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153059288357104e-05,
+      "loss": 3.1943,
+      "step": 3665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150896569022033e-05,
+      "loss": 1.0021,
+      "step": 3666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148733386799422e-05,
+      "loss": 0.9328,
+      "step": 3667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146569741852066e-05,
+      "loss": 0.9813,
+      "step": 3668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144405634342786e-05,
+      "loss": 0.9868,
+      "step": 3669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614224106443445e-05,
+      "loss": 0.9617,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6140076032289945e-05,
+      "loss": 1.0325,
+      "step": 3671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613791053807221e-05,
+      "loss": 0.9746,
+      "step": 3672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61357445819442e-05,
+      "loss": 1.0064,
+      "step": 3673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613357816406892e-05,
+      "loss": 0.9391,
+      "step": 3674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613141128460941e-05,
+      "loss": 0.9839,
+      "step": 3675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129243943728733e-05,
+      "loss": 0.9885,
+      "step": 3676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127076141589988e-05,
+      "loss": 3.0751,
+      "step": 3677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124907878356322e-05,
+      "loss": 0.947,
+      "step": 3678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122739154190904e-05,
+      "loss": 0.9351,
+      "step": 3679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120569969256947e-05,
+      "loss": 0.938,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118400323717684e-05,
+      "loss": 0.9859,
+      "step": 3681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61162302177364e-05,
+      "loss": 0.8969,
+      "step": 3682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114059651476404e-05,
+      "loss": 0.9488,
+      "step": 3683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611188862510105e-05,
+      "loss": 0.9466,
+      "step": 3684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610971713877371e-05,
+      "loss": 0.9588,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107545192657797e-05,
+      "loss": 1.0144,
+      "step": 3686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105372786916776e-05,
+      "loss": 0.9735,
+      "step": 3687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610319992171412e-05,
+      "loss": 1.0004,
+      "step": 3688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610102659721335e-05,
+      "loss": 0.9218,
+      "step": 3689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098852813578024e-05,
+      "loss": 3.1859,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096678570971728e-05,
+      "loss": 0.9519,
+      "step": 3691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094503869558086e-05,
+      "loss": 0.9561,
+      "step": 3692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092328709500755e-05,
+      "loss": 0.9871,
+      "step": 3693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090153090963426e-05,
+      "loss": 0.9737,
+      "step": 3694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087977014109837e-05,
+      "loss": 0.9799,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085800479103734e-05,
+      "loss": 3.3003,
+      "step": 3696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083623486108918e-05,
+      "loss": 1.0085,
+      "step": 3697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081446035289223e-05,
+      "loss": 0.9758,
+      "step": 3698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607926812680851e-05,
+      "loss": 1.0042,
+      "step": 3699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607708976083068e-05,
+      "loss": 0.9885,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074910937519663e-05,
+      "loss": 1.0062,
+      "step": 3701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072731657039433e-05,
+      "loss": 0.9907,
+      "step": 3702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070551919553988e-05,
+      "loss": 0.9607,
+      "step": 3703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068371725227367e-05,
+      "loss": 0.9143,
+      "step": 3704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066191074223638e-05,
+      "loss": 0.985,
+      "step": 3705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064009966706914e-05,
+      "loss": 0.998,
+      "step": 3706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606182840284132e-05,
+      "loss": 0.964,
+      "step": 3707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605964638279105e-05,
+      "loss": 0.9688,
+      "step": 3708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057463906720295e-05,
+      "loss": 0.9897,
+      "step": 3709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605528097479331e-05,
+      "loss": 0.9558,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605309758717436e-05,
+      "loss": 0.9003,
+      "step": 3711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050913744027772e-05,
+      "loss": 0.9687,
+      "step": 3712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604872944551788e-05,
+      "loss": 1.0042,
+      "step": 3713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604654469180907e-05,
+      "loss": 0.9805,
+      "step": 3714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604435948306575e-05,
+      "loss": 0.9622,
+      "step": 3715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042173819452375e-05,
+      "loss": 1.0268,
+      "step": 3716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603998770113343e-05,
+      "loss": 0.986,
+      "step": 3717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603780112827342e-05,
+      "loss": 3.1177,
+      "step": 3718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035614101036903e-05,
+      "loss": 0.981,
+      "step": 3719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603342661958847e-05,
+      "loss": 0.986,
+      "step": 3720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603123868409273e-05,
+      "loss": 0.9782,
+      "step": 3721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602905029471435e-05,
+      "loss": 0.9894,
+      "step": 3722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026861451618002e-05,
+      "loss": 0.9992,
+      "step": 3723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602467215496842e-05,
+      "loss": 0.9881,
+      "step": 3724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022482404930353e-05,
+      "loss": 1.0093,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60202922016686e-05,
+      "loss": 0.9596,
+      "step": 3726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6018101545347975e-05,
+      "loss": 1.015,
+      "step": 3727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015910436133345e-05,
+      "loss": 0.9728,
+      "step": 3728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013718874189594e-05,
+      "loss": 0.9309,
+      "step": 3729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011526859681658e-05,
+      "loss": 0.96,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009334392774493e-05,
+      "loss": 0.9798,
+      "step": 3731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600714147363309e-05,
+      "loss": 0.9517,
+      "step": 3732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004948102422483e-05,
+      "loss": 0.9953,
+      "step": 3733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002754279307738e-05,
+      "loss": 0.9456,
+      "step": 3734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000560004453944e-05,
+      "loss": 0.9992,
+      "step": 3735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998365278026232e-05,
+      "loss": 0.9746,
+      "step": 3736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996170100189777e-05,
+      "loss": 0.9477,
+      "step": 3737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993974471109766e-05,
+      "loss": 0.9954,
+      "step": 3738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599177839095144e-05,
+      "loss": 0.9934,
+      "step": 3739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598958185988006e-05,
+      "loss": 0.9477,
+      "step": 3740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987384878060928e-05,
+      "loss": 0.961,
+      "step": 3741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985187445659377e-05,
+      "loss": 0.9944,
+      "step": 3742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982989562840785e-05,
+      "loss": 1.0069,
+      "step": 3743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980791229770538e-05,
+      "loss": 0.9248,
+      "step": 3744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978592446614088e-05,
+      "loss": 0.9989,
+      "step": 3745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976393213536897e-05,
+      "loss": 0.9633,
+      "step": 3746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974193530704466e-05,
+      "loss": 3.2522,
+      "step": 3747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971993398282342e-05,
+      "loss": 1.0138,
+      "step": 3748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596979281643609e-05,
+      "loss": 0.971,
+      "step": 3749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5967591785331315e-05,
+      "loss": 0.9691,
+      "step": 3750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965390305133662e-05,
+      "loss": 0.9791,
+      "step": 3751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963188376008795e-05,
+      "loss": 0.9297,
+      "step": 3752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596098599812243e-05,
+      "loss": 0.9614,
+      "step": 3753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958783171640304e-05,
+      "loss": 0.9918,
+      "step": 3754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595657989672819e-05,
+      "loss": 0.9396,
+      "step": 3755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954376173551896e-05,
+      "loss": 1.005,
+      "step": 3756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952172002277268e-05,
+      "loss": 0.9804,
+      "step": 3757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949967383070177e-05,
+      "loss": 0.9223,
+      "step": 3758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947762316096534e-05,
+      "loss": 0.965,
+      "step": 3759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945556801522282e-05,
+      "loss": 1.0229,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59433508395134e-05,
+      "loss": 1.01,
+      "step": 3761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941144430235887e-05,
+      "loss": 0.9538,
+      "step": 3762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938937573855805e-05,
+      "loss": 1.0348,
+      "step": 3763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936730270539218e-05,
+      "loss": 0.9872,
+      "step": 3764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934522520452244e-05,
+      "loss": 0.9818,
+      "step": 3765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932314323761026e-05,
+      "loss": 0.9416,
+      "step": 3766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930105680631738e-05,
+      "loss": 0.9733,
+      "step": 3767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927896591230596e-05,
+      "loss": 0.9894,
+      "step": 3768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592568705572385e-05,
+      "loss": 1.0002,
+      "step": 3769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923477074277774e-05,
+      "loss": 1.0278,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921266647058683e-05,
+      "loss": 0.9331,
+      "step": 3771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591905577423292e-05,
+      "loss": 0.9821,
+      "step": 3772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916844455966865e-05,
+      "loss": 0.9807,
+      "step": 3773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914632692426936e-05,
+      "loss": 1.0099,
+      "step": 3774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912420483779576e-05,
+      "loss": 0.9774,
+      "step": 3775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591020783019127e-05,
+      "loss": 1.0194,
+      "step": 3776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907994731828526e-05,
+      "loss": 0.955,
+      "step": 3777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905781188857895e-05,
+      "loss": 1.0106,
+      "step": 3778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590356720144596e-05,
+      "loss": 0.9661,
+      "step": 3779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901352769759326e-05,
+      "loss": 0.9298,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589913789396465e-05,
+      "loss": 0.9876,
+      "step": 3781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896922574228613e-05,
+      "loss": 0.9752,
+      "step": 3782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894706810717923e-05,
+      "loss": 0.9649,
+      "step": 3783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892490603599334e-05,
+      "loss": 1.0253,
+      "step": 3784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589027395303962e-05,
+      "loss": 0.9625,
+      "step": 3785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888056859205604e-05,
+      "loss": 0.9417,
+      "step": 3786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588583932226413e-05,
+      "loss": 0.9377,
+      "step": 3787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883621342382083e-05,
+      "loss": 1.0141,
+      "step": 3788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588140291972637e-05,
+      "loss": 0.983,
+      "step": 3789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5879184054463946e-05,
+      "loss": 1.0313,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876964746761792e-05,
+      "loss": 0.9612,
+      "step": 3791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874744996786918e-05,
+      "loss": 0.9942,
+      "step": 3792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5872524804706375e-05,
+      "loss": 1.0036,
+      "step": 3793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870304170687246e-05,
+      "loss": 0.9676,
+      "step": 3794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868083094896643e-05,
+      "loss": 1.0059,
+      "step": 3795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586586157750171e-05,
+      "loss": 0.9692,
+      "step": 3796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5863639618669634e-05,
+      "loss": 1.0002,
+      "step": 3797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586141721856763e-05,
+      "loss": 0.9831,
+      "step": 3798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859194377362942e-05,
+      "loss": 0.9787,
+      "step": 3799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585697109522285e-05,
+      "loss": 0.9763,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854747372314668e-05,
+      "loss": 0.9954,
+      "step": 3801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585252320880575e-05,
+      "loss": 0.9858,
+      "step": 3802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585029860486346e-05,
+      "loss": 0.9895,
+      "step": 3803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584807356065523e-05,
+      "loss": 0.966,
+      "step": 3804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845848076348495e-05,
+      "loss": 0.9388,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584362215211073e-05,
+      "loss": 1.0015,
+      "step": 3806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841395788109463e-05,
+      "loss": 0.9848,
+      "step": 3807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839168984512227e-05,
+      "loss": 1.0511,
+      "step": 3808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836941741486607e-05,
+      "loss": 1.014,
+      "step": 3809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834714059200215e-05,
+      "loss": 0.9962,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832485937820693e-05,
+      "loss": 0.9433,
+      "step": 3811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830257377515716e-05,
+      "loss": 0.9248,
+      "step": 3812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5828028378453e-05,
+      "loss": 0.9659,
+      "step": 3813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582579894080029e-05,
+      "loss": 0.9596,
+      "step": 3814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823569064725356e-05,
+      "loss": 0.984,
+      "step": 3815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821338750396016e-05,
+      "loss": 0.9646,
+      "step": 3816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819107997980112e-05,
+      "loss": 1.0425,
+      "step": 3817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816876807645514e-05,
+      "loss": 1.0455,
+      "step": 3818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814645179560135e-05,
+      "loss": 0.939,
+      "step": 3819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812413113891918e-05,
+      "loss": 0.9546,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810180610808833e-05,
+      "loss": 0.9508,
+      "step": 3821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580794767047889e-05,
+      "loss": 0.9635,
+      "step": 3822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805714293070135e-05,
+      "loss": 0.9682,
+      "step": 3823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803480478750636e-05,
+      "loss": 0.9846,
+      "step": 3824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801246227688497e-05,
+      "loss": 0.9758,
+      "step": 3825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799011540051862e-05,
+      "loss": 1.01,
+      "step": 3826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796776416008897e-05,
+      "loss": 0.9309,
+      "step": 3827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794540855727817e-05,
+      "loss": 3.2372,
+      "step": 3828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5792304859376852e-05,
+      "loss": 0.9827,
+      "step": 3829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790068427124274e-05,
+      "loss": 0.9297,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787831559138388e-05,
+      "loss": 1.0201,
+      "step": 3831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578559425558753e-05,
+      "loss": 1.0022,
+      "step": 3832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578335651664007e-05,
+      "loss": 0.9925,
+      "step": 3833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781118342464402e-05,
+      "loss": 0.9898,
+      "step": 3834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577887973322897e-05,
+      "loss": 1.0046,
+      "step": 3835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5776640689102235e-05,
+      "loss": 0.9706,
+      "step": 3836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774401210252702e-05,
+      "loss": 3.1263,
+      "step": 3837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57721612968489e-05,
+      "loss": 0.9823,
+      "step": 3838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769920949059395e-05,
+      "loss": 0.9748,
+      "step": 3839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767680167052787e-05,
+      "loss": 0.9736,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765438950997704e-05,
+      "loss": 1.004,
+      "step": 3841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576319730106281e-05,
+      "loss": 0.9639,
+      "step": 3842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57609552174168e-05,
+      "loss": 0.966,
+      "step": 3843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575871270022841e-05,
+      "loss": 0.9756,
+      "step": 3844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575646974966639e-05,
+      "loss": 0.9662,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575422636589954e-05,
+      "loss": 0.9559,
+      "step": 3846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751982549096686e-05,
+      "loss": 0.9447,
+      "step": 3847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574973829942669e-05,
+      "loss": 1.0089,
+      "step": 3848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574749361705844e-05,
+      "loss": 1.0147,
+      "step": 3849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745248502160858e-05,
+      "loss": 0.9905,
+      "step": 3850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743002954902905e-05,
+      "loss": 0.9912,
+      "step": 3851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740756975453567e-05,
+      "loss": 0.9181,
+      "step": 3852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573851056398187e-05,
+      "loss": 1.0035,
+      "step": 3853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5736263720656867e-05,
+      "loss": 0.9888,
+      "step": 3854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573401644564764e-05,
+      "loss": 3.091,
+      "step": 3855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731768739123318e-05,
+      "loss": 0.9948,
+      "step": 3856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729520601253046e-05,
+      "loss": 0.9931,
+      "step": 3857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572727203220601e-05,
+      "loss": 1.0113,
+      "step": 3858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572502303215142e-05,
+      "loss": 1.0061,
+      "step": 3859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572277360125854e-05,
+      "loss": 1.005,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720523739696637e-05,
+      "loss": 1.0336,
+      "step": 3861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5718273447635037e-05,
+      "loss": 0.9979,
+      "step": 3862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571602272524307e-05,
+      "loss": 0.9764,
+      "step": 3863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713771572690135e-05,
+      "loss": 0.9876,
+      "step": 3864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571151999014563e-05,
+      "loss": 0.9468,
+      "step": 3865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5709267977779e-05,
+      "loss": 0.9786,
+      "step": 3866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707015535759725e-05,
+      "loss": 0.9737,
+      "step": 3867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570476266425731e-05,
+      "loss": 0.9852,
+      "step": 3868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702509363441298e-05,
+      "loss": 0.9633,
+      "step": 3869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700255633481257e-05,
+      "loss": 0.9743,
+      "step": 3870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698001474546794e-05,
+      "loss": 1.0156,
+      "step": 3871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695746886807553e-05,
+      "loss": 0.9805,
+      "step": 3872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693491870433195e-05,
+      "loss": 0.9425,
+      "step": 3873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691236425593427e-05,
+      "loss": 1.0135,
+      "step": 3874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688980552457982e-05,
+      "loss": 0.9912,
+      "step": 3875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686724251196622e-05,
+      "loss": 0.9712,
+      "step": 3876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684467521979153e-05,
+      "loss": 0.978,
+      "step": 3877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682210364975402e-05,
+      "loss": 0.968,
+      "step": 3878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679952780355233e-05,
+      "loss": 0.9709,
+      "step": 3879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677694768288537e-05,
+      "loss": 1.0008,
+      "step": 3880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675436328945246e-05,
+      "loss": 0.9973,
+      "step": 3881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673177462495322e-05,
+      "loss": 1.0184,
+      "step": 3882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567091816910875e-05,
+      "loss": 0.9754,
+      "step": 3883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566865844895556e-05,
+      "loss": 0.9575,
+      "step": 3884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666398302205798e-05,
+      "loss": 0.9788,
+      "step": 3885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566413772902956e-05,
+      "loss": 0.9748,
+      "step": 3886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661876729596973e-05,
+      "loss": 0.9817,
+      "step": 3887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659615304078173e-05,
+      "loss": 1.0092,
+      "step": 3888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657353452643357e-05,
+      "loss": 1.0041,
+      "step": 3889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655091175462733e-05,
+      "loss": 1.0176,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652828472706554e-05,
+      "loss": 0.9916,
+      "step": 3891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650565344545098e-05,
+      "loss": 0.9922,
+      "step": 3892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648301791148685e-05,
+      "loss": 0.9529,
+      "step": 3893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646037812687645e-05,
+      "loss": 0.9749,
+      "step": 3894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564377340933237e-05,
+      "loss": 0.9399,
+      "step": 3895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641508581253254e-05,
+      "loss": 0.9551,
+      "step": 3896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639243328620745e-05,
+      "loss": 1.0171,
+      "step": 3897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563697765160532e-05,
+      "loss": 0.9747,
+      "step": 3898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563471155037747e-05,
+      "loss": 0.9716,
+      "step": 3899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632445025107743e-05,
+      "loss": 0.9714,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56301780759667e-05,
+      "loss": 0.9422,
+      "step": 3901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627910703124947e-05,
+      "loss": 1.0013,
+      "step": 3902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562564290675311e-05,
+      "loss": 0.906,
+      "step": 3903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623374687021856e-05,
+      "loss": 0.963,
+      "step": 3904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562110604410188e-05,
+      "loss": 1.0039,
+      "step": 3905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561883697816391e-05,
+      "loss": 0.9808,
+      "step": 3906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56165674893787e-05,
+      "loss": 1.0016,
+      "step": 3907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561429757791705e-05,
+      "loss": 0.995,
+      "step": 3908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612027243949774e-05,
+      "loss": 0.9436,
+      "step": 3909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609756487647734e-05,
+      "loss": 1.0069,
+      "step": 3910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607485309181813e-05,
+      "loss": 0.9733,
+      "step": 3911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560521370872293e-05,
+      "loss": 0.9905,
+      "step": 3912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560294168644203e-05,
+      "loss": 0.9555,
+      "step": 3913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600669242510103e-05,
+      "loss": 0.936,
+      "step": 3914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559839637709816e-05,
+      "loss": 1.0052,
+      "step": 3915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559612309037724e-05,
+      "loss": 0.9229,
+      "step": 3916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593849382518426e-05,
+      "loss": 0.9847,
+      "step": 3917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559157525369283e-05,
+      "loss": 0.9292,
+      "step": 3918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558930070407158e-05,
+      "loss": 1.0123,
+      "step": 3919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558702573382586e-05,
+      "loss": 1.019,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584750343126867e-05,
+      "loss": 0.9655,
+      "step": 3921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582474532145838e-05,
+      "loss": 0.9428,
+      "step": 3922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558019830105404e-05,
+      "loss": 0.9685,
+      "step": 3923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557792165002277e-05,
+      "loss": 0.9891,
+      "step": 3924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557564457922336e-05,
+      "loss": 1.02,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573367088827175e-05,
+      "loss": 0.9952,
+      "step": 3926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55710891790056e-05,
+      "loss": 0.9801,
+      "step": 3927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568810849930067e-05,
+      "loss": 0.9688,
+      "step": 3928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566532101772027e-05,
+      "loss": 0.9815,
+      "step": 3929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556425293470297e-05,
+      "loss": 1.0014,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561973348894417e-05,
+      "loss": 0.9705,
+      "step": 3931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559693344517916e-05,
+      "loss": 0.9992,
+      "step": 3932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555741292174505e-05,
+      "loss": 0.9713,
+      "step": 3933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555132080747438e-05,
+      "loss": 0.9384,
+      "step": 3934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5552850821696716e-05,
+      "loss": 0.9911,
+      "step": 3935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555056914476457e-05,
+      "loss": 0.9955,
+      "step": 3936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55482870501227e-05,
+      "loss": 0.9739,
+      "step": 3937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546004537942853e-05,
+      "loss": 1.0358,
+      "step": 3938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55437216083968e-05,
+      "loss": 1.006,
+      "step": 3939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541438261656335e-05,
+      "loss": 0.9389,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539154497893298e-05,
+      "loss": 1.0038,
+      "step": 3941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536870317279557e-05,
+      "loss": 0.9804,
+      "step": 3942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534585719987e-05,
+      "loss": 0.9703,
+      "step": 3943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532300706187567e-05,
+      "loss": 0.95,
+      "step": 3944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530015276053208e-05,
+      "loss": 0.9979,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527729429755915e-05,
+      "loss": 0.9974,
+      "step": 3946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525443167467714e-05,
+      "loss": 0.9855,
+      "step": 3947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523156489360655e-05,
+      "loss": 1.0037,
+      "step": 3948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5520869395606827e-05,
+      "loss": 1.0266,
+      "step": 3949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518581886378338e-05,
+      "loss": 0.9705,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516293961847342e-05,
+      "loss": 0.9805,
+      "step": 3951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514005622186016e-05,
+      "loss": 0.9985,
+      "step": 3952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511716867566567e-05,
+      "loss": 0.9971,
+      "step": 3953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550942769816124e-05,
+      "loss": 0.983,
+      "step": 3954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55071381141423e-05,
+      "loss": 0.9636,
+      "step": 3955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550484811568206e-05,
+      "loss": 1.0211,
+      "step": 3956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502557702952848e-05,
+      "loss": 1.0032,
+      "step": 3957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500266876127033e-05,
+      "loss": 0.9878,
+      "step": 3958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549797563537701e-05,
+      "loss": 1.0097,
+      "step": 3959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495683980875207e-05,
+      "loss": 0.9855,
+      "step": 3960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493391912794083e-05,
+      "loss": 0.972,
+      "step": 3961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491099431306126e-05,
+      "loss": 1.0059,
+      "step": 3962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488806536583862e-05,
+      "loss": 0.9739,
+      "step": 3963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486513228799846e-05,
+      "loss": 0.9726,
+      "step": 3964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548421950812665e-05,
+      "loss": 1.0306,
+      "step": 3965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54819253747369e-05,
+      "loss": 0.9905,
+      "step": 3966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479630828803235e-05,
+      "loss": 0.9625,
+      "step": 3967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477335870498336e-05,
+      "loss": 0.9317,
+      "step": 3968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547504049999491e-05,
+      "loss": 1.0121,
+      "step": 3969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547274471746569e-05,
+      "loss": 0.9313,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470448523083453e-05,
+      "loss": 0.9269,
+      "step": 3971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5468151917021e-05,
+      "loss": 0.9435,
+      "step": 3972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465854899451156e-05,
+      "loss": 0.9288,
+      "step": 3973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546355747054679e-05,
+      "loss": 0.9744,
+      "step": 3974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546125963048079e-05,
+      "loss": 0.9698,
+      "step": 3975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545896137942609e-05,
+      "loss": 0.9829,
+      "step": 3976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456662717555637e-05,
+      "loss": 0.9772,
+      "step": 3977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545436364504242e-05,
+      "loss": 0.9611,
+      "step": 3978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545206416205945e-05,
+      "loss": 0.978,
+      "step": 3979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544976426877979e-05,
+      "loss": 0.9962,
+      "step": 3980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544746396537651e-05,
+      "loss": 0.9794,
+      "step": 3981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445163252022722e-05,
+      "loss": 1.0333,
+      "step": 3982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442862128891564e-05,
+      "loss": 3.0476,
+      "step": 3983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440560596156208e-05,
+      "loss": 0.964,
+      "step": 3984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543825865398986e-05,
+      "loss": 1.0135,
+      "step": 3985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435956302565748e-05,
+      "loss": 0.9285,
+      "step": 3986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543365354205714e-05,
+      "loss": 2.9801,
+      "step": 3987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431350372637335e-05,
+      "loss": 1.0006,
+      "step": 3988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542904679447965e-05,
+      "loss": 0.9779,
+      "step": 3989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426742807757447e-05,
+      "loss": 0.9511,
+      "step": 3990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424438412644113e-05,
+      "loss": 0.9669,
+      "step": 3991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422133609313062e-05,
+      "loss": 0.9729,
+      "step": 3992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419828397937747e-05,
+      "loss": 0.9502,
+      "step": 3993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417522778691645e-05,
+      "loss": 0.975,
+      "step": 3994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415216751748264e-05,
+      "loss": 0.9695,
+      "step": 3995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541291031728115e-05,
+      "loss": 0.9214,
+      "step": 3996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410603475463875e-05,
+      "loss": 1.0169,
+      "step": 3997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408296226470036e-05,
+      "loss": 0.9609,
+      "step": 3998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405988570473266e-05,
+      "loss": 0.9262,
+      "step": 3999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403680507647233e-05,
+      "loss": 0.9409,
+      "step": 4000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401372038165625e-05,
+      "loss": 0.9708,
+      "step": 4001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539906316220217e-05,
+      "loss": 0.9379,
+      "step": 4002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396753879930622e-05,
+      "loss": 0.9183,
+      "step": 4003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394444191524768e-05,
+      "loss": 1.0039,
+      "step": 4004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392134097158424e-05,
+      "loss": 1.0454,
+      "step": 4005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538982359700544e-05,
+      "loss": 0.9784,
+      "step": 4006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387512691239682e-05,
+      "loss": 0.9953,
+      "step": 4007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385201380035073e-05,
+      "loss": 0.9303,
+      "step": 4008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538288966356554e-05,
+      "loss": 0.9796,
+      "step": 4009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538057754200506e-05,
+      "loss": 0.944,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378265015527624e-05,
+      "loss": 0.949,
+      "step": 4011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537595208430727e-05,
+      "loss": 0.9876,
+      "step": 4012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537363874851805e-05,
+      "loss": 0.9705,
+      "step": 4013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371325008334067e-05,
+      "loss": 3.273,
+      "step": 4014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369010863929427e-05,
+      "loss": 0.9696,
+      "step": 4015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536669631547829e-05,
+      "loss": 0.963,
+      "step": 4016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5364381363154842e-05,
+      "loss": 0.9552,
+      "step": 4017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362066007133288e-05,
+      "loss": 0.9601,
+      "step": 4018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535975024758787e-05,
+      "loss": 0.9718,
+      "step": 4019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535743408469287e-05,
+      "loss": 0.9985,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355117518622584e-05,
+      "loss": 0.9696,
+      "step": 4021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535280054955135e-05,
+      "loss": 0.9788,
+      "step": 4022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350483177653528e-05,
+      "loss": 0.9936,
+      "step": 4023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5348165403103516e-05,
+      "loss": 0.9668,
+      "step": 4024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5345847226075736e-05,
+      "loss": 0.9788,
+      "step": 4025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343528646744644e-05,
+      "loss": 0.9696,
+      "step": 4026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534120966528473e-05,
+      "loss": 0.9597,
+      "step": 4027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53388902818705e-05,
+      "loss": 0.9799,
+      "step": 4028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533657049667651e-05,
+      "loss": 0.9754,
+      "step": 4029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5334250309877334e-05,
+      "loss": 0.9793,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533192972164757e-05,
+      "loss": 0.9798,
+      "step": 4031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532960873216186e-05,
+      "loss": 0.9683,
+      "step": 4032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327287341594875e-05,
+      "loss": 1.0114,
+      "step": 4033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532496555012131e-05,
+      "loss": 0.9241,
+      "step": 4034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532264335791589e-05,
+      "loss": 1.0251,
+      "step": 4035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.9689,
+      "step": 4036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531799777200854e-05,
+      "loss": 0.9927,
+      "step": 4037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315674378656215e-05,
+      "loss": 0.958,
+      "step": 4038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531335058527125e-05,
+      "loss": 3.297,
+      "step": 4039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5311026392028512e-05,
+      "loss": 0.9524,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308701799102918e-05,
+      "loss": 1.0129,
+      "step": 4041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530637680666941e-05,
+      "loss": 0.9984,
+      "step": 4042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530405141490294e-05,
+      "loss": 0.9811,
+      "step": 4043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301725623978514e-05,
+      "loss": 0.9151,
+      "step": 4044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299399434071163e-05,
+      "loss": 1.0161,
+      "step": 4045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297072845355944e-05,
+      "loss": 0.9683,
+      "step": 4046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294745858007944e-05,
+      "loss": 0.9982,
+      "step": 4047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292418472202283e-05,
+      "loss": 1.0117,
+      "step": 4048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5290090688114103e-05,
+      "loss": 0.9452,
+      "step": 4049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5287762505918592e-05,
+      "loss": 0.9851,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285433925790946e-05,
+      "loss": 0.917,
+      "step": 4051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283104947906413e-05,
+      "loss": 1.0004,
+      "step": 4052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280775572440257e-05,
+      "loss": 1.0179,
+      "step": 4053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278445799567775e-05,
+      "loss": 0.9853,
+      "step": 4054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276115629464297e-05,
+      "loss": 1.0023,
+      "step": 4055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527378506230518e-05,
+      "loss": 0.9604,
+      "step": 4056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527145409826581e-05,
+      "loss": 0.9805,
+      "step": 4057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269122737521605e-05,
+      "loss": 0.9634,
+      "step": 4058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266790980248015e-05,
+      "loss": 0.9325,
+      "step": 4059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526445882662051e-05,
+      "loss": 0.9555,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262126276814604e-05,
+      "loss": 0.9572,
+      "step": 4061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259793331005832e-05,
+      "loss": 0.9756,
+      "step": 4062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525745998936976e-05,
+      "loss": 0.9882,
+      "step": 4063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255126252081986e-05,
+      "loss": 0.9644,
+      "step": 4064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525279211931813e-05,
+      "loss": 1.0161,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250457591253857e-05,
+      "loss": 1.0043,
+      "step": 4066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248122668064844e-05,
+      "loss": 0.9671,
+      "step": 4067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245787349926812e-05,
+      "loss": 1.0298,
+      "step": 4068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243451637015502e-05,
+      "loss": 0.9831,
+      "step": 4069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241115529506694e-05,
+      "loss": 0.9679,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238779027576187e-05,
+      "loss": 0.9622,
+      "step": 4071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523644213139982e-05,
+      "loss": 3.0398,
+      "step": 4072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234104841153454e-05,
+      "loss": 0.9934,
+      "step": 4073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231767157012981e-05,
+      "loss": 0.9565,
+      "step": 4074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229429079154327e-05,
+      "loss": 0.9938,
+      "step": 4075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227090607753442e-05,
+      "loss": 0.9779,
+      "step": 4076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224751742986313e-05,
+      "loss": 1.0248,
+      "step": 4077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222412485028946e-05,
+      "loss": 0.9496,
+      "step": 4078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220072834057387e-05,
+      "loss": 0.9704,
+      "step": 4079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217732790247705e-05,
+      "loss": 0.9498,
+      "step": 4080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215392353776004e-05,
+      "loss": 0.9804,
+      "step": 4081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213051524818405e-05,
+      "loss": 0.9668,
+      "step": 4082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521071030355108e-05,
+      "loss": 0.9365,
+      "step": 4083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208368690150215e-05,
+      "loss": 1.0012,
+      "step": 4084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206026684792022e-05,
+      "loss": 1.0024,
+      "step": 4085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520368428765275e-05,
+      "loss": 0.9763,
+      "step": 4086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201341498908688e-05,
+      "loss": 0.9767,
+      "step": 4087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198998318736134e-05,
+      "loss": 0.9813,
+      "step": 4088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196654747311423e-05,
+      "loss": 0.9932,
+      "step": 4089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194310784810929e-05,
+      "loss": 0.9598,
+      "step": 4090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519196643141104e-05,
+      "loss": 0.9569,
+      "step": 4091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189621687288185e-05,
+      "loss": 0.9483,
+      "step": 4092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187276552618818e-05,
+      "loss": 0.9902,
+      "step": 4093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184931027579422e-05,
+      "loss": 0.9868,
+      "step": 4094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182585112346513e-05,
+      "loss": 0.981,
+      "step": 4095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518023880709663e-05,
+      "loss": 0.9308,
+      "step": 4096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177892112006345e-05,
+      "loss": 0.9638,
+      "step": 4097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517554502725226e-05,
+      "loss": 0.9758,
+      "step": 4098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173197553011005e-05,
+      "loss": 1.001,
+      "step": 4099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170849689459247e-05,
+      "loss": 1.0399,
+      "step": 4100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168501436773664e-05,
+      "loss": 0.9663,
+      "step": 4101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166152795130979e-05,
+      "loss": 1.0002,
+      "step": 4102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516380376470794e-05,
+      "loss": 1.0155,
+      "step": 4103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161454345681327e-05,
+      "loss": 0.9513,
+      "step": 4104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515910453822794e-05,
+      "loss": 0.9698,
+      "step": 4105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515675434252462e-05,
+      "loss": 0.9642,
+      "step": 4106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154403758748228e-05,
+      "loss": 0.9802,
+      "step": 4107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515205278707566e-05,
+      "loss": 0.9745,
+      "step": 4108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149701427683838e-05,
+      "loss": 3.3201,
+      "step": 4109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147349680749713e-05,
+      "loss": 1.0277,
+      "step": 4110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144997546450273e-05,
+      "loss": 0.9961,
+      "step": 4111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142645024962521e-05,
+      "loss": 1.0132,
+      "step": 4112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140292116463498e-05,
+      "loss": 0.9931,
+      "step": 4113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137938821130273e-05,
+      "loss": 0.9655,
+      "step": 4114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135585139139948e-05,
+      "loss": 0.995,
+      "step": 4115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133231070669645e-05,
+      "loss": 1.0161,
+      "step": 4116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130876615896523e-05,
+      "loss": 0.9854,
+      "step": 4117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128521774997769e-05,
+      "loss": 0.9797,
+      "step": 4118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126166548150592e-05,
+      "loss": 1.0076,
+      "step": 4119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123810935532235e-05,
+      "loss": 0.9755,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121454937319976e-05,
+      "loss": 0.9571,
+      "step": 4121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119098553691116e-05,
+      "loss": 0.9508,
+      "step": 4122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116741784822981e-05,
+      "loss": 0.9523,
+      "step": 4123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511438463089293e-05,
+      "loss": 0.965,
+      "step": 4124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112027092078356e-05,
+      "loss": 0.9758,
+      "step": 4125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109669168556674e-05,
+      "loss": 0.9945,
+      "step": 4126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107310860505329e-05,
+      "loss": 3.422,
+      "step": 4127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104952168101798e-05,
+      "loss": 0.9696,
+      "step": 4128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102593091523585e-05,
+      "loss": 0.9707,
+      "step": 4129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510023363094822e-05,
+      "loss": 0.9925,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097873786553268e-05,
+      "loss": 0.9535,
+      "step": 4131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095513558516319e-05,
+      "loss": 0.9898,
+      "step": 4132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093152947014997e-05,
+      "loss": 0.9495,
+      "step": 4133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509079195222694e-05,
+      "loss": 0.9638,
+      "step": 4134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088430574329836e-05,
+      "loss": 1.014,
+      "step": 4135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086068813501381e-05,
+      "loss": 1.0006,
+      "step": 4136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083706669919323e-05,
+      "loss": 0.9573,
+      "step": 4137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081344143761416e-05,
+      "loss": 0.9358,
+      "step": 4138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078981235205453e-05,
+      "loss": 1.038,
+      "step": 4139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507661794442926e-05,
+      "loss": 0.9775,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074254271610682e-05,
+      "loss": 0.9826,
+      "step": 4141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071890216927604e-05,
+      "loss": 3.3052,
+      "step": 4142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069525780557927e-05,
+      "loss": 0.9621,
+      "step": 4143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067160962679595e-05,
+      "loss": 0.9134,
+      "step": 4144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064795763470565e-05,
+      "loss": 0.969,
+      "step": 4145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5062430183108835e-05,
+      "loss": 0.9758,
+      "step": 4146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060064221772428e-05,
+      "loss": 0.9455,
+      "step": 4147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505769787963939e-05,
+      "loss": 0.9279,
+      "step": 4148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055331156887811e-05,
+      "loss": 0.9365,
+      "step": 4149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052964053695788e-05,
+      "loss": 1.0084,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050596570241467e-05,
+      "loss": 0.9774,
+      "step": 4151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048228706703008e-05,
+      "loss": 0.9821,
+      "step": 4152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045860463258607e-05,
+      "loss": 0.9665,
+      "step": 4153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043491840086489e-05,
+      "loss": 1.0207,
+      "step": 4154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50411228373649e-05,
+      "loss": 0.9959,
+      "step": 4155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038753455272126e-05,
+      "loss": 0.9559,
+      "step": 4156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036383693986473e-05,
+      "loss": 0.9845,
+      "step": 4157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034013553686277e-05,
+      "loss": 1.0334,
+      "step": 4158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031643034549908e-05,
+      "loss": 1.001,
+      "step": 4159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029272136755753e-05,
+      "loss": 0.9467,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026900860482241e-05,
+      "loss": 0.9518,
+      "step": 4161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024529205907818e-05,
+      "loss": 0.9769,
+      "step": 4162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022157173210969e-05,
+      "loss": 0.9428,
+      "step": 4163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019784762570196e-05,
+      "loss": 0.9547,
+      "step": 4164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017411974164037e-05,
+      "loss": 1.0149,
+      "step": 4165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015038808171065e-05,
+      "loss": 0.9839,
+      "step": 4166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012665264769864e-05,
+      "loss": 0.9614,
+      "step": 4167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010291344139056e-05,
+      "loss": 0.9357,
+      "step": 4168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007917046457295e-05,
+      "loss": 1.0006,
+      "step": 4169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005542371903258e-05,
+      "loss": 3.2053,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003167320655654e-05,
+      "loss": 0.9575,
+      "step": 4171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000791892893216e-05,
+      "loss": 0.94,
+      "step": 4172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998416088794705e-05,
+      "loss": 1.0149,
+      "step": 4173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996039908538917e-05,
+      "loss": 0.9193,
+      "step": 4174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993663352304668e-05,
+      "loss": 0.9163,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991286420270813e-05,
+      "loss": 3.1115,
+      "step": 4176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988909112616221e-05,
+      "loss": 0.9322,
+      "step": 4177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986531429519805e-05,
+      "loss": 3.2206,
+      "step": 4178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498415337116049e-05,
+      "loss": 0.9636,
+      "step": 4179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981774937717243e-05,
+      "loss": 0.9661,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4979396129369055e-05,
+      "loss": 1.0051,
+      "step": 4181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977016946294937e-05,
+      "loss": 0.9492,
+      "step": 4182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974637388673941e-05,
+      "loss": 0.9664,
+      "step": 4183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497225745668514e-05,
+      "loss": 0.9854,
+      "step": 4184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969877150507635e-05,
+      "loss": 0.9959,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967496470320557e-05,
+      "loss": 0.9585,
+      "step": 4186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496511541630307e-05,
+      "loss": 0.967,
+      "step": 4187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962733988634352e-05,
+      "loss": 0.9995,
+      "step": 4188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960352187493623e-05,
+      "loss": 0.9323,
+      "step": 4189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957970013060124e-05,
+      "loss": 0.9915,
+      "step": 4190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495558746551313e-05,
+      "loss": 0.9596,
+      "step": 4191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.495320454503194e-05,
+      "loss": 0.9479,
+      "step": 4192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950821251795873e-05,
+      "loss": 0.9589,
+      "step": 4193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948437585984298e-05,
+      "loss": 0.8895,
+      "step": 4194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946053547776589e-05,
+      "loss": 1.0006,
+      "step": 4195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494366913735216e-05,
+      "loss": 0.9813,
+      "step": 4196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941284354890449e-05,
+      "loss": 0.9796,
+      "step": 4197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938899200570926e-05,
+      "loss": 0.9707,
+      "step": 4198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936513674573088e-05,
+      "loss": 1.0338,
+      "step": 4199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934127777076453e-05,
+      "loss": 0.9482,
+      "step": 4200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931741508260574e-05,
+      "loss": 0.9254,
+      "step": 4201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929354868305035e-05,
+      "loss": 0.9929,
+      "step": 4202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926967857389435e-05,
+      "loss": 0.9233,
+      "step": 4203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492458047569342e-05,
+      "loss": 1.0074,
+      "step": 4204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922192723396644e-05,
+      "loss": 0.9107,
+      "step": 4205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919804600678805e-05,
+      "loss": 0.9712,
+      "step": 4206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917416107719614e-05,
+      "loss": 1.0036,
+      "step": 4207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915027244698824e-05,
+      "loss": 0.9998,
+      "step": 4208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912638011796205e-05,
+      "loss": 3.345,
+      "step": 4209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910248409191566e-05,
+      "loss": 0.9157,
+      "step": 4210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490785843706473e-05,
+      "loss": 0.9581,
+      "step": 4211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905468095595561e-05,
+      "loss": 0.9684,
+      "step": 4212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490307738496394e-05,
+      "loss": 0.9527,
+      "step": 4213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900686305349782e-05,
+      "loss": 0.9602,
+      "step": 4214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489829485693303e-05,
+      "loss": 1.0,
+      "step": 4215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895903039893655e-05,
+      "loss": 3.0051,
+      "step": 4216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893510854411647e-05,
+      "loss": 0.968,
+      "step": 4217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891118300667038e-05,
+      "loss": 3.064,
+      "step": 4218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888725378839877e-05,
+      "loss": 0.9562,
+      "step": 4219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886332089110242e-05,
+      "loss": 0.987,
+      "step": 4220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883938431658245e-05,
+      "loss": 0.9652,
+      "step": 4221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881544406664016e-05,
+      "loss": 0.9274,
+      "step": 4222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879150014307726e-05,
+      "loss": 0.99,
+      "step": 4223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876755254769557e-05,
+      "loss": 0.9554,
+      "step": 4224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874360128229736e-05,
+      "loss": 0.9677,
+      "step": 4225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871964634868501e-05,
+      "loss": 0.9862,
+      "step": 4226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486956877486613e-05,
+      "loss": 0.9947,
+      "step": 4227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867172548402922e-05,
+      "loss": 0.9961,
+      "step": 4228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864775955659209e-05,
+      "loss": 0.9467,
+      "step": 4229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862378996815344e-05,
+      "loss": 0.9867,
+      "step": 4230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485998167205171e-05,
+      "loss": 0.9776,
+      "step": 4231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857583981548722e-05,
+      "loss": 0.9297,
+      "step": 4232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485518592548682e-05,
+      "loss": 0.9439,
+      "step": 4233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852787504046465e-05,
+      "loss": 0.9668,
+      "step": 4234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850388717408151e-05,
+      "loss": 0.9769,
+      "step": 4235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847989565752405e-05,
+      "loss": 0.954,
+      "step": 4236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845590049259775e-05,
+      "loss": 0.9723,
+      "step": 4237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484319016811083e-05,
+      "loss": 0.9764,
+      "step": 4238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840789922486183e-05,
+      "loss": 0.9235,
+      "step": 4239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838389312566459e-05,
+      "loss": 0.9529,
+      "step": 4240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835988338532318e-05,
+      "loss": 0.9772,
+      "step": 4241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833587000564448e-05,
+      "loss": 0.9616,
+      "step": 4242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831185298843563e-05,
+      "loss": 0.9767,
+      "step": 4243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828783233550402e-05,
+      "loss": 0.9812,
+      "step": 4244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482638080486573e-05,
+      "loss": 0.9716,
+      "step": 4245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823978012970347e-05,
+      "loss": 0.8901,
+      "step": 4246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821574858045073e-05,
+      "loss": 2.9682,
+      "step": 4247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819171340270764e-05,
+      "loss": 1.0124,
+      "step": 4248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816767459828288e-05,
+      "loss": 1.0175,
+      "step": 4249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814363216898555e-05,
+      "loss": 1.0165,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48119586116625e-05,
+      "loss": 0.9715,
+      "step": 4251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809553644301076e-05,
+      "loss": 0.9592,
+      "step": 4252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807148314995274e-05,
+      "loss": 0.9279,
+      "step": 4253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804742623926108e-05,
+      "loss": 0.9661,
+      "step": 4254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4802336571274614e-05,
+      "loss": 0.9684,
+      "step": 4255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799930157221862e-05,
+      "loss": 0.9372,
+      "step": 4256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797523381948953e-05,
+      "loss": 0.9253,
+      "step": 4257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795116245637004e-05,
+      "loss": 0.9713,
+      "step": 4258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792708748467162e-05,
+      "loss": 0.9585,
+      "step": 4259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790300890620615e-05,
+      "loss": 0.9676,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787892672278557e-05,
+      "loss": 0.9835,
+      "step": 4261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478548409362222e-05,
+      "loss": 0.9953,
+      "step": 4262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783075154832867e-05,
+      "loss": 0.9723,
+      "step": 4263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780665856091783e-05,
+      "loss": 0.9327,
+      "step": 4264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778256197580276e-05,
+      "loss": 0.9391,
+      "step": 4265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775846179479687e-05,
+      "loss": 0.97,
+      "step": 4266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773435801971385e-05,
+      "loss": 0.9895,
+      "step": 4267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771025065236763e-05,
+      "loss": 0.9437,
+      "step": 4268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476861396945724e-05,
+      "loss": 1.0262,
+      "step": 4269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766202514814267e-05,
+      "loss": 0.9661,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763790701489314e-05,
+      "loss": 3.1805,
+      "step": 4271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761378529663889e-05,
+      "loss": 0.981,
+      "step": 4272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758965999519513e-05,
+      "loss": 0.9864,
+      "step": 4273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756553111237745e-05,
+      "loss": 0.9744,
+      "step": 4274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475413986500017e-05,
+      "loss": 0.9697,
+      "step": 4275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751726260988397e-05,
+      "loss": 0.9539,
+      "step": 4276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474931229938406e-05,
+      "loss": 0.9894,
+      "step": 4277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746897980368824e-05,
+      "loss": 0.9689,
+      "step": 4278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744483304124377e-05,
+      "loss": 0.9872,
+      "step": 4279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742068270832439e-05,
+      "loss": 0.974,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739652880674756e-05,
+      "loss": 1.0244,
+      "step": 4281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737237133833096e-05,
+      "loss": 0.9834,
+      "step": 4282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734821030489253e-05,
+      "loss": 1.0127,
+      "step": 4283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732404570825057e-05,
+      "loss": 0.9707,
+      "step": 4284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472998775502236e-05,
+      "loss": 1.0029,
+      "step": 4285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727570583263036e-05,
+      "loss": 0.9719,
+      "step": 4286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725153055728988e-05,
+      "loss": 0.9844,
+      "step": 4287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722735172602156e-05,
+      "loss": 1.0182,
+      "step": 4288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4720316934064496e-05,
+      "loss": 1.0427,
+      "step": 4289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717898340297988e-05,
+      "loss": 1.0088,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715479391484651e-05,
+      "loss": 0.9606,
+      "step": 4291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713060087806517e-05,
+      "loss": 0.9624,
+      "step": 4292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710640429445656e-05,
+      "loss": 0.9334,
+      "step": 4293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708220416584155e-05,
+      "loss": 0.9817,
+      "step": 4294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705800049404143e-05,
+      "loss": 0.9515,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703379328087754e-05,
+      "loss": 0.998,
+      "step": 4296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700958252817165e-05,
+      "loss": 1.002,
+      "step": 4297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698536823774578e-05,
+      "loss": 0.9578,
+      "step": 4298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696115041142214e-05,
+      "loss": 0.9592,
+      "step": 4299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693692905102324e-05,
+      "loss": 0.9574,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691270415837189e-05,
+      "loss": 1.0093,
+      "step": 4301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688847573529115e-05,
+      "loss": 0.9889,
+      "step": 4302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686424378360434e-05,
+      "loss": 0.9734,
+      "step": 4303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684000830513498e-05,
+      "loss": 0.9794,
+      "step": 4304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46815769301707e-05,
+      "loss": 0.9649,
+      "step": 4305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679152677514448e-05,
+      "loss": 0.9825,
+      "step": 4306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676728072727178e-05,
+      "loss": 1.0003,
+      "step": 4307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674303115991358e-05,
+      "loss": 0.981,
+      "step": 4308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4671877807489475e-05,
+      "loss": 0.9824,
+      "step": 4309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4669452147404051e-05,
+      "loss": 0.9667,
+      "step": 4310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667026135917624e-05,
+      "loss": 0.9127,
+      "step": 4311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664599773212766e-05,
+      "loss": 0.9578,
+      "step": 4312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466217305947208e-05,
+      "loss": 0.9107,
+      "step": 4313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465974599487818e-05,
+      "loss": 0.9409,
+      "step": 4314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657318579613721e-05,
+      "loss": 0.9707,
+      "step": 4315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654890813861375e-05,
+      "loss": 0.9645,
+      "step": 4316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652462697803848e-05,
+      "loss": 0.9823,
+      "step": 4317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650034231623868e-05,
+      "loss": 0.9274,
+      "step": 4318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464760541550419e-05,
+      "loss": 0.9887,
+      "step": 4319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464517624962759e-05,
+      "loss": 0.968,
+      "step": 4320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642746734176885e-05,
+      "loss": 1.0024,
+      "step": 4321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640316869334898e-05,
+      "loss": 0.9787,
+      "step": 4322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46378866552845e-05,
+      "loss": 0.9917,
+      "step": 4323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635456092208571e-05,
+      "loss": 0.9623,
+      "step": 4324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633025180290021e-05,
+      "loss": 0.9637,
+      "step": 4325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630593919711797e-05,
+      "loss": 0.924,
+      "step": 4326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628162310656861e-05,
+      "loss": 1.027,
+      "step": 4327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625730353308203e-05,
+      "loss": 0.9956,
+      "step": 4328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623298047848841e-05,
+      "loss": 0.9602,
+      "step": 4329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462086539446182e-05,
+      "loss": 0.9816,
+      "step": 4330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4618432393330211e-05,
+      "loss": 0.9757,
+      "step": 4331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615999044637106e-05,
+      "loss": 0.9507,
+      "step": 4332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613565348565631e-05,
+      "loss": 0.995,
+      "step": 4333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611131305298934e-05,
+      "loss": 0.9965,
+      "step": 4334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608696915020186e-05,
+      "loss": 0.9761,
+      "step": 4335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606262177912593e-05,
+      "loss": 0.9644,
+      "step": 4336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603827094159382e-05,
+      "loss": 0.9417,
+      "step": 4337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601391663943802e-05,
+      "loss": 0.9865,
+      "step": 4338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598955887449133e-05,
+      "loss": 1.0112,
+      "step": 4339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596519764858683e-05,
+      "loss": 0.9707,
+      "step": 4340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459408329635578e-05,
+      "loss": 0.9755,
+      "step": 4341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459164648212378e-05,
+      "loss": 0.957,
+      "step": 4342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458920932234607e-05,
+      "loss": 0.9643,
+      "step": 4343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586771817206057e-05,
+      "loss": 0.9791,
+      "step": 4344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584333966887178e-05,
+      "loss": 0.8906,
+      "step": 4345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458189577157289e-05,
+      "loss": 0.9709,
+      "step": 4346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579457231446688e-05,
+      "loss": 0.957,
+      "step": 4347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577018346692075e-05,
+      "loss": 1.022,
+      "step": 4348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574579117492598e-05,
+      "loss": 0.9313,
+      "step": 4349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572139544031815e-05,
+      "loss": 0.96,
+      "step": 4350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569699626493326e-05,
+      "loss": 0.9736,
+      "step": 4351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456725936506074e-05,
+      "loss": 0.9521,
+      "step": 4352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4564818759917698e-05,
+      "loss": 0.9486,
+      "step": 4353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562377811247879e-05,
+      "loss": 0.9115,
+      "step": 4354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559936519234967e-05,
+      "loss": 0.995,
+      "step": 4355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557494884062686e-05,
+      "loss": 0.9968,
+      "step": 4356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555052905914784e-05,
+      "loss": 1.0143,
+      "step": 4357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552610584975029e-05,
+      "loss": 0.9467,
+      "step": 4358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455016792142722e-05,
+      "loss": 0.9587,
+      "step": 4359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547724915455181e-05,
+      "loss": 0.9634,
+      "step": 4360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545281567242761e-05,
+      "loss": 0.9856,
+      "step": 4361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542837876973836e-05,
+      "loss": 0.9937,
+      "step": 4362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540393844832301e-05,
+      "loss": 0.9878,
+      "step": 4363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537949471002094e-05,
+      "loss": 0.983,
+      "step": 4364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535504755667153e-05,
+      "loss": 0.9176,
+      "step": 4365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4533059699011464e-05,
+      "loss": 0.9775,
+      "step": 4366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453061430121903e-05,
+      "loss": 0.9327,
+      "step": 4367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452816856247388e-05,
+      "loss": 0.9494,
+      "step": 4368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452572248296007e-05,
+      "loss": 0.9249,
+      "step": 4369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523276062861672e-05,
+      "loss": 0.9342,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45208293023628e-05,
+      "loss": 0.9927,
+      "step": 4371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518382201647588e-05,
+      "loss": 0.9938,
+      "step": 4372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515934760900185e-05,
+      "loss": 0.9958,
+      "step": 4373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451348698030478e-05,
+      "loss": 0.9227,
+      "step": 4374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451103886004558e-05,
+      "loss": 1.0054,
+      "step": 4375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508590400306818e-05,
+      "loss": 0.9772,
+      "step": 4376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506141601272757e-05,
+      "loss": 0.9665,
+      "step": 4377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503692463127674e-05,
+      "loss": 0.965,
+      "step": 4378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450124298605589e-05,
+      "loss": 0.9444,
+      "step": 4379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498793170241737e-05,
+      "loss": 1.0386,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496343015869572e-05,
+      "loss": 0.9595,
+      "step": 4381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493892523123784e-05,
+      "loss": 0.9815,
+      "step": 4382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491441692188793e-05,
+      "loss": 1.033,
+      "step": 4383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488990523249027e-05,
+      "loss": 1.0131,
+      "step": 4384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486539016488956e-05,
+      "loss": 0.9895,
+      "step": 4385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484087172093065e-05,
+      "loss": 0.9752,
+      "step": 4386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481634990245871e-05,
+      "loss": 0.9792,
+      "step": 4387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4479182471131912e-05,
+      "loss": 0.978,
+      "step": 4388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476729614935754e-05,
+      "loss": 3.0979,
+      "step": 4389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474276421841987e-05,
+      "loss": 0.906,
+      "step": 4390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471822892035226e-05,
+      "loss": 0.9424,
+      "step": 4391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469369025700116e-05,
+      "loss": 0.9771,
+      "step": 4392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466914823021319e-05,
+      "loss": 0.9608,
+      "step": 4393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464460284183525e-05,
+      "loss": 0.9858,
+      "step": 4394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462005409371456e-05,
+      "loss": 0.9468,
+      "step": 4395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459550198769853e-05,
+      "loss": 0.9499,
+      "step": 4396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457094652563486e-05,
+      "loss": 0.989,
+      "step": 4397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454638770937141e-05,
+      "loss": 0.9711,
+      "step": 4398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445218255407564e-05,
+      "loss": 0.9845,
+      "step": 4399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449726002163828e-05,
+      "loss": 0.9182,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447269115386574e-05,
+      "loss": 1.016,
+      "step": 4401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444811893928765e-05,
+      "loss": 0.9706,
+      "step": 4402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442354337975326e-05,
+      "loss": 0.9822,
+      "step": 4403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439896447711204e-05,
+      "loss": 1.0341,
+      "step": 4404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437438223321359e-05,
+      "loss": 0.9688,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434979664990792e-05,
+      "loss": 0.9521,
+      "step": 4406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443252077290452e-05,
+      "loss": 0.919,
+      "step": 4407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443006154724759e-05,
+      "loss": 0.9716,
+      "step": 4408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4427601988205067e-05,
+      "loss": 0.9356,
+      "step": 4409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425142095962055e-05,
+      "loss": 0.9363,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422681870703664e-05,
+      "loss": 0.9681,
+      "step": 4411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420221312615041e-05,
+      "loss": 0.9619,
+      "step": 4412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417760421881361e-05,
+      "loss": 0.979,
+      "step": 4413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415299198687818e-05,
+      "loss": 0.9471,
+      "step": 4414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412837643219625e-05,
+      "loss": 0.9816,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4410375755662036e-05,
+      "loss": 1.0074,
+      "step": 4416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407913536200318e-05,
+      "loss": 1.0217,
+      "step": 4417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405450985019764e-05,
+      "loss": 0.9942,
+      "step": 4418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402988102305693e-05,
+      "loss": 0.9522,
+      "step": 4419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400524888243456e-05,
+      "loss": 0.9618,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439806134301842e-05,
+      "loss": 2.9827,
+      "step": 4421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395597466815976e-05,
+      "loss": 0.9708,
+      "step": 4422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393133259821546e-05,
+      "loss": 0.9874,
+      "step": 4423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4390668722220578e-05,
+      "loss": 0.968,
+      "step": 4424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438820385419854e-05,
+      "loss": 0.9909,
+      "step": 4425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4385738655940921e-05,
+      "loss": 0.9485,
+      "step": 4426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438327312763325e-05,
+      "loss": 0.9481,
+      "step": 4427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.438080726946106e-05,
+      "loss": 1.0031,
+      "step": 4428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378341081609928e-05,
+      "loss": 0.9798,
+      "step": 4429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375874564265448e-05,
+      "loss": 0.9677,
+      "step": 4430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373407717613234e-05,
+      "loss": 0.9704,
+      "step": 4431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370940541838929e-05,
+      "loss": 0.9691,
+      "step": 4432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368473037128203e-05,
+      "loss": 0.9881,
+      "step": 4433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4366005203666751e-05,
+      "loss": 1.0004,
+      "step": 4434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363537041640289e-05,
+      "loss": 0.9769,
+      "step": 4435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361068551234556e-05,
+      "loss": 0.9583,
+      "step": 4436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435859973263532e-05,
+      "loss": 0.9883,
+      "step": 4437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356130586028378e-05,
+      "loss": 0.9341,
+      "step": 4438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353661111599541e-05,
+      "loss": 0.9745,
+      "step": 4439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351191309534651e-05,
+      "loss": 0.9766,
+      "step": 4440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348721180019576e-05,
+      "loss": 0.9373,
+      "step": 4441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346250723240205e-05,
+      "loss": 0.9847,
+      "step": 4442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343779939382451e-05,
+      "loss": 0.9718,
+      "step": 4443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341308828632254e-05,
+      "loss": 0.9987,
+      "step": 4444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 3.1191,
+      "step": 4445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336365627198424e-05,
+      "loss": 0.9632,
+      "step": 4446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333893536886783e-05,
+      "loss": 0.9744,
+      "step": 4447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331421120426706e-05,
+      "loss": 0.9941,
+      "step": 4448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328948378004257e-05,
+      "loss": 0.9622,
+      "step": 4449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326475309805518e-05,
+      "loss": 0.9204,
+      "step": 4450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324001916016603e-05,
+      "loss": 0.9475,
+      "step": 4451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321528196823647e-05,
+      "loss": 1.0103,
+      "step": 4452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431905415241281e-05,
+      "loss": 0.9582,
+      "step": 4453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316579782970277e-05,
+      "loss": 0.9917,
+      "step": 4454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431410508868226e-05,
+      "loss": 0.9833,
+      "step": 4455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311630069734988e-05,
+      "loss": 0.9186,
+      "step": 4456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309154726314722e-05,
+      "loss": 0.9661,
+      "step": 4457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306679058607747e-05,
+      "loss": 0.9903,
+      "step": 4458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304203066800368e-05,
+      "loss": 0.9934,
+      "step": 4459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4301726751078915e-05,
+      "loss": 0.9716,
+      "step": 4460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299250111629746e-05,
+      "loss": 0.9906,
+      "step": 4461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429677314863924e-05,
+      "loss": 1.0076,
+      "step": 4462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294295862293804e-05,
+      "loss": 0.9992,
+      "step": 4463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291818252779863e-05,
+      "loss": 1.0136,
+      "step": 4464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428934032028387e-05,
+      "loss": 0.924,
+      "step": 4465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428686206499231e-05,
+      "loss": 0.9613,
+      "step": 4466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284383487091675e-05,
+      "loss": 1.0028,
+      "step": 4467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281904586768499e-05,
+      "loss": 0.9354,
+      "step": 4468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279425364209327e-05,
+      "loss": 0.9634,
+      "step": 4469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276945819600739e-05,
+      "loss": 0.967,
+      "step": 4470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4274465953129326e-05,
+      "loss": 0.9596,
+      "step": 4471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271985764981717e-05,
+      "loss": 0.9691,
+      "step": 4472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269505255344558e-05,
+      "loss": 0.9242,
+      "step": 4473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267024424404523e-05,
+      "loss": 1.0022,
+      "step": 4474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42645432723483e-05,
+      "loss": 0.9803,
+      "step": 4475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262061799362621e-05,
+      "loss": 0.9831,
+      "step": 4476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259580005634218e-05,
+      "loss": 0.9741,
+      "step": 4477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257097891349863e-05,
+      "loss": 0.9848,
+      "step": 4478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254615456696352e-05,
+      "loss": 0.9363,
+      "step": 4479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42521327018605e-05,
+      "loss": 0.9587,
+      "step": 4480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249649627029143e-05,
+      "loss": 0.9665,
+      "step": 4481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247166232389149e-05,
+      "loss": 0.9863,
+      "step": 4482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244682518127406e-05,
+      "loss": 1.0218,
+      "step": 4483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424219848443083e-05,
+      "loss": 0.9778,
+      "step": 4484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423971413148635e-05,
+      "loss": 0.9944,
+      "step": 4485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237229459480934e-05,
+      "loss": 0.9982,
+      "step": 4486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234744468601564e-05,
+      "loss": 1.0205,
+      "step": 4487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232259159035246e-05,
+      "loss": 0.9596,
+      "step": 4488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229773530969018e-05,
+      "loss": 0.9483,
+      "step": 4489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227287584589936e-05,
+      "loss": 0.9682,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224801320085075e-05,
+      "loss": 0.9401,
+      "step": 4491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222314737641545e-05,
+      "loss": 0.952,
+      "step": 4492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219827837446472e-05,
+      "loss": 0.9428,
+      "step": 4493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421734061968701e-05,
+      "loss": 0.9971,
+      "step": 4494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214853084550332e-05,
+      "loss": 0.9971,
+      "step": 4495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212365232223644e-05,
+      "loss": 0.9701,
+      "step": 4496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209877062894166e-05,
+      "loss": 0.9676,
+      "step": 4497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207388576749146e-05,
+      "loss": 0.9867,
+      "step": 4498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204899773975855e-05,
+      "loss": 0.9739,
+      "step": 4499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202410654761593e-05,
+      "loss": 0.9922,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199921219293678e-05,
+      "loss": 0.9848,
+      "step": 4501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197431467759446e-05,
+      "loss": 0.9295,
+      "step": 4502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194941400346271e-05,
+      "loss": 0.9378,
+      "step": 4503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192451017241545e-05,
+      "loss": 0.9785,
+      "step": 4504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189960318632677e-05,
+      "loss": 0.9404,
+      "step": 4505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187469304707109e-05,
+      "loss": 2.9685,
+      "step": 4506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184977975652303e-05,
+      "loss": 0.9795,
+      "step": 4507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182486331655746e-05,
+      "loss": 0.9894,
+      "step": 4508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179994372904942e-05,
+      "loss": 1.0013,
+      "step": 4509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177502099587425e-05,
+      "loss": 0.9558,
+      "step": 4510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417500951189076e-05,
+      "loss": 0.9869,
+      "step": 4511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172516610002518e-05,
+      "loss": 0.9668,
+      "step": 4512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170023394110306e-05,
+      "loss": 1.0091,
+      "step": 4513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167529864401755e-05,
+      "loss": 0.9906,
+      "step": 4514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165036021064511e-05,
+      "loss": 3.0066,
+      "step": 4515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162541864286252e-05,
+      "loss": 0.9955,
+      "step": 4516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160047394254677e-05,
+      "loss": 0.9671,
+      "step": 4517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157552611157507e-05,
+      "loss": 0.9859,
+      "step": 4518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415505751518249e-05,
+      "loss": 0.9272,
+      "step": 4519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152562106517387e-05,
+      "loss": 0.9529,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150066385350001e-05,
+      "loss": 0.9695,
+      "step": 4521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147570351868143e-05,
+      "loss": 0.9984,
+      "step": 4522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4145074006259651e-05,
+      "loss": 0.9612,
+      "step": 4523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142577348712397e-05,
+      "loss": 0.963,
+      "step": 4524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140080379414255e-05,
+      "loss": 0.9329,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137583098553144e-05,
+      "loss": 0.9904,
+      "step": 4526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135085506316997e-05,
+      "loss": 0.9921,
+      "step": 4527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132587602893768e-05,
+      "loss": 0.9969,
+      "step": 4528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.413008938847144e-05,
+      "loss": 0.9688,
+      "step": 4529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127590863238015e-05,
+      "loss": 0.9507,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125092027381517e-05,
+      "loss": 0.9309,
+      "step": 4531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122592881090005e-05,
+      "loss": 0.9889,
+      "step": 4532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412009342455155e-05,
+      "loss": 0.9379,
+      "step": 4533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117593657954242e-05,
+      "loss": 0.9888,
+      "step": 4534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115093581486211e-05,
+      "loss": 2.9959,
+      "step": 4535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41125931953356e-05,
+      "loss": 3.0445,
+      "step": 4536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411009249969057e-05,
+      "loss": 0.9349,
+      "step": 4537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107591494739316e-05,
+      "loss": 0.978,
+      "step": 4538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105090180670052e-05,
+      "loss": 0.9766,
+      "step": 4539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4102588557671015e-05,
+      "loss": 0.9366,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100086625930464e-05,
+      "loss": 1.007,
+      "step": 4541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097584385636684e-05,
+      "loss": 0.969,
+      "step": 4542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095081836977981e-05,
+      "loss": 2.9229,
+      "step": 4543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092578980142684e-05,
+      "loss": 0.9667,
+      "step": 4544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409007581531915e-05,
+      "loss": 0.9158,
+      "step": 4545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4087572342695751e-05,
+      "loss": 3.0067,
+      "step": 4546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085068562460888e-05,
+      "loss": 0.8877,
+      "step": 4547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4082564474802987e-05,
+      "loss": 0.9704,
+      "step": 4548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080060079910489e-05,
+      "loss": 0.9483,
+      "step": 4549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077555377971867e-05,
+      "loss": 0.975,
+      "step": 4550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075050369175607e-05,
+      "loss": 0.9876,
+      "step": 4551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072545053710231e-05,
+      "loss": 0.9795,
+      "step": 4552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070039431764275e-05,
+      "loss": 0.9938,
+      "step": 4553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067533503526298e-05,
+      "loss": 0.9955,
+      "step": 4554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065027269184888e-05,
+      "loss": 0.9812,
+      "step": 4555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406252072892865e-05,
+      "loss": 0.9206,
+      "step": 4556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060013882946217e-05,
+      "loss": 0.9101,
+      "step": 4557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057506731426238e-05,
+      "loss": 0.9836,
+      "step": 4558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054999274557392e-05,
+      "loss": 0.9619,
+      "step": 4559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405249151252838e-05,
+      "loss": 0.9932,
+      "step": 4560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049983445527922e-05,
+      "loss": 0.9607,
+      "step": 4561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047475073744762e-05,
+      "loss": 1.0078,
+      "step": 4562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044966397367673e-05,
+      "loss": 0.9778,
+      "step": 4563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042457416585444e-05,
+      "loss": 0.9718,
+      "step": 4564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039948131586884e-05,
+      "loss": 2.9548,
+      "step": 4565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403743854256084e-05,
+      "loss": 0.9762,
+      "step": 4566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034928649696165e-05,
+      "loss": 0.9736,
+      "step": 4567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032418453181742e-05,
+      "loss": 0.99,
+      "step": 4568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029907953206476e-05,
+      "loss": 0.9731,
+      "step": 4569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027397149959302e-05,
+      "loss": 0.9564,
+      "step": 4570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4024886043629163e-05,
+      "loss": 0.973,
+      "step": 4571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022374634405033e-05,
+      "loss": 0.9711,
+      "step": 4572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401986292247592e-05,
+      "loss": 0.9997,
+      "step": 4573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017350908030833e-05,
+      "loss": 0.9849,
+      "step": 4574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014838591258815e-05,
+      "loss": 1.0201,
+      "step": 4575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4012325972348935e-05,
+      "loss": 2.9601,
+      "step": 4576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009813051490279e-05,
+      "loss": 0.9626,
+      "step": 4577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007299828871959e-05,
+      "loss": 0.9966,
+      "step": 4578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004786304683106e-05,
+      "loss": 0.9932,
+      "step": 4579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4002272479112877e-05,
+      "loss": 0.9698,
+      "step": 4580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999758352350452e-05,
+      "loss": 0.9752,
+      "step": 4581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399724392458503e-05,
+      "loss": 0.9637,
+      "step": 4582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994729196005839e-05,
+      "loss": 1.0037,
+      "step": 4583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992214166802119e-05,
+      "loss": 0.9434,
+      "step": 4584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989698837163145e-05,
+      "loss": 0.8971,
+      "step": 4585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987183207278207e-05,
+      "loss": 0.9652,
+      "step": 4586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3984667277336622e-05,
+      "loss": 0.9722,
+      "step": 4587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982151047527719e-05,
+      "loss": 0.9891,
+      "step": 4588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979634518040865e-05,
+      "loss": 0.9616,
+      "step": 4589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977117689065445e-05,
+      "loss": 0.9659,
+      "step": 4590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974600560790856e-05,
+      "loss": 0.9744,
+      "step": 4591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972083133406527e-05,
+      "loss": 1.0128,
+      "step": 4592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969565407101909e-05,
+      "loss": 0.947,
+      "step": 4593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967047382066476e-05,
+      "loss": 0.9473,
+      "step": 4594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3964529058489721e-05,
+      "loss": 0.9624,
+      "step": 4595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962010436561162e-05,
+      "loss": 0.914,
+      "step": 4596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959491516470333e-05,
+      "loss": 0.9777,
+      "step": 4597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956972298406806e-05,
+      "loss": 0.9813,
+      "step": 4598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954452782560157e-05,
+      "loss": 0.9393,
+      "step": 4599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395193296912e-05,
+      "loss": 0.9826,
+      "step": 4600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949412858275956e-05,
+      "loss": 0.9457,
+      "step": 4601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3946892450217688e-05,
+      "loss": 0.9406,
+      "step": 4602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944371745134857e-05,
+      "loss": 0.9885,
+      "step": 4603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941850743217168e-05,
+      "loss": 0.9945,
+      "step": 4604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939329444654337e-05,
+      "loss": 1.0047,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936807849636107e-05,
+      "loss": 0.9453,
+      "step": 4606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934285958352238e-05,
+      "loss": 0.9969,
+      "step": 4607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931763770992519e-05,
+      "loss": 0.9785,
+      "step": 4608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929241287746756e-05,
+      "loss": 0.9678,
+      "step": 4609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392671850880478e-05,
+      "loss": 1.015,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924195434356443e-05,
+      "loss": 0.9853,
+      "step": 4611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392167206459162e-05,
+      "loss": 0.9379,
+      "step": 4612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391914839970021e-05,
+      "loss": 0.9423,
+      "step": 4613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916624439872126e-05,
+      "loss": 0.8756,
+      "step": 4614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914100185297315e-05,
+      "loss": 0.9828,
+      "step": 4615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911575636165742e-05,
+      "loss": 0.9456,
+      "step": 4616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909050792667383e-05,
+      "loss": 1.0398,
+      "step": 4617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906525654992255e-05,
+      "loss": 0.9647,
+      "step": 4618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390400022333039e-05,
+      "loss": 0.9711,
+      "step": 4619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901474497871829e-05,
+      "loss": 0.9012,
+      "step": 4620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898948478806657e-05,
+      "loss": 0.9563,
+      "step": 4621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3896422166324965e-05,
+      "loss": 1.0024,
+      "step": 4622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893895560616875e-05,
+      "loss": 3.0736,
+      "step": 4623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891368661872522e-05,
+      "loss": 0.9582,
+      "step": 4624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888841470282071e-05,
+      "loss": 0.9734,
+      "step": 4625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388631398603571e-05,
+      "loss": 0.9686,
+      "step": 4626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3883786209323638e-05,
+      "loss": 0.9814,
+      "step": 4627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388125814033609e-05,
+      "loss": 1.0237,
+      "step": 4628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878729779263316e-05,
+      "loss": 0.9929,
+      "step": 4629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876201126295588e-05,
+      "loss": 0.9838,
+      "step": 4630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873672181623197e-05,
+      "loss": 0.9591,
+      "step": 4631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871142945436465e-05,
+      "loss": 0.9881,
+      "step": 4632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868613417925728e-05,
+      "loss": 0.9951,
+      "step": 4633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386608359928134e-05,
+      "loss": 3.3261,
+      "step": 4634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863553489693693e-05,
+      "loss": 1.0042,
+      "step": 4635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386102308935319e-05,
+      "loss": 1.0,
+      "step": 4636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858492398450252e-05,
+      "loss": 0.9482,
+      "step": 4637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855961417175328e-05,
+      "loss": 1.0012,
+      "step": 4638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853430145718892e-05,
+      "loss": 0.9792,
+      "step": 4639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850898584271432e-05,
+      "loss": 0.9652,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384836673302346e-05,
+      "loss": 0.9885,
+      "step": 4641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845834592165514e-05,
+      "loss": 0.9371,
+      "step": 4642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843302161888153e-05,
+      "loss": 0.9691,
+      "step": 4643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840769442381952e-05,
+      "loss": 0.988,
+      "step": 4644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383823643383751e-05,
+      "loss": 0.978,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835703136445453e-05,
+      "loss": 0.9518,
+      "step": 4646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833169550396426e-05,
+      "loss": 0.9622,
+      "step": 4647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383063567588109e-05,
+      "loss": 0.9286,
+      "step": 4648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828101513090137e-05,
+      "loss": 0.9397,
+      "step": 4649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825567062214273e-05,
+      "loss": 0.9402,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823032323444232e-05,
+      "loss": 1.0079,
+      "step": 4651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820497296970765e-05,
+      "loss": 0.9574,
+      "step": 4652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817961982984644e-05,
+      "loss": 0.9672,
+      "step": 4653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815426381676669e-05,
+      "loss": 0.9526,
+      "step": 4654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812890493237652e-05,
+      "loss": 0.9203,
+      "step": 4655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810354317858436e-05,
+      "loss": 0.9541,
+      "step": 4656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807817855729883e-05,
+      "loss": 0.9464,
+      "step": 4657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805281107042872e-05,
+      "loss": 0.9964,
+      "step": 4658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802744071988305e-05,
+      "loss": 0.9553,
+      "step": 4659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800206750757114e-05,
+      "loss": 0.9631,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797669143540241e-05,
+      "loss": 0.9639,
+      "step": 4661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795131250528655e-05,
+      "loss": 2.978,
+      "step": 4662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3792593071913344e-05,
+      "loss": 0.934,
+      "step": 4663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3790054607885325e-05,
+      "loss": 0.9372,
+      "step": 4664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787515858635626e-05,
+      "loss": 0.9159,
+      "step": 4665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784976824355302e-05,
+      "loss": 0.9816,
+      "step": 4666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378243750523543e-05,
+      "loss": 0.985,
+      "step": 4667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377989790146711e-05,
+      "loss": 0.9392,
+      "step": 4668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777358013241452e-05,
+      "loss": 1.0036,
+      "step": 4669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774817840749607e-05,
+      "loss": 0.976,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772277384182726e-05,
+      "loss": 0.9023,
+      "step": 4671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769736643732e-05,
+      "loss": 0.8897,
+      "step": 4672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767195619588626e-05,
+      "loss": 0.9765,
+      "step": 4673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376465431194384e-05,
+      "loss": 1.0334,
+      "step": 4674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762112720988876e-05,
+      "loss": 0.9489,
+      "step": 4675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759570846915009e-05,
+      "loss": 0.9994,
+      "step": 4676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757028689913532e-05,
+      "loss": 0.9171,
+      "step": 4677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754486250175748e-05,
+      "loss": 0.9618,
+      "step": 4678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751943527892995e-05,
+      "loss": 0.9647,
+      "step": 4679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749400523256622e-05,
+      "loss": 0.978,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746857236458007e-05,
+      "loss": 0.9396,
+      "step": 4681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3744313667688544e-05,
+      "loss": 0.9856,
+      "step": 4682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741769817139648e-05,
+      "loss": 0.9732,
+      "step": 4683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739225685002762e-05,
+      "loss": 0.9714,
+      "step": 4684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736681271469345e-05,
+      "loss": 0.9599,
+      "step": 4685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734136576730873e-05,
+      "loss": 0.9795,
+      "step": 4686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373159160097885e-05,
+      "loss": 0.9516,
+      "step": 4687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729046344404802e-05,
+      "loss": 0.9823,
+      "step": 4688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726500807200269e-05,
+      "loss": 0.9422,
+      "step": 4689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723954989556815e-05,
+      "loss": 0.9635,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721408891666031e-05,
+      "loss": 0.9633,
+      "step": 4691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718862513719524e-05,
+      "loss": 0.9945,
+      "step": 4692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3716315855908918e-05,
+      "loss": 0.9412,
+      "step": 4693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713768918425868e-05,
+      "loss": 3.0209,
+      "step": 4694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711221701462037e-05,
+      "loss": 0.9128,
+      "step": 4695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708674205209124e-05,
+      "loss": 0.888,
+      "step": 4696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706126429858836e-05,
+      "loss": 0.9797,
+      "step": 4697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703578375602913e-05,
+      "loss": 0.9969,
+      "step": 4698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701030042633103e-05,
+      "loss": 0.9655,
+      "step": 4699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698481431141185e-05,
+      "loss": 0.9919,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3695932541318958e-05,
+      "loss": 1.011,
+      "step": 4701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693383373358234e-05,
+      "loss": 0.9646,
+      "step": 4702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690833927450853e-05,
+      "loss": 0.9621,
+      "step": 4703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688284203788675e-05,
+      "loss": 0.9717,
+      "step": 4704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368573420256358e-05,
+      "loss": 0.9592,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368318392396747e-05,
+      "loss": 0.9654,
+      "step": 4706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680633368192269e-05,
+      "loss": 0.9781,
+      "step": 4707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678082535429914e-05,
+      "loss": 0.9792,
+      "step": 4708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675531425872373e-05,
+      "loss": 0.9421,
+      "step": 4709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367298003971163e-05,
+      "loss": 0.9627,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670428377139689e-05,
+      "loss": 1.011,
+      "step": 4711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667876438348576e-05,
+      "loss": 0.9756,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665324223530338e-05,
+      "loss": 0.9967,
+      "step": 4713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662771732877044e-05,
+      "loss": 0.9304,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660218966580785e-05,
+      "loss": 0.9884,
+      "step": 4715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657665924833663e-05,
+      "loss": 0.999,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655112607827814e-05,
+      "loss": 0.9501,
+      "step": 4717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3652559015755385e-05,
+      "loss": 0.9522,
+      "step": 4718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650005148808549e-05,
+      "loss": 0.9489,
+      "step": 4719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647451007179497e-05,
+      "loss": 0.9333,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644896591060444e-05,
+      "loss": 0.9417,
+      "step": 4721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642341900643623e-05,
+      "loss": 0.9554,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639786936121287e-05,
+      "loss": 1.0303,
+      "step": 4723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363723169768571e-05,
+      "loss": 0.9457,
+      "step": 4724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634676185529183e-05,
+      "loss": 0.9311,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632120399844033e-05,
+      "loss": 0.9547,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629564340822587e-05,
+      "loss": 0.9955,
+      "step": 4727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627008008657206e-05,
+      "loss": 0.9454,
+      "step": 4728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624451403540264e-05,
+      "loss": 0.9939,
+      "step": 4729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621894525664164e-05,
+      "loss": 0.9635,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361933737522132e-05,
+      "loss": 0.9532,
+      "step": 4731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616779952404175e-05,
+      "loss": 0.9297,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614222257405187e-05,
+      "loss": 0.9854,
+      "step": 4733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611664290416835e-05,
+      "loss": 0.9566,
+      "step": 4734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360910605163162e-05,
+      "loss": 0.9641,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606547541242066e-05,
+      "loss": 0.9758,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603988759440709e-05,
+      "loss": 0.977,
+      "step": 4737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3601429706420116e-05,
+      "loss": 0.9612,
+      "step": 4738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598870382372864e-05,
+      "loss": 0.9817,
+      "step": 4739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596310787491564e-05,
+      "loss": 0.949,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593750921968829e-05,
+      "loss": 0.9708,
+      "step": 4741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591190785997309e-05,
+      "loss": 0.9803,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588630379769667e-05,
+      "loss": 0.9444,
+      "step": 4743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586069703478586e-05,
+      "loss": 0.9531,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583508757316767e-05,
+      "loss": 0.9215,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580947541476942e-05,
+      "loss": 0.9209,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578386056151854e-05,
+      "loss": 0.9808,
+      "step": 4747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575824301534267e-05,
+      "loss": 0.9685,
+      "step": 4748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357326227781696e-05,
+      "loss": 0.9967,
+      "step": 4749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357069998519275e-05,
+      "loss": 0.9632,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356813742385446e-05,
+      "loss": 0.9382,
+      "step": 4751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356557459399493e-05,
+      "loss": 0.9874,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563011495807033e-05,
+      "loss": 0.988,
+      "step": 4753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560448129483656e-05,
+      "loss": 0.9637,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557884495217703e-05,
+      "loss": 0.9741,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555320593202101e-05,
+      "loss": 0.9717,
+      "step": 4756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552756423629798e-05,
+      "loss": 0.9526,
+      "step": 4757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3550191986693762e-05,
+      "loss": 0.9308,
+      "step": 4758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547627282586979e-05,
+      "loss": 0.9692,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354506231150246e-05,
+      "loss": 0.9115,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542497073633224e-05,
+      "loss": 0.9638,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353993156917233e-05,
+      "loss": 0.9609,
+      "step": 4762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353736579831284e-05,
+      "loss": 0.9661,
+      "step": 4763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534799761247843e-05,
+      "loss": 1.007,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532233458170446e-05,
+      "loss": 0.9643,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529666889273776e-05,
+      "loss": 0.9711,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527100054750981e-05,
+      "loss": 0.9548,
+      "step": 4767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524532954795233e-05,
+      "loss": 0.9675,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521965589599715e-05,
+      "loss": 0.9206,
+      "step": 4769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519397959357634e-05,
+      "loss": 0.9618,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516830064262226e-05,
+      "loss": 2.9733,
+      "step": 4771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514261904506732e-05,
+      "loss": 0.9818,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511693480284417e-05,
+      "loss": 1.0033,
+      "step": 4773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509124791788576e-05,
+      "loss": 0.9791,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506555839212511e-05,
+      "loss": 0.9348,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503986622749553e-05,
+      "loss": 0.9397,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501417142593046e-05,
+      "loss": 0.9729,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498847398936357e-05,
+      "loss": 0.9398,
+      "step": 4778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496277391972874e-05,
+      "loss": 0.9509,
+      "step": 4779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493707121896003e-05,
+      "loss": 0.9475,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491136588899174e-05,
+      "loss": 0.9691,
+      "step": 4781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488565793175827e-05,
+      "loss": 0.9652,
+      "step": 4782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485994734919434e-05,
+      "loss": 0.9844,
+      "step": 4783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483423414323476e-05,
+      "loss": 0.9887,
+      "step": 4784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480851831581463e-05,
+      "loss": 1.0052,
+      "step": 4785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478279986886915e-05,
+      "loss": 0.9737,
+      "step": 4786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347570788043338e-05,
+      "loss": 0.9655,
+      "step": 4787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473135512414423e-05,
+      "loss": 1.0212,
+      "step": 4788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3470562883023628e-05,
+      "loss": 0.993,
+      "step": 4789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467989992454598e-05,
+      "loss": 0.8928,
+      "step": 4790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465416840900956e-05,
+      "loss": 0.9586,
+      "step": 4791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3462843428556349e-05,
+      "loss": 0.9597,
+      "step": 4792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3460269755614437e-05,
+      "loss": 0.9729,
+      "step": 4793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3457695822268902e-05,
+      "loss": 3.0089,
+      "step": 4794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455121628713444e-05,
+      "loss": 0.9527,
+      "step": 4795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3452547175141793e-05,
+      "loss": 0.9737,
+      "step": 4796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344997246174768e-05,
+      "loss": 0.9827,
+      "step": 4797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447397488724874e-05,
+      "loss": 0.9646,
+      "step": 4798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444822256267152e-05,
+      "loss": 0.9812,
+      "step": 4799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442246764568314e-05,
+      "loss": 0.94,
+      "step": 4800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439671013822177e-05,
+      "loss": 2.8698,
+      "step": 4801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437095004222585e-05,
+      "loss": 0.9538,
+      "step": 4802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434518735963393e-05,
+      "loss": 0.9484,
+      "step": 4803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343194220923848e-05,
+      "loss": 0.9347,
+      "step": 4804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342936542424174e-05,
+      "loss": 0.9599,
+      "step": 4805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426788381167097e-05,
+      "loss": 3.0536,
+      "step": 4806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424211080208478e-05,
+      "loss": 0.9665,
+      "step": 4807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421633521559845e-05,
+      "loss": 0.9405,
+      "step": 4808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419055705415175e-05,
+      "loss": 0.9365,
+      "step": 4809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416477631968454e-05,
+      "loss": 0.9418,
+      "step": 4810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34138993014137e-05,
+      "loss": 0.9804,
+      "step": 4811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411320713944946e-05,
+      "loss": 0.9656,
+      "step": 4812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408741869756247e-05,
+      "loss": 0.9625,
+      "step": 4813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340616276904167e-05,
+      "loss": 0.9641,
+      "step": 4814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403583411995306e-05,
+      "loss": 0.9515,
+      "step": 4815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3401003798811273e-05,
+      "loss": 3.0627,
+      "step": 4816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398423929683693e-05,
+      "loss": 0.9365,
+      "step": 4817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395843804806714e-05,
+      "loss": 0.9552,
+      "step": 4818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393263424374508e-05,
+      "loss": 0.988,
+      "step": 4819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390682788581265e-05,
+      "loss": 0.9559,
+      "step": 4820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388101897621185e-05,
+      "loss": 0.9651,
+      "step": 4821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385520751688495e-05,
+      "loss": 0.9795,
+      "step": 4822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338293935097744e-05,
+      "loss": 0.9255,
+      "step": 4823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380357695682289e-05,
+      "loss": 0.9472,
+      "step": 4824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377775785997318e-05,
+      "loss": 0.9369,
+      "step": 4825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375193622116837e-05,
+      "loss": 0.9598,
+      "step": 4826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337261120423516e-05,
+      "loss": 0.9589,
+      "step": 4827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370028532546635e-05,
+      "loss": 0.9509,
+      "step": 4828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367445607245613e-05,
+      "loss": 0.9645,
+      "step": 4829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364862428526481e-05,
+      "loss": 0.9485,
+      "step": 4830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362278996583632e-05,
+      "loss": 0.9887,
+      "step": 4831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359695311611485e-05,
+      "loss": 0.9858,
+      "step": 4832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357111373804474e-05,
+      "loss": 1.0063,
+      "step": 4833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354527183357057e-05,
+      "loss": 3.003,
+      "step": 4834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351942740463705e-05,
+      "loss": 0.9433,
+      "step": 4835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349358045318912e-05,
+      "loss": 0.966,
+      "step": 4836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346773098117193e-05,
+      "loss": 0.9748,
+      "step": 4837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344187899053079e-05,
+      "loss": 0.9248,
+      "step": 4838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334160244832111e-05,
+      "loss": 0.9863,
+      "step": 4839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339016746115864e-05,
+      "loss": 1.0115,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333643079263193e-05,
+      "loss": 0.9722,
+      "step": 4841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333844588063911e-05,
+      "loss": 0.9814,
+      "step": 4842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331258132606432e-05,
+      "loss": 0.9696,
+      "step": 4843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328671426454139e-05,
+      "loss": 0.9522,
+      "step": 4844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3326084469801695e-05,
+      "loss": 2.9266,
+      "step": 4845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332349726284378e-05,
+      "loss": 0.998,
+      "step": 4846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33209098057751e-05,
+      "loss": 0.9425,
+      "step": 4847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318322098790374e-05,
+      "loss": 0.9792,
+      "step": 4848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3315734142084335e-05,
+      "loss": 0.9846,
+      "step": 4849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313145935851744e-05,
+      "loss": 0.9241,
+      "step": 4850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310557480287378e-05,
+      "loss": 0.9063,
+      "step": 4851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3307968775586033e-05,
+      "loss": 0.9449,
+      "step": 4852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305379821942519e-05,
+      "loss": 0.957,
+      "step": 4853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 0.9028,
+      "step": 4854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300201168608342e-05,
+      "loss": 0.9885,
+      "step": 4855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297611469307398e-05,
+      "loss": 0.962,
+      "step": 4856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295021521843729e-05,
+      "loss": 0.9582,
+      "step": 4857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292431326412244e-05,
+      "loss": 0.9664,
+      "step": 4858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289840883207867e-05,
+      "loss": 0.9249,
+      "step": 4859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287250192425542e-05,
+      "loss": 0.9546,
+      "step": 4860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328465925426023e-05,
+      "loss": 0.93,
+      "step": 4861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282068068906923e-05,
+      "loss": 0.9789,
+      "step": 4862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279476636560608e-05,
+      "loss": 0.9516,
+      "step": 4863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276884957416314e-05,
+      "loss": 1.0042,
+      "step": 4864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274293031669076e-05,
+      "loss": 0.9587,
+      "step": 4865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271700859513948e-05,
+      "loss": 0.9497,
+      "step": 4866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269108441146004e-05,
+      "loss": 0.8957,
+      "step": 4867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326651577676034e-05,
+      "loss": 1.0077,
+      "step": 4868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263922866552068e-05,
+      "loss": 0.9654,
+      "step": 4869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261329710716317e-05,
+      "loss": 0.9663,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258736309448235e-05,
+      "loss": 0.9673,
+      "step": 4871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325614266294299e-05,
+      "loss": 0.9795,
+      "step": 4872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3253548771395767e-05,
+      "loss": 0.9588,
+      "step": 4873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3250954635001771e-05,
+      "loss": 0.9406,
+      "step": 4874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248360253956225e-05,
+      "loss": 0.9607,
+      "step": 4875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324576562845437e-05,
+      "loss": 0.9688,
+      "step": 4876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324317075869146e-05,
+      "loss": 0.9966,
+      "step": 4877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324057564486278e-05,
+      "loss": 0.974,
+      "step": 4878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237980287163623e-05,
+      "loss": 0.9986,
+      "step": 4879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235384685789302e-05,
+      "loss": 0.9568,
+      "step": 4880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323278884093515e-05,
+      "loss": 0.9636,
+      "step": 4881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230192752796524e-05,
+      "loss": 1.0007,
+      "step": 4882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227596421568787e-05,
+      "loss": 2.8844,
+      "step": 4883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224999847447326e-05,
+      "loss": 1.0107,
+      "step": 4884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222403030627551e-05,
+      "loss": 0.9567,
+      "step": 4885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219805971304885e-05,
+      "loss": 0.9879,
+      "step": 4886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321720866967477e-05,
+      "loss": 0.935,
+      "step": 4887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214611125932666e-05,
+      "loss": 0.9421,
+      "step": 4888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321201334027405e-05,
+      "loss": 0.9459,
+      "step": 4889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209415312894427e-05,
+      "loss": 0.9338,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206817043989301e-05,
+      "loss": 0.937,
+      "step": 4891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204218533754216e-05,
+      "loss": 0.9574,
+      "step": 4892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201619782384719e-05,
+      "loss": 0.9673,
+      "step": 4893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319902079007638e-05,
+      "loss": 0.9575,
+      "step": 4894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196421557024786e-05,
+      "loss": 0.9333,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193822083425545e-05,
+      "loss": 1.0063,
+      "step": 4896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191222369474278e-05,
+      "loss": 0.9399,
+      "step": 4897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318862241536663e-05,
+      "loss": 0.9679,
+      "step": 4898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318602222129826e-05,
+      "loss": 0.9612,
+      "step": 4899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318342178746485e-05,
+      "loss": 0.9796,
+      "step": 4900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318082111406209e-05,
+      "loss": 0.9991,
+      "step": 4901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31782202012857e-05,
+      "loss": 0.9496,
+      "step": 4902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175619049331409e-05,
+      "loss": 0.9381,
+      "step": 4903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173017658394973e-05,
+      "loss": 0.9588,
+      "step": 4904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317041602867215e-05,
+      "loss": 0.9297,
+      "step": 4905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167814160358735e-05,
+      "loss": 0.9966,
+      "step": 4906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165212053650533e-05,
+      "loss": 0.9741,
+      "step": 4907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162609708743363e-05,
+      "loss": 0.9453,
+      "step": 4908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160007125833063e-05,
+      "loss": 0.9848,
+      "step": 4909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157404305115492e-05,
+      "loss": 1.01,
+      "step": 4910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154801246786533e-05,
+      "loss": 2.8941,
+      "step": 4911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152197951042069e-05,
+      "loss": 0.9973,
+      "step": 4912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149594418078025e-05,
+      "loss": 0.9297,
+      "step": 4913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146990648090316e-05,
+      "loss": 0.9525,
+      "step": 4914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31443866412749e-05,
+      "loss": 0.9383,
+      "step": 4915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141782397827739e-05,
+      "loss": 0.9547,
+      "step": 4916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139177917944816e-05,
+      "loss": 0.925,
+      "step": 4917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136573201822129e-05,
+      "loss": 0.9535,
+      "step": 4918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133968249655701e-05,
+      "loss": 0.9652,
+      "step": 4919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131363061641569e-05,
+      "loss": 0.9994,
+      "step": 4920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128757637975786e-05,
+      "loss": 2.9679,
+      "step": 4921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312615197885442e-05,
+      "loss": 0.9561,
+      "step": 4922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123546084473564e-05,
+      "loss": 0.9653,
+      "step": 4923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120939955029328e-05,
+      "loss": 0.9249,
+      "step": 4924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118333590717833e-05,
+      "loss": 0.9874,
+      "step": 4925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115726991735223e-05,
+      "loss": 0.9523,
+      "step": 4926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113120158277657e-05,
+      "loss": 0.9421,
+      "step": 4927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3110513090541317e-05,
+      "loss": 0.975,
+      "step": 4928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107905788722392e-05,
+      "loss": 0.9816,
+      "step": 4929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31052982530171e-05,
+      "loss": 0.9453,
+      "step": 4930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102690483621674e-05,
+      "loss": 0.9631,
+      "step": 4931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310008248073236e-05,
+      "loss": 0.9544,
+      "step": 4932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309747424454542e-05,
+      "loss": 0.9733,
+      "step": 4933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094865775257142e-05,
+      "loss": 0.9422,
+      "step": 4934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309225707306383e-05,
+      "loss": 0.9245,
+      "step": 4935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089648138161798e-05,
+      "loss": 0.9875,
+      "step": 4936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087038970747379e-05,
+      "loss": 0.959,
+      "step": 4937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084429571016932e-05,
+      "loss": 0.9022,
+      "step": 4938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308181993916683e-05,
+      "loss": 0.9392,
+      "step": 4939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3079210075393456e-05,
+      "loss": 0.9737,
+      "step": 4940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076599979893221e-05,
+      "loss": 0.9955,
+      "step": 4941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073989652862542e-05,
+      "loss": 3.0122,
+      "step": 4942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071379094497866e-05,
+      "loss": 3.0057,
+      "step": 4943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068768304995651e-05,
+      "loss": 1.0132,
+      "step": 4944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066157284552371e-05,
+      "loss": 2.832,
+      "step": 4945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306354603336452e-05,
+      "loss": 0.961,
+      "step": 4946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060934551628603e-05,
+      "loss": 0.9522,
+      "step": 4947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058322839541158e-05,
+      "loss": 0.9631,
+      "step": 4948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055710897298723e-05,
+      "loss": 2.8236,
+      "step": 4949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053098725097862e-05,
+      "loss": 0.9468,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050486323135151e-05,
+      "loss": 0.9382,
+      "step": 4951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047873691607196e-05,
+      "loss": 1.0161,
+      "step": 4952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045260830710608e-05,
+      "loss": 0.9463,
+      "step": 4953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042647740642012e-05,
+      "loss": 1.0365,
+      "step": 4954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040034421598064e-05,
+      "loss": 0.9279,
+      "step": 4955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037420873775429e-05,
+      "loss": 1.0347,
+      "step": 4956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034807097370788e-05,
+      "loss": 0.9565,
+      "step": 4957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032193092580845e-05,
+      "loss": 0.9207,
+      "step": 4958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029578859602313e-05,
+      "loss": 0.9815,
+      "step": 4959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026964398631932e-05,
+      "loss": 0.9309,
+      "step": 4960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024349709866448e-05,
+      "loss": 0.9777,
+      "step": 4961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302173479350264e-05,
+      "loss": 1.0087,
+      "step": 4962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301911964973728e-05,
+      "loss": 0.9701,
+      "step": 4963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016504278767183e-05,
+      "loss": 0.9596,
+      "step": 4964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013888680789167e-05,
+      "loss": 0.9586,
+      "step": 4965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011272856000069e-05,
+      "loss": 0.9392,
+      "step": 4966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300865680459674e-05,
+      "loss": 0.9082,
+      "step": 4967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006040526776054e-05,
+      "loss": 0.9332,
+      "step": 4968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003424022734904e-05,
+      "loss": 0.94,
+      "step": 4969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3000807292670192e-05,
+      "loss": 0.9297,
+      "step": 4970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299819033677884e-05,
+      "loss": 0.9662,
+      "step": 4971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2995573155257788e-05,
+      "loss": 0.9564,
+      "step": 4972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992955748303998e-05,
+      "loss": 1.0012,
+      "step": 4973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990338116114436e-05,
+      "loss": 0.9661,
+      "step": 4974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987720258886094e-05,
+      "loss": 0.9608,
+      "step": 4975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985102176815987e-05,
+      "loss": 0.9798,
+      "step": 4976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982483870101132e-05,
+      "loss": 0.9488,
+      "step": 4977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297986533893857e-05,
+      "loss": 0.9658,
+      "step": 4978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977246583525367e-05,
+      "loss": 0.9614,
+      "step": 4979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974627604058592e-05,
+      "loss": 0.9312,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972008400735339e-05,
+      "loss": 0.9226,
+      "step": 4981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969388973752712e-05,
+      "loss": 0.9279,
+      "step": 4982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966769323307846e-05,
+      "loss": 0.958,
+      "step": 4983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964149449597876e-05,
+      "loss": 0.9601,
+      "step": 4984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961529352819964e-05,
+      "loss": 0.97,
+      "step": 4985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958909033171288e-05,
+      "loss": 0.9285,
+      "step": 4986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956288490849039e-05,
+      "loss": 2.8131,
+      "step": 4987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953667726050422e-05,
+      "loss": 0.9258,
+      "step": 4988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951046738972673e-05,
+      "loss": 0.9447,
+      "step": 4989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948425529813032e-05,
+      "loss": 0.9493,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945804098768754e-05,
+      "loss": 0.9319,
+      "step": 4991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943182446037117e-05,
+      "loss": 0.9554,
+      "step": 4992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940560571815417e-05,
+      "loss": 0.9626,
+      "step": 4993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937938476300966e-05,
+      "loss": 0.9717,
+      "step": 4994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935316159691084e-05,
+      "loss": 0.9215,
+      "step": 4995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932693622183118e-05,
+      "loss": 0.9534,
+      "step": 4996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930070863974433e-05,
+      "loss": 0.9604,
+      "step": 4997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927447885262396e-05,
+      "loss": 0.9713,
+      "step": 4998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924824686244404e-05,
+      "loss": 0.9009,
+      "step": 4999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922201267117865e-05,
+      "loss": 0.966,
+      "step": 5000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291957762808021e-05,
+      "loss": 0.918,
+      "step": 5001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916953769328879e-05,
+      "loss": 0.9638,
+      "step": 5002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914329691061327e-05,
+      "loss": 0.9316,
+      "step": 5003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911705393475036e-05,
+      "loss": 0.9336,
+      "step": 5004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909080876767497e-05,
+      "loss": 0.9293,
+      "step": 5005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906456141136215e-05,
+      "loss": 0.9746,
+      "step": 5006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903831186778722e-05,
+      "loss": 0.9965,
+      "step": 5007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290120601389255e-05,
+      "loss": 2.9442,
+      "step": 5008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898580622675267e-05,
+      "loss": 0.9493,
+      "step": 5009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895955013324442e-05,
+      "loss": 0.9739,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893329186037668e-05,
+      "loss": 0.9368,
+      "step": 5011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890703141012554e-05,
+      "loss": 0.9238,
+      "step": 5012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888076878446716e-05,
+      "loss": 0.9384,
+      "step": 5013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885450398537804e-05,
+      "loss": 0.9378,
+      "step": 5014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288282370148347e-05,
+      "loss": 0.9735,
+      "step": 5015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880196787481388e-05,
+      "loss": 0.9438,
+      "step": 5016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877569656729242e-05,
+      "loss": 0.956,
+      "step": 5017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2874942309424747e-05,
+      "loss": 0.9329,
+      "step": 5018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872314745765619e-05,
+      "loss": 0.9852,
+      "step": 5019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869686965949595e-05,
+      "loss": 2.9063,
+      "step": 5020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867058970174429e-05,
+      "loss": 0.9793,
+      "step": 5021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864430758637899e-05,
+      "loss": 0.9804,
+      "step": 5022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286180233153778e-05,
+      "loss": 2.8734,
+      "step": 5023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859173689071887e-05,
+      "loss": 0.9162,
+      "step": 5024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856544831438032e-05,
+      "loss": 0.9792,
+      "step": 5025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853915758834053e-05,
+      "loss": 0.9577,
+      "step": 5026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28512864714578e-05,
+      "loss": 0.9782,
+      "step": 5027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848656969507143e-05,
+      "loss": 0.9644,
+      "step": 5028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846027253179963e-05,
+      "loss": 0.9447,
+      "step": 5029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843397322674162e-05,
+      "loss": 0.9891,
+      "step": 5030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2840767178187657e-05,
+      "loss": 0.9553,
+      "step": 5031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283813681991838e-05,
+      "loss": 0.9188,
+      "step": 5032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835506248064275e-05,
+      "loss": 0.9585,
+      "step": 5033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283287546282331e-05,
+      "loss": 0.9663,
+      "step": 5034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283024446439347e-05,
+      "loss": 0.9248,
+      "step": 5035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827613252972749e-05,
+      "loss": 0.9232,
+      "step": 5036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824981828759153e-05,
+      "loss": 0.9972,
+      "step": 5037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822350191950715e-05,
+      "loss": 0.9635,
+      "step": 5038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2819718342745481e-05,
+      "loss": 0.9589,
+      "step": 5039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817086281341514e-05,
+      "loss": 0.9344,
+      "step": 5040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281445400793688e-05,
+      "loss": 2.8399,
+      "step": 5041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811821522729684e-05,
+      "loss": 0.9168,
+      "step": 5042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280918882591803e-05,
+      "loss": 0.971,
+      "step": 5043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280655591770004e-05,
+      "loss": 0.9772,
+      "step": 5044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803922798273853e-05,
+      "loss": 2.8999,
+      "step": 5045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801289467837632e-05,
+      "loss": 0.9526,
+      "step": 5046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2798655926589545e-05,
+      "loss": 0.9826,
+      "step": 5047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796022174727775e-05,
+      "loss": 0.9786,
+      "step": 5048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793388212450538e-05,
+      "loss": 0.9553,
+      "step": 5049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790754039956039e-05,
+      "loss": 0.9277,
+      "step": 5050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788119657442524e-05,
+      "loss": 0.9766,
+      "step": 5051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785485065108243e-05,
+      "loss": 0.9663,
+      "step": 5052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278285026315146e-05,
+      "loss": 0.9504,
+      "step": 5053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780215251770457e-05,
+      "loss": 0.9409,
+      "step": 5054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777580031163534e-05,
+      "loss": 0.9491,
+      "step": 5055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774944601529007e-05,
+      "loss": 0.9887,
+      "step": 5056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772308963065206e-05,
+      "loss": 0.9489,
+      "step": 5057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2769673115970473e-05,
+      "loss": 0.9269,
+      "step": 5058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767037060443173e-05,
+      "loss": 0.9415,
+      "step": 5059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764400796681681e-05,
+      "loss": 0.9589,
+      "step": 5060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761764324884391e-05,
+      "loss": 0.96,
+      "step": 5061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759127645249711e-05,
+      "loss": 0.9298,
+      "step": 5062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275649075797607e-05,
+      "loss": 0.9381,
+      "step": 5063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753853663261898e-05,
+      "loss": 3.065,
+      "step": 5064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751216361305659e-05,
+      "loss": 0.9914,
+      "step": 5065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748578852305817e-05,
+      "loss": 0.9436,
+      "step": 5066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2745941136460865e-05,
+      "loss": 0.9774,
+      "step": 5067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27433032139693e-05,
+      "loss": 0.9646,
+      "step": 5068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740665085029642e-05,
+      "loss": 0.9511,
+      "step": 5069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738026749840425e-05,
+      "loss": 0.9615,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735388208600195e-05,
+      "loss": 0.9294,
+      "step": 5071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732749461507519e-05,
+      "loss": 0.9871,
+      "step": 5072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730110508760976e-05,
+      "loss": 0.9748,
+      "step": 5073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727471350559158e-05,
+      "loss": 0.9298,
+      "step": 5074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.272483198710068e-05,
+      "loss": 1.0201,
+      "step": 5075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722192418584165e-05,
+      "loss": 0.9615,
+      "step": 5076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271955264520826e-05,
+      "loss": 0.9357,
+      "step": 5077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716912667171613e-05,
+      "loss": 0.9515,
+      "step": 5078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714272484672901e-05,
+      "loss": 0.9984,
+      "step": 5079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711632097910813e-05,
+      "loss": 0.9381,
+      "step": 5080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708991507084055e-05,
+      "loss": 0.9545,
+      "step": 5081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706350712391335e-05,
+      "loss": 0.9629,
+      "step": 5082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703709714031393e-05,
+      "loss": 0.962,
+      "step": 5083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701068512202983e-05,
+      "loss": 0.9662,
+      "step": 5084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2698427107104862e-05,
+      "loss": 0.9798,
+      "step": 5085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695785498935813e-05,
+      "loss": 0.9158,
+      "step": 5086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269314368789463e-05,
+      "loss": 0.9839,
+      "step": 5087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690501674180124e-05,
+      "loss": 0.9076,
+      "step": 5088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687859457991118e-05,
+      "loss": 0.9573,
+      "step": 5089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685217039526459e-05,
+      "loss": 0.9516,
+      "step": 5090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682574418984996e-05,
+      "loss": 0.9399,
+      "step": 5091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679931596565605e-05,
+      "loss": 0.9628,
+      "step": 5092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677288572467169e-05,
+      "loss": 0.9648,
+      "step": 5093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267464534688859e-05,
+      "loss": 0.9612,
+      "step": 5094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2672001920028787e-05,
+      "loss": 0.9837,
+      "step": 5095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266935829208669e-05,
+      "loss": 0.959,
+      "step": 5096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666714463261247e-05,
+      "loss": 0.8751,
+      "step": 5097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664070433751423e-05,
+      "loss": 0.9551,
+      "step": 5098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661426203756186e-05,
+      "loss": 0.9658,
+      "step": 5099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658781773474535e-05,
+      "loss": 0.9425,
+      "step": 5100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656137143105483e-05,
+      "loss": 0.939,
+      "step": 5101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653492312848041e-05,
+      "loss": 0.9291,
+      "step": 5102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265084728290125e-05,
+      "loss": 0.9489,
+      "step": 5103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648202053464166e-05,
+      "loss": 0.9784,
+      "step": 5104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645556624735855e-05,
+      "loss": 0.9335,
+      "step": 5105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642910996915401e-05,
+      "loss": 0.9449,
+      "step": 5106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640265170201895e-05,
+      "loss": 0.9717,
+      "step": 5107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637619144794458e-05,
+      "loss": 0.957,
+      "step": 5108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634972920892215e-05,
+      "loss": 1.0062,
+      "step": 5109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26323264986943e-05,
+      "loss": 0.9854,
+      "step": 5110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629679878399881e-05,
+      "loss": 0.9897,
+      "step": 5111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262703306020813e-05,
+      "loss": 0.9642,
+      "step": 5112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624386044318226e-05,
+      "loss": 1.0051,
+      "step": 5113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621738830929376e-05,
+      "loss": 0.9552,
+      "step": 5114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619091420240795e-05,
+      "loss": 1.0024,
+      "step": 5115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261644381245172e-05,
+      "loss": 0.9531,
+      "step": 5116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613796007761387e-05,
+      "loss": 0.8784,
+      "step": 5117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261114800636907e-05,
+      "loss": 0.9728,
+      "step": 5118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608499808474038e-05,
+      "loss": 0.9616,
+      "step": 5119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605851414275581e-05,
+      "loss": 0.9585,
+      "step": 5120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2603202823973004e-05,
+      "loss": 0.9991,
+      "step": 5121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600554037765633e-05,
+      "loss": 0.941,
+      "step": 5122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25979050558528e-05,
+      "loss": 0.938,
+      "step": 5123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595255878433854e-05,
+      "loss": 0.9626,
+      "step": 5124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259260650570816e-05,
+      "loss": 0.9723,
+      "step": 5125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25899569378751e-05,
+      "loss": 0.9571,
+      "step": 5126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587307175134065e-05,
+      "loss": 0.8943,
+      "step": 5127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584657217684461e-05,
+      "loss": 3.0301,
+      "step": 5128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582007065725721e-05,
+      "loss": 0.9461,
+      "step": 5129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579356719457276e-05,
+      "loss": 0.9419,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576706179078577e-05,
+      "loss": 0.9309,
+      "step": 5131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574055444789095e-05,
+      "loss": 0.9654,
+      "step": 5132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257140451678831e-05,
+      "loss": 0.9173,
+      "step": 5133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568753395275721e-05,
+      "loss": 0.9439,
+      "step": 5134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566102080450832e-05,
+      "loss": 0.966,
+      "step": 5135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2563450572513177e-05,
+      "loss": 0.9923,
+      "step": 5136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560798871662292e-05,
+      "loss": 0.9055,
+      "step": 5137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558146978097732e-05,
+      "loss": 0.9862,
+      "step": 5138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555494892019064e-05,
+      "loss": 2.8271,
+      "step": 5139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552842613625876e-05,
+      "loss": 0.9391,
+      "step": 5140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550190143117761e-05,
+      "loss": 0.9126,
+      "step": 5141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547537480694334e-05,
+      "loss": 0.9937,
+      "step": 5142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544884626555225e-05,
+      "loss": 0.9192,
+      "step": 5143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542231580900069e-05,
+      "loss": 0.9519,
+      "step": 5144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539578343928525e-05,
+      "loss": 1.0031,
+      "step": 5145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536924915840264e-05,
+      "loss": 0.9641,
+      "step": 5146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534271296834968e-05,
+      "loss": 0.978,
+      "step": 5147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531617487112338e-05,
+      "loss": 0.9703,
+      "step": 5148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528963486872085e-05,
+      "loss": 0.9706,
+      "step": 5149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526309296313941e-05,
+      "loss": 0.9641,
+      "step": 5150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523654915637645e-05,
+      "loss": 0.9486,
+      "step": 5151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521000345042951e-05,
+      "loss": 1.0092,
+      "step": 5152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518345584729632e-05,
+      "loss": 0.9183,
+      "step": 5153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515690634897473e-05,
+      "loss": 0.9838,
+      "step": 5154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513035495746272e-05,
+      "loss": 0.9876,
+      "step": 5155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510380167475844e-05,
+      "loss": 0.9876,
+      "step": 5156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507724650286014e-05,
+      "loss": 0.9691,
+      "step": 5157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505068944376629e-05,
+      "loss": 0.9483,
+      "step": 5158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502413049947535e-05,
+      "loss": 0.9759,
+      "step": 5159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499756967198611e-05,
+      "loss": 0.9705,
+      "step": 5160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497100696329737e-05,
+      "loss": 0.9261,
+      "step": 5161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494444237540817e-05,
+      "loss": 0.9599,
+      "step": 5162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491787591031754e-05,
+      "loss": 1.0031,
+      "step": 5163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489130757002484e-05,
+      "loss": 0.9841,
+      "step": 5164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486473735652942e-05,
+      "loss": 0.9714,
+      "step": 5165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483816527183084e-05,
+      "loss": 0.9885,
+      "step": 5166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481159131792881e-05,
+      "loss": 0.9459,
+      "step": 5167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478501549682318e-05,
+      "loss": 0.9656,
+      "step": 5168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475843781051387e-05,
+      "loss": 0.9607,
+      "step": 5169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473185826100098e-05,
+      "loss": 0.9911,
+      "step": 5170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2470527685028485e-05,
+      "loss": 0.9628,
+      "step": 5171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467869358036578e-05,
+      "loss": 0.9412,
+      "step": 5172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465210845324432e-05,
+      "loss": 0.9948,
+      "step": 5173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246255214709212e-05,
+      "loss": 0.9735,
+      "step": 5174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459893263539717e-05,
+      "loss": 0.9126,
+      "step": 5175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245723419486732e-05,
+      "loss": 0.9619,
+      "step": 5176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454574941275039e-05,
+      "loss": 2.8542,
+      "step": 5177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451915502962997e-05,
+      "loss": 2.873,
+      "step": 5178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244925588013133e-05,
+      "loss": 0.93,
+      "step": 5179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446596072980185e-05,
+      "loss": 0.9241,
+      "step": 5180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2443936081709733e-05,
+      "loss": 0.958,
+      "step": 5181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244127590652015e-05,
+      "loss": 0.9884,
+      "step": 5182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438615547611626e-05,
+      "loss": 3.0202,
+      "step": 5183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435955005184373e-05,
+      "loss": 0.9423,
+      "step": 5184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433294279438604e-05,
+      "loss": 0.9602,
+      "step": 5185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430633370574554e-05,
+      "loss": 0.9328,
+      "step": 5186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427972278792472e-05,
+      "loss": 0.9353,
+      "step": 5187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425311004292624e-05,
+      "loss": 0.9617,
+      "step": 5188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422649547275278e-05,
+      "loss": 0.9378,
+      "step": 5189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419987907940722e-05,
+      "loss": 0.97,
+      "step": 5190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417326086489264e-05,
+      "loss": 0.9447,
+      "step": 5191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414664083121217e-05,
+      "loss": 2.9096,
+      "step": 5192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412001898036911e-05,
+      "loss": 0.9349,
+      "step": 5193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240933953143669e-05,
+      "loss": 0.9071,
+      "step": 5194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406676983520913e-05,
+      "loss": 0.9248,
+      "step": 5195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240401425448995e-05,
+      "loss": 0.9891,
+      "step": 5196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240135134454418e-05,
+      "loss": 0.9527,
+      "step": 5197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398688253884009e-05,
+      "loss": 0.9213,
+      "step": 5198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396024982709845e-05,
+      "loss": 0.9911,
+      "step": 5199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393361531222114e-05,
+      "loss": 0.9696,
+      "step": 5200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390697899621252e-05,
+      "loss": 0.9661,
+      "step": 5201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388034088107714e-05,
+      "loss": 0.9362,
+      "step": 5202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385370096881968e-05,
+      "loss": 0.9696,
+      "step": 5203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382705926144487e-05,
+      "loss": 2.9069,
+      "step": 5204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380041576095773e-05,
+      "loss": 0.9281,
+      "step": 5205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377377046936325e-05,
+      "loss": 0.962,
+      "step": 5206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374712338866665e-05,
+      "loss": 0.9686,
+      "step": 5207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372047452087328e-05,
+      "loss": 0.9549,
+      "step": 5208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369382386798859e-05,
+      "loss": 0.9423,
+      "step": 5209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366717143201819e-05,
+      "loss": 0.9621,
+      "step": 5210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236405172149678e-05,
+      "loss": 0.982,
+      "step": 5211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361386121884333e-05,
+      "loss": 0.9682,
+      "step": 5212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358720344565077e-05,
+      "loss": 0.9119,
+      "step": 5213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356054389739623e-05,
+      "loss": 0.9477,
+      "step": 5214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353388257608598e-05,
+      "loss": 0.9937,
+      "step": 5215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350721948372648e-05,
+      "loss": 0.9536,
+      "step": 5216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348055462232427e-05,
+      "loss": 0.9487,
+      "step": 5217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345388799388592e-05,
+      "loss": 0.9871,
+      "step": 5218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342721960041835e-05,
+      "loss": 0.9486,
+      "step": 5219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340054944392844e-05,
+      "loss": 0.9607,
+      "step": 5220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337387752642327e-05,
+      "loss": 0.9396,
+      "step": 5221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2334720384991003e-05,
+      "loss": 0.9679,
+      "step": 5222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233205284163961e-05,
+      "loss": 0.9656,
+      "step": 5223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329385122788891e-05,
+      "loss": 0.9256,
+      "step": 5224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326717228639607e-05,
+      "loss": 0.9346,
+      "step": 5225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232404915939253e-05,
+      "loss": 0.9293,
+      "step": 5226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321380915248446e-05,
+      "loss": 0.9658,
+      "step": 5227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231871249640816e-05,
+      "loss": 0.9232,
+      "step": 5228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316043903072477e-05,
+      "loss": 0.9676,
+      "step": 5229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313375135442226e-05,
+      "loss": 0.9558,
+      "step": 5230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231070619371825e-05,
+      "loss": 0.9395,
+      "step": 5231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308037078101394e-05,
+      "loss": 0.9709,
+      "step": 5232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305367788792526e-05,
+      "loss": 0.9717,
+      "step": 5233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302698325992527e-05,
+      "loss": 0.9485,
+      "step": 5234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300028689902283e-05,
+      "loss": 0.9269,
+      "step": 5235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22973588807227e-05,
+      "loss": 0.9636,
+      "step": 5236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.22946888986547e-05,
+      "loss": 0.9476,
+      "step": 5237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292018743899208e-05,
+      "loss": 0.9095,
+      "step": 5238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289348416657166e-05,
+      "loss": 0.9496,
+      "step": 5239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286677917129534e-05,
+      "loss": 0.9489,
+      "step": 5240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284007245517283e-05,
+      "loss": 0.9829,
+      "step": 5241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228133640202139e-05,
+      "loss": 0.946,
+      "step": 5242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278665386842851e-05,
+      "loss": 1.0234,
+      "step": 5243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275994200182679e-05,
+      "loss": 0.988,
+      "step": 5244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273322842241892e-05,
+      "loss": 1.0057,
+      "step": 5245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227065131322152e-05,
+      "loss": 0.9503,
+      "step": 5246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267979613322614e-05,
+      "loss": 0.9184,
+      "step": 5247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265307742746232e-05,
+      "loss": 0.9352,
+      "step": 5248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262635701693448e-05,
+      "loss": 0.9272,
+      "step": 5249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259963490365347e-05,
+      "loss": 1.0216,
+      "step": 5250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257291108963026e-05,
+      "loss": 0.9423,
+      "step": 5251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254618557687596e-05,
+      "loss": 0.9531,
+      "step": 5252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225194583674018e-05,
+      "loss": 0.912,
+      "step": 5253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249272946321917e-05,
+      "loss": 0.9314,
+      "step": 5254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246599886633951e-05,
+      "loss": 0.9209,
+      "step": 5255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224392665787745e-05,
+      "loss": 0.9555,
+      "step": 5256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241253260253584e-05,
+      "loss": 0.9694,
+      "step": 5257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238579693963544e-05,
+      "loss": 0.9435,
+      "step": 5258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235905959208527e-05,
+      "loss": 0.9481,
+      "step": 5259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233232056189746e-05,
+      "loss": 0.9341,
+      "step": 5260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230557985108428e-05,
+      "loss": 0.9444,
+      "step": 5261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227883746165815e-05,
+      "loss": 0.931,
+      "step": 5262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.9312,
+      "step": 5263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222534765501693e-05,
+      "loss": 0.973,
+      "step": 5264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221986002418273e-05,
+      "loss": 0.9652,
+      "step": 5265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217185115807549e-05,
+      "loss": 0.9574,
+      "step": 5266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214510040577442e-05,
+      "loss": 0.9837,
+      "step": 5267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2211834798693731e-05,
+      "loss": 0.9778,
+      "step": 5268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209159390357739e-05,
+      "loss": 2.6776,
+      "step": 5269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206483815770804e-05,
+      "loss": 0.9642,
+      "step": 5270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220380807513428e-05,
+      "loss": 0.9887,
+      "step": 5271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201132168649528e-05,
+      "loss": 0.9556,
+      "step": 5272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198456096517924e-05,
+      "loss": 2.7468,
+      "step": 5273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195779858940857e-05,
+      "loss": 0.9107,
+      "step": 5274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219310345611973e-05,
+      "loss": 0.9321,
+      "step": 5275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190426888255953e-05,
+      "loss": 0.9625,
+      "step": 5276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187750155550952e-05,
+      "loss": 0.9761,
+      "step": 5277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218507325820617e-05,
+      "loss": 0.9913,
+      "step": 5278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182396196423057e-05,
+      "loss": 0.9329,
+      "step": 5279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217971897040307e-05,
+      "loss": 0.9666,
+      "step": 5280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177041580347688e-05,
+      "loss": 0.9384,
+      "step": 5281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174364026458404e-05,
+      "loss": 0.9319,
+      "step": 5282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217168630893671e-05,
+      "loss": 0.9648,
+      "step": 5283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169008427984119e-05,
+      "loss": 0.9406,
+      "step": 5284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166330383802161e-05,
+      "loss": 0.9388,
+      "step": 5285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216365217659237e-05,
+      "loss": 0.948,
+      "step": 5286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2160973806556298e-05,
+      "loss": 0.9131,
+      "step": 5287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158295273895501e-05,
+      "loss": 0.9496,
+      "step": 5288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155616578811558e-05,
+      "loss": 2.9602,
+      "step": 5289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152937721506052e-05,
+      "loss": 0.9612,
+      "step": 5290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150258702180582e-05,
+      "loss": 0.9074,
+      "step": 5291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147579521036759e-05,
+      "loss": 0.9224,
+      "step": 5292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2144900178276205e-05,
+      "loss": 0.9201,
+      "step": 5293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142220674100553e-05,
+      "loss": 0.9366,
+      "step": 5294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139541008711453e-05,
+      "loss": 0.9887,
+      "step": 5295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136861182310564e-05,
+      "loss": 1.0158,
+      "step": 5296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134181195099555e-05,
+      "loss": 0.9506,
+      "step": 5297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131501047280107e-05,
+      "loss": 0.9498,
+      "step": 5298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128820739053922e-05,
+      "loss": 0.95,
+      "step": 5299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126140270622703e-05,
+      "loss": 0.9334,
+      "step": 5300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123459642188171e-05,
+      "loss": 0.9503,
+      "step": 5301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120778853952053e-05,
+      "loss": 0.9745,
+      "step": 5302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118097906116103e-05,
+      "loss": 0.9528,
+      "step": 5303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115416798882065e-05,
+      "loss": 0.9806,
+      "step": 5304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112735532451706e-05,
+      "loss": 0.9448,
+      "step": 5305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110054107026818e-05,
+      "loss": 0.9125,
+      "step": 5306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107372522809186e-05,
+      "loss": 0.9244,
+      "step": 5307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210469078000061e-05,
+      "loss": 0.9428,
+      "step": 5308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102008878802907e-05,
+      "loss": 0.9069,
+      "step": 5309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099326819417907e-05,
+      "loss": 0.9631,
+      "step": 5310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096644602047447e-05,
+      "loss": 0.9557,
+      "step": 5311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093962226893378e-05,
+      "loss": 0.8882,
+      "step": 5312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091279694157563e-05,
+      "loss": 0.9561,
+      "step": 5313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088597004041878e-05,
+      "loss": 0.9994,
+      "step": 5314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085914156748208e-05,
+      "loss": 0.9347,
+      "step": 5315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083231152478454e-05,
+      "loss": 0.9391,
+      "step": 5316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080547991434522e-05,
+      "loss": 0.9554,
+      "step": 5317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207786467381834e-05,
+      "loss": 0.957,
+      "step": 5318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075181199831834e-05,
+      "loss": 0.9854,
+      "step": 5319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072497569676957e-05,
+      "loss": 1.0024,
+      "step": 5320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069813783555664e-05,
+      "loss": 0.9648,
+      "step": 5321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067129841669922e-05,
+      "loss": 0.9652,
+      "step": 5322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064445744221713e-05,
+      "loss": 0.9659,
+      "step": 5323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061761491413032e-05,
+      "loss": 0.8793,
+      "step": 5324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2059077083445881e-05,
+      "loss": 0.9712,
+      "step": 5325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056392520522273e-05,
+      "loss": 0.9223,
+      "step": 5326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053707802844243e-05,
+      "loss": 0.9134,
+      "step": 5327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051022930613826e-05,
+      "loss": 0.9424,
+      "step": 5328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204833790403307e-05,
+      "loss": 0.932,
+      "step": 5329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204565272330404e-05,
+      "loss": 0.9436,
+      "step": 5330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042967388628812e-05,
+      "loss": 0.9001,
+      "step": 5331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040281900209471e-05,
+      "loss": 0.972,
+      "step": 5332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203759625824811e-05,
+      "loss": 0.9288,
+      "step": 5333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034910462946843e-05,
+      "loss": 0.9695,
+      "step": 5334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032224514507789e-05,
+      "loss": 0.9806,
+      "step": 5335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029538413133076e-05,
+      "loss": 0.9311,
+      "step": 5336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026852159024853e-05,
+      "loss": 0.9238,
+      "step": 5337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202416575238527e-05,
+      "loss": 0.9856,
+      "step": 5338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021479193416502e-05,
+      "loss": 0.9302,
+      "step": 5339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018792482320713e-05,
+      "loss": 0.931,
+      "step": 5340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016105619300105e-05,
+      "loss": 0.9717,
+      "step": 5341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013418604556868e-05,
+      "loss": 0.9941,
+      "step": 5342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010731438293222e-05,
+      "loss": 0.9644,
+      "step": 5343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200804412071139e-05,
+      "loss": 0.8977,
+      "step": 5344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005356652013604e-05,
+      "loss": 0.9428,
+      "step": 5345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200266903240211e-05,
+      "loss": 0.9559,
+      "step": 5346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999981262079165e-05,
+      "loss": 0.9586,
+      "step": 5347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997293341247042e-05,
+      "loss": 0.9522,
+      "step": 5348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994605270108021e-05,
+      "loss": 0.9385,
+      "step": 5349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991917048864385e-05,
+      "loss": 0.9513,
+      "step": 5350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198922867771845e-05,
+      "loss": 0.9317,
+      "step": 5351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986540156872522e-05,
+      "loss": 0.9512,
+      "step": 5352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983851486528926e-05,
+      "loss": 0.9648,
+      "step": 5353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981162666890001e-05,
+      "loss": 0.8992,
+      "step": 5354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978473698158094e-05,
+      "loss": 0.9559,
+      "step": 5355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975784580535568e-05,
+      "loss": 0.9657,
+      "step": 5356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973095314224787e-05,
+      "loss": 0.9483,
+      "step": 5357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970405899428137e-05,
+      "loss": 0.9494,
+      "step": 5358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1967716336348007e-05,
+      "loss": 0.9528,
+      "step": 5359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965026625186806e-05,
+      "loss": 0.9688,
+      "step": 5360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962336766146945e-05,
+      "loss": 0.9497,
+      "step": 5361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959646759430852e-05,
+      "loss": 0.9604,
+      "step": 5362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195695660524096e-05,
+      "loss": 0.9244,
+      "step": 5363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954266303779725e-05,
+      "loss": 0.9229,
+      "step": 5364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951575855249601e-05,
+      "loss": 0.9853,
+      "step": 5365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948885259853063e-05,
+      "loss": 0.9583,
+      "step": 5366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946194517792584e-05,
+      "loss": 0.9489,
+      "step": 5367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943503629270663e-05,
+      "loss": 1.0272,
+      "step": 5368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940812594489805e-05,
+      "loss": 0.9391,
+      "step": 5369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938121413652522e-05,
+      "loss": 0.9351,
+      "step": 5370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935430086961335e-05,
+      "loss": 0.9502,
+      "step": 5371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1932738614618791e-05,
+      "loss": 0.9554,
+      "step": 5372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193004699682743e-05,
+      "loss": 0.9247,
+      "step": 5373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927355233789812e-05,
+      "loss": 0.9701,
+      "step": 5374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924663325708505e-05,
+      "loss": 0.9225,
+      "step": 5375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921971272786098e-05,
+      "loss": 0.9641,
+      "step": 5376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191927907522517e-05,
+      "loss": 2.9611,
+      "step": 5377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916586733228328e-05,
+      "loss": 0.9489,
+      "step": 5378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913894246998185e-05,
+      "loss": 0.9587,
+      "step": 5379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911201616737369e-05,
+      "loss": 0.979,
+      "step": 5380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908508842648508e-05,
+      "loss": 0.9105,
+      "step": 5381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905815924934252e-05,
+      "loss": 0.9373,
+      "step": 5382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903122863797255e-05,
+      "loss": 0.9342,
+      "step": 5383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900429659440187e-05,
+      "loss": 0.9515,
+      "step": 5384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897736312065721e-05,
+      "loss": 0.9139,
+      "step": 5385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895042821876552e-05,
+      "loss": 0.9762,
+      "step": 5386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892349189075372e-05,
+      "loss": 0.904,
+      "step": 5387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889655413864898e-05,
+      "loss": 0.9811,
+      "step": 5388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886961496447847e-05,
+      "loss": 2.9315,
+      "step": 5389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884267437026953e-05,
+      "loss": 0.9788,
+      "step": 5390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881573235804953e-05,
+      "loss": 0.9904,
+      "step": 5391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878878892984604e-05,
+      "loss": 0.9353,
+      "step": 5392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876184408768672e-05,
+      "loss": 0.9034,
+      "step": 5393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873489783359928e-05,
+      "loss": 0.9535,
+      "step": 5394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870795016961157e-05,
+      "loss": 0.9441,
+      "step": 5395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186810010977515e-05,
+      "loss": 1.0063,
+      "step": 5396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865405062004725e-05,
+      "loss": 0.9698,
+      "step": 5397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862709873852689e-05,
+      "loss": 0.9886,
+      "step": 5398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186001454552187e-05,
+      "loss": 1.0079,
+      "step": 5399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857319077215106e-05,
+      "loss": 1.0069,
+      "step": 5400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1854623469135249e-05,
+      "loss": 0.9448,
+      "step": 5401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851927721485153e-05,
+      "loss": 2.7821,
+      "step": 5402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849231834467694e-05,
+      "loss": 0.9513,
+      "step": 5403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846535808285742e-05,
+      "loss": 0.983,
+      "step": 5404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843839643142196e-05,
+      "loss": 0.9597,
+      "step": 5405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841143339239952e-05,
+      "loss": 0.9247,
+      "step": 5406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838446896781924e-05,
+      "loss": 0.9287,
+      "step": 5407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835750315971029e-05,
+      "loss": 0.9517,
+      "step": 5408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833053597010202e-05,
+      "loss": 1.0031,
+      "step": 5409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830356740102387e-05,
+      "loss": 0.926,
+      "step": 5410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827659745450537e-05,
+      "loss": 0.9915,
+      "step": 5411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824962613257609e-05,
+      "loss": 0.9349,
+      "step": 5412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822265343726583e-05,
+      "loss": 0.986,
+      "step": 5413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819567937060441e-05,
+      "loss": 0.9559,
+      "step": 5414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816870393462176e-05,
+      "loss": 0.9522,
+      "step": 5415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181417271313479e-05,
+      "loss": 0.9166,
+      "step": 5416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811474896281305e-05,
+      "loss": 0.9377,
+      "step": 5417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808776943104743e-05,
+      "loss": 0.9583,
+      "step": 5418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806078853808134e-05,
+      "loss": 0.9618,
+      "step": 5419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803380628594529e-05,
+      "loss": 0.9424,
+      "step": 5420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800682267666983e-05,
+      "loss": 0.9499,
+      "step": 5421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797983771228563e-05,
+      "loss": 0.945,
+      "step": 5422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795285139482341e-05,
+      "loss": 0.9372,
+      "step": 5423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792586372631404e-05,
+      "loss": 0.9619,
+      "step": 5424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789887470878856e-05,
+      "loss": 0.9568,
+      "step": 5425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787188434427795e-05,
+      "loss": 0.9084,
+      "step": 5426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784489263481338e-05,
+      "loss": 2.8563,
+      "step": 5427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781789958242618e-05,
+      "loss": 1.0091,
+      "step": 5428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779090518914767e-05,
+      "loss": 0.9216,
+      "step": 5429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776390945700933e-05,
+      "loss": 1.0143,
+      "step": 5430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773691238804277e-05,
+      "loss": 0.994,
+      "step": 5431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770991398427963e-05,
+      "loss": 0.9164,
+      "step": 5432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768291424775164e-05,
+      "loss": 0.9259,
+      "step": 5433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765591318049074e-05,
+      "loss": 0.9268,
+      "step": 5434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762891078452889e-05,
+      "loss": 0.9543,
+      "step": 5435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760190706189815e-05,
+      "loss": 0.9227,
+      "step": 5436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757490201463066e-05,
+      "loss": 0.9323,
+      "step": 5437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175478956447588e-05,
+      "loss": 0.9541,
+      "step": 5438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752088795431484e-05,
+      "loss": 0.968,
+      "step": 5439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749387894533126e-05,
+      "loss": 0.9398,
+      "step": 5440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746686861984067e-05,
+      "loss": 0.9252,
+      "step": 5441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743985697987575e-05,
+      "loss": 0.9694,
+      "step": 5442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741284402746924e-05,
+      "loss": 0.9659,
+      "step": 5443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738582976465398e-05,
+      "loss": 0.9646,
+      "step": 5444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735881419346299e-05,
+      "loss": 0.9835,
+      "step": 5445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733179731592932e-05,
+      "loss": 0.9403,
+      "step": 5446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730477913408608e-05,
+      "loss": 3.0166,
+      "step": 5447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727775964996664e-05,
+      "loss": 0.948,
+      "step": 5448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725073886560426e-05,
+      "loss": 0.9679,
+      "step": 5449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722371678303244e-05,
+      "loss": 0.9676,
+      "step": 5450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719669340428472e-05,
+      "loss": 0.9629,
+      "step": 5451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716966873139477e-05,
+      "loss": 0.9485,
+      "step": 5452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714264276639633e-05,
+      "loss": 0.9866,
+      "step": 5453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711561551132324e-05,
+      "loss": 0.9872,
+      "step": 5454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708858696820947e-05,
+      "loss": 0.9656,
+      "step": 5455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706155713908902e-05,
+      "loss": 0.9714,
+      "step": 5456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703452602599606e-05,
+      "loss": 0.9608,
+      "step": 5457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170074936309648e-05,
+      "loss": 0.9558,
+      "step": 5458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698045995602962e-05,
+      "loss": 0.9487,
+      "step": 5459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695342500322488e-05,
+      "loss": 0.9594,
+      "step": 5460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692638877458516e-05,
+      "loss": 0.9067,
+      "step": 5461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689935127214501e-05,
+      "loss": 0.9481,
+      "step": 5462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687231249793925e-05,
+      "loss": 0.9725,
+      "step": 5463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684527245400261e-05,
+      "loss": 0.9571,
+      "step": 5464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681823114237001e-05,
+      "loss": 0.9286,
+      "step": 5465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679118856507645e-05,
+      "loss": 3.2012,
+      "step": 5466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676414472415707e-05,
+      "loss": 0.9487,
+      "step": 5467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673709962164699e-05,
+      "loss": 0.9388,
+      "step": 5468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671005325958156e-05,
+      "loss": 0.96,
+      "step": 5469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1668300563999613e-05,
+      "loss": 0.9687,
+      "step": 5470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166559567649262e-05,
+      "loss": 0.9886,
+      "step": 5471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166289066364073e-05,
+      "loss": 0.9464,
+      "step": 5472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660185525647517e-05,
+      "loss": 0.9768,
+      "step": 5473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657480262716545e-05,
+      "loss": 0.983,
+      "step": 5474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165477487505141e-05,
+      "loss": 0.9633,
+      "step": 5475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652069362855704e-05,
+      "loss": 0.9863,
+      "step": 5476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164936372633303e-05,
+      "loss": 0.9834,
+      "step": 5477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646657965687e-05,
+      "loss": 0.9542,
+      "step": 5478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164395208112124e-05,
+      "loss": 0.9793,
+      "step": 5479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641246072839382e-05,
+      "loss": 0.9178,
+      "step": 5480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638539941045067e-05,
+      "loss": 0.952,
+      "step": 5481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163583368594194e-05,
+      "loss": 0.9844,
+      "step": 5482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633127307733669e-05,
+      "loss": 0.9469,
+      "step": 5483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630420806623922e-05,
+      "loss": 0.9615,
+      "step": 5484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1627714182816372e-05,
+      "loss": 0.976,
+      "step": 5485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625007436514713e-05,
+      "loss": 0.9517,
+      "step": 5486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162230056792264e-05,
+      "loss": 0.9953,
+      "step": 5487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619593577243857e-05,
+      "loss": 0.9164,
+      "step": 5488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616886464682082e-05,
+      "loss": 1.0129,
+      "step": 5489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614179230441038e-05,
+      "loss": 0.9758,
+      "step": 5490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611471874724463e-05,
+      "loss": 0.9377,
+      "step": 5491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608764397736092e-05,
+      "loss": 0.93,
+      "step": 5492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606056799679684e-05,
+      "loss": 0.9495,
+      "step": 5493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603349080758995e-05,
+      "loss": 0.9514,
+      "step": 5494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.16006412411778e-05,
+      "loss": 0.9536,
+      "step": 5495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597933281139873e-05,
+      "loss": 0.9305,
+      "step": 5496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595225200849007e-05,
+      "loss": 0.9717,
+      "step": 5497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592517000508998e-05,
+      "loss": 0.9305,
+      "step": 5498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158980868032365e-05,
+      "loss": 0.9427,
+      "step": 5499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158710024049678e-05,
+      "loss": 0.9786,
+      "step": 5500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584391681232217e-05,
+      "loss": 0.9574,
+      "step": 5501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581683002733788e-05,
+      "loss": 0.9099,
+      "step": 5502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578974205205333e-05,
+      "loss": 0.9529,
+      "step": 5503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1576265288850714e-05,
+      "loss": 0.9553,
+      "step": 5504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573556253873784e-05,
+      "loss": 0.9639,
+      "step": 5505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570847100478412e-05,
+      "loss": 0.9541,
+      "step": 5506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568137828868478e-05,
+      "loss": 1.0023,
+      "step": 5507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156542843924787e-05,
+      "loss": 0.9046,
+      "step": 5508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562718931820482e-05,
+      "loss": 0.9275,
+      "step": 5509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560009306790222e-05,
+      "loss": 0.9022,
+      "step": 5510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557299564360996e-05,
+      "loss": 0.9739,
+      "step": 5511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554589704736734e-05,
+      "loss": 2.8521,
+      "step": 5512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551879728121364e-05,
+      "loss": 0.9076,
+      "step": 5513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154916963471883e-05,
+      "loss": 0.9584,
+      "step": 5514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546459424733072e-05,
+      "loss": 0.9803,
+      "step": 5515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543749098368057e-05,
+      "loss": 0.947,
+      "step": 5516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541038655827747e-05,
+      "loss": 0.9695,
+      "step": 5517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538328097316121e-05,
+      "loss": 0.9582,
+      "step": 5518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535617423037155e-05,
+      "loss": 0.9544,
+      "step": 5519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532906633194849e-05,
+      "loss": 0.9403,
+      "step": 5520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530195727993202e-05,
+      "loss": 0.9814,
+      "step": 5521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527484707636222e-05,
+      "loss": 0.9303,
+      "step": 5522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152477357232793e-05,
+      "loss": 0.9602,
+      "step": 5523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1522062322272352e-05,
+      "loss": 0.9399,
+      "step": 5524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1519350957673528e-05,
+      "loss": 0.9503,
+      "step": 5525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516639478735496e-05,
+      "loss": 0.9286,
+      "step": 5526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513927885662312e-05,
+      "loss": 0.9622,
+      "step": 5527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511216178658037e-05,
+      "loss": 0.9571,
+      "step": 5528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508504357926746e-05,
+      "loss": 0.9485,
+      "step": 5529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505792423672515e-05,
+      "loss": 0.9264,
+      "step": 5530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503080376099427e-05,
+      "loss": 0.9663,
+      "step": 5531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500368215411581e-05,
+      "loss": 0.9602,
+      "step": 5532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497655941813083e-05,
+      "loss": 0.907,
+      "step": 5533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494943555508044e-05,
+      "loss": 0.9534,
+      "step": 5534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492231056700592e-05,
+      "loss": 0.9833,
+      "step": 5535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489518445594845e-05,
+      "loss": 0.9852,
+      "step": 5536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148680572239495e-05,
+      "loss": 0.9573,
+      "step": 5537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148409288730505e-05,
+      "loss": 0.9928,
+      "step": 5538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481379940529302e-05,
+      "loss": 0.9361,
+      "step": 5539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478666882271869e-05,
+      "loss": 0.9693,
+      "step": 5540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475953712736922e-05,
+      "loss": 0.9552,
+      "step": 5541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473240432128647e-05,
+      "loss": 0.8939,
+      "step": 5542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470527040651227e-05,
+      "loss": 0.9216,
+      "step": 5543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467813538508858e-05,
+      "loss": 0.9041,
+      "step": 5544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465099925905747e-05,
+      "loss": 0.9339,
+      "step": 5545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462386203046112e-05,
+      "loss": 0.9218,
+      "step": 5546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459672370134172e-05,
+      "loss": 0.9748,
+      "step": 5547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145695842737415e-05,
+      "loss": 0.9413,
+      "step": 5548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454244374970298e-05,
+      "loss": 0.9751,
+      "step": 5549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451530213126857e-05,
+      "loss": 0.89,
+      "step": 5550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448815942048074e-05,
+      "loss": 0.9078,
+      "step": 5551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446101561938225e-05,
+      "loss": 0.9131,
+      "step": 5552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443387073001574e-05,
+      "loss": 2.9511,
+      "step": 5553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440672475442408e-05,
+      "loss": 0.8782,
+      "step": 5554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437957769465002e-05,
+      "loss": 0.9236,
+      "step": 5555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435242955273661e-05,
+      "loss": 2.7577,
+      "step": 5556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432528033072692e-05,
+      "loss": 0.9619,
+      "step": 5557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14298130030664e-05,
+      "loss": 0.9822,
+      "step": 5558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1427097865459108e-05,
+      "loss": 2.7936,
+      "step": 5559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424382620455145e-05,
+      "loss": 0.9751,
+      "step": 5560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421667268258849e-05,
+      "loss": 2.8623,
+      "step": 5561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418951809074561e-05,
+      "loss": 0.9543,
+      "step": 5562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416236243106638e-05,
+      "loss": 0.946,
+      "step": 5563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413520570559438e-05,
+      "loss": 0.9794,
+      "step": 5564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410804791637326e-05,
+      "loss": 0.9838,
+      "step": 5565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408088906544685e-05,
+      "loss": 0.9018,
+      "step": 5566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405372915485899e-05,
+      "loss": 0.9901,
+      "step": 5567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402656818665361e-05,
+      "loss": 0.95,
+      "step": 5568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399940616287464e-05,
+      "loss": 0.9761,
+      "step": 5569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397224308556627e-05,
+      "loss": 0.8843,
+      "step": 5570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394507895677263e-05,
+      "loss": 0.9504,
+      "step": 5571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391791377853792e-05,
+      "loss": 0.9091,
+      "step": 5572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138907475529065e-05,
+      "loss": 0.9668,
+      "step": 5573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386358028192282e-05,
+      "loss": 0.914,
+      "step": 5574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383641196763129e-05,
+      "loss": 0.9379,
+      "step": 5575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380924261207646e-05,
+      "loss": 0.9703,
+      "step": 5576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378207221730301e-05,
+      "loss": 0.9012,
+      "step": 5577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137549007853557e-05,
+      "loss": 0.9571,
+      "step": 5578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137277283182792e-05,
+      "loss": 0.9749,
+      "step": 5579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1370055481811851e-05,
+      "loss": 0.9112,
+      "step": 5580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367338028691848e-05,
+      "loss": 0.9566,
+      "step": 5581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136462047267242e-05,
+      "loss": 0.9878,
+      "step": 5582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361902813958075e-05,
+      "loss": 0.9273,
+      "step": 5583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359185052753332e-05,
+      "loss": 0.967,
+      "step": 5584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356467189262714e-05,
+      "loss": 0.9207,
+      "step": 5585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135374922369076e-05,
+      "loss": 0.9146,
+      "step": 5586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351031156242007e-05,
+      "loss": 0.9313,
+      "step": 5587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348312987121008e-05,
+      "loss": 0.9289,
+      "step": 5588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345594716532313e-05,
+      "loss": 1.0041,
+      "step": 5589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342876344680494e-05,
+      "loss": 0.894,
+      "step": 5590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340157871770117e-05,
+      "loss": 0.9165,
+      "step": 5591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337439298005765e-05,
+      "loss": 0.9749,
+      "step": 5592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334720623592024e-05,
+      "loss": 0.9365,
+      "step": 5593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332001848733485e-05,
+      "loss": 0.9355,
+      "step": 5594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329282973634758e-05,
+      "loss": 0.9378,
+      "step": 5595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326563998500448e-05,
+      "loss": 0.9709,
+      "step": 5596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323844923535168e-05,
+      "loss": 0.9172,
+      "step": 5597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321125748943551e-05,
+      "loss": 0.9573,
+      "step": 5598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318406474930226e-05,
+      "loss": 0.9681,
+      "step": 5599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315687101699831e-05,
+      "loss": 0.9582,
+      "step": 5600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312967629457017e-05,
+      "loss": 0.9829,
+      "step": 5601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310248058406432e-05,
+      "loss": 0.9285,
+      "step": 5602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1307528388752749e-05,
+      "loss": 2.7942,
+      "step": 5603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304808620700625e-05,
+      "loss": 1.0084,
+      "step": 5604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302088754454745e-05,
+      "loss": 0.9667,
+      "step": 5605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1299368790219791e-05,
+      "loss": 0.9437,
+      "step": 5606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296648728200456e-05,
+      "loss": 0.9417,
+      "step": 5607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129392856860144e-05,
+      "loss": 0.9,
+      "step": 5608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1291208311627446e-05,
+      "loss": 0.9058,
+      "step": 5609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1288487957483187e-05,
+      "loss": 0.9149,
+      "step": 5610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285767506373387e-05,
+      "loss": 0.9256,
+      "step": 5611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283046958502777e-05,
+      "loss": 0.9547,
+      "step": 5612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280326314076087e-05,
+      "loss": 0.9543,
+      "step": 5613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1277605573298059e-05,
+      "loss": 0.9496,
+      "step": 5614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127488473637345e-05,
+      "loss": 0.9391,
+      "step": 5615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272163803507015e-05,
+      "loss": 0.9824,
+      "step": 5616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269442774903517e-05,
+      "loss": 0.9736,
+      "step": 5617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266721650767725e-05,
+      "loss": 0.9433,
+      "step": 5618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264000431304422e-05,
+      "loss": 0.9795,
+      "step": 5619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261279116718398e-05,
+      "loss": 0.9671,
+      "step": 5620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1258557707214437e-05,
+      "loss": 0.9541,
+      "step": 5621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255836202997343e-05,
+      "loss": 0.9483,
+      "step": 5622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1253114604271928e-05,
+      "loss": 0.9238,
+      "step": 5623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250392911243005e-05,
+      "loss": 0.9504,
+      "step": 5624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247671124115389e-05,
+      "loss": 0.9468,
+      "step": 5625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244949243093916e-05,
+      "loss": 0.9737,
+      "step": 5626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242227268383424e-05,
+      "loss": 2.956,
+      "step": 5627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239505200188748e-05,
+      "loss": 0.9855,
+      "step": 5628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236783038714743e-05,
+      "loss": 0.9474,
+      "step": 5629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234060784166266e-05,
+      "loss": 0.8968,
+      "step": 5630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123133843674818e-05,
+      "loss": 1.0108,
+      "step": 5631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228615996665355e-05,
+      "loss": 0.9398,
+      "step": 5632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225893464122672e-05,
+      "loss": 0.944,
+      "step": 5633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223170839325018e-05,
+      "loss": 0.9466,
+      "step": 5634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220448122477273e-05,
+      "loss": 0.9368,
+      "step": 5635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217725313784349e-05,
+      "loss": 0.9332,
+      "step": 5636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215002413451148e-05,
+      "loss": 2.9989,
+      "step": 5637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121227942168258e-05,
+      "loss": 0.9121,
+      "step": 5638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209556338683564e-05,
+      "loss": 0.9282,
+      "step": 5639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206833164659035e-05,
+      "loss": 0.9796,
+      "step": 5640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204109899813916e-05,
+      "loss": 2.7669,
+      "step": 5641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120138654435315e-05,
+      "loss": 0.9619,
+      "step": 5642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198663098481685e-05,
+      "loss": 0.9312,
+      "step": 5643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195939562404477e-05,
+      "loss": 0.9483,
+      "step": 5644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193215936326482e-05,
+      "loss": 0.9112,
+      "step": 5645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1190492220452671e-05,
+      "loss": 0.9443,
+      "step": 5646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187768414988015e-05,
+      "loss": 0.9517,
+      "step": 5647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185044520137497e-05,
+      "loss": 0.9173,
+      "step": 5648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182320536106101e-05,
+      "loss": 0.9188,
+      "step": 5649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117959646309883e-05,
+      "loss": 0.9868,
+      "step": 5650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176872301320673e-05,
+      "loss": 0.9844,
+      "step": 5651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174148050976648e-05,
+      "loss": 0.9511,
+      "step": 5652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171423712271761e-05,
+      "loss": 0.9611,
+      "step": 5653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116869928541104e-05,
+      "loss": 0.962,
+      "step": 5654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165974770599506e-05,
+      "loss": 0.9409,
+      "step": 5655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163250168042197e-05,
+      "loss": 0.9449,
+      "step": 5656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160525477944156e-05,
+      "loss": 0.8935,
+      "step": 5657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157800700510427e-05,
+      "loss": 0.98,
+      "step": 5658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155075835946061e-05,
+      "loss": 0.9398,
+      "step": 5659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152350884456122e-05,
+      "loss": 0.9969,
+      "step": 5660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149625846245682e-05,
+      "loss": 0.9535,
+      "step": 5661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146900721519807e-05,
+      "loss": 0.9383,
+      "step": 5662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1144175510483575e-05,
+      "loss": 0.9634,
+      "step": 5663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141450213342082e-05,
+      "loss": 0.9212,
+      "step": 5664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138724830300417e-05,
+      "loss": 0.9648,
+      "step": 5665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135999361563676e-05,
+      "loss": 0.9421,
+      "step": 5666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133273807336968e-05,
+      "loss": 0.9151,
+      "step": 5667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130548167825409e-05,
+      "loss": 0.9421,
+      "step": 5668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127822443234109e-05,
+      "loss": 2.7969,
+      "step": 5669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11250966337682e-05,
+      "loss": 0.9736,
+      "step": 5670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112237073963281e-05,
+      "loss": 0.9658,
+      "step": 5671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 0.9495,
+      "step": 5672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111691869817415e-05,
+      "loss": 0.95,
+      "step": 5673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114192551261178e-05,
+      "loss": 0.9463,
+      "step": 5674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111466320499318e-05,
+      "loss": 1.0052,
+      "step": 5675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108740006093728e-05,
+      "loss": 0.9521,
+      "step": 5676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110601360824958e-05,
+      "loss": 0.8787,
+      "step": 5677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103287127172055e-05,
+      "loss": 0.9745,
+      "step": 5678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100560563066331e-05,
+      "loss": 0.9637,
+      "step": 5679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097833916137593e-05,
+      "loss": 0.9378,
+      "step": 5680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095107186591044e-05,
+      "loss": 0.9817,
+      "step": 5681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092380374631879e-05,
+      "loss": 0.9838,
+      "step": 5682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089653480465305e-05,
+      "loss": 2.894,
+      "step": 5683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086926504296539e-05,
+      "loss": 0.9548,
+      "step": 5684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084199446330796e-05,
+      "loss": 0.9659,
+      "step": 5685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081472306773306e-05,
+      "loss": 0.9427,
+      "step": 5686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078745085829297e-05,
+      "loss": 0.9812,
+      "step": 5687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076017783704006e-05,
+      "loss": 0.9278,
+      "step": 5688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073290400602681e-05,
+      "loss": 0.9558,
+      "step": 5689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070562936730572e-05,
+      "loss": 0.9562,
+      "step": 5690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106783539229293e-05,
+      "loss": 0.9483,
+      "step": 5691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106510776749502e-05,
+      "loss": 0.8784,
+      "step": 5692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062380062542112e-05,
+      "loss": 0.939,
+      "step": 5693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059652277639479e-05,
+      "loss": 1.0122,
+      "step": 5694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10569244129924e-05,
+      "loss": 0.9311,
+      "step": 5695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054196468806162e-05,
+      "loss": 0.9774,
+      "step": 5696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105146844528606e-05,
+      "loss": 0.9619,
+      "step": 5697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048740342637384e-05,
+      "loss": 1.0127,
+      "step": 5698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046012161065451e-05,
+      "loss": 0.9583,
+      "step": 5699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043283900775558e-05,
+      "loss": 0.9494,
+      "step": 5700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104055556197303e-05,
+      "loss": 0.9499,
+      "step": 5701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037827144863187e-05,
+      "loss": 0.9567,
+      "step": 5702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035098649651355e-05,
+      "loss": 0.968,
+      "step": 5703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032370076542867e-05,
+      "loss": 0.93,
+      "step": 5704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029641425743063e-05,
+      "loss": 0.9469,
+      "step": 5705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026912697457292e-05,
+      "loss": 0.968,
+      "step": 5706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024183891890905e-05,
+      "loss": 0.9827,
+      "step": 5707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021455009249254e-05,
+      "loss": 0.9377,
+      "step": 5708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018726049737703e-05,
+      "loss": 0.9249,
+      "step": 5709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015997013561625e-05,
+      "loss": 0.9449,
+      "step": 5710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013267900926389e-05,
+      "loss": 0.8957,
+      "step": 5711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010538712037383e-05,
+      "loss": 0.9126,
+      "step": 5712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007809447099981e-05,
+      "loss": 0.9647,
+      "step": 5713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005080106319588e-05,
+      "loss": 0.9547,
+      "step": 5714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002350689901591e-05,
+      "loss": 0.9193,
+      "step": 5715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999621198051399e-05,
+      "loss": 0.9438,
+      "step": 5716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0996891630974416e-05,
+      "loss": 0.9423,
+      "step": 5717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994161988876062e-05,
+      "loss": 0.9587,
+      "step": 5718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991432271961751e-05,
+      "loss": 0.9284,
+      "step": 5719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988702480436914e-05,
+      "loss": 0.9198,
+      "step": 5720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985972614506978e-05,
+      "loss": 0.9656,
+      "step": 5721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983242674377382e-05,
+      "loss": 0.9491,
+      "step": 5722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098051266025357e-05,
+      "loss": 0.9861,
+      "step": 5723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977782572340988e-05,
+      "loss": 0.8623,
+      "step": 5724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975052410845089e-05,
+      "loss": 0.9555,
+      "step": 5725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972322175971332e-05,
+      "loss": 0.9589,
+      "step": 5726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969591867925184e-05,
+      "loss": 0.9449,
+      "step": 5727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966861486912114e-05,
+      "loss": 1.0049,
+      "step": 5728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964131033137595e-05,
+      "loss": 0.9796,
+      "step": 5729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961400506807114e-05,
+      "loss": 0.9626,
+      "step": 5730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958669908126151e-05,
+      "loss": 0.9152,
+      "step": 5731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955939237300202e-05,
+      "loss": 0.955,
+      "step": 5732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953208494534764e-05,
+      "loss": 0.93,
+      "step": 5733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950477680035338e-05,
+      "loss": 0.9318,
+      "step": 5734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947746794007437e-05,
+      "loss": 0.9836,
+      "step": 5735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945015836656567e-05,
+      "loss": 1.0114,
+      "step": 5736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942284808188255e-05,
+      "loss": 0.9305,
+      "step": 5737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0939553708808017e-05,
+      "loss": 0.9321,
+      "step": 5738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093682253872139e-05,
+      "loss": 0.9327,
+      "step": 5739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934091298133906e-05,
+      "loss": 0.9252,
+      "step": 5740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931359987251106e-05,
+      "loss": 0.95,
+      "step": 5741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928628606278535e-05,
+      "loss": 0.9391,
+      "step": 5742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925897155421743e-05,
+      "loss": 0.974,
+      "step": 5743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923165634886288e-05,
+      "loss": 0.9693,
+      "step": 5744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920434044877733e-05,
+      "loss": 0.9295,
+      "step": 5745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917702385601637e-05,
+      "loss": 0.905,
+      "step": 5746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091497065726358e-05,
+      "loss": 0.9395,
+      "step": 5747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912238860069138e-05,
+      "loss": 1.0123,
+      "step": 5748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909506994223886e-05,
+      "loss": 0.9727,
+      "step": 5749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0906775059933419e-05,
+      "loss": 0.9501,
+      "step": 5750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0904043057403328e-05,
+      "loss": 0.9174,
+      "step": 5751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090131098683921e-05,
+      "loss": 0.9003,
+      "step": 5752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898578848446668e-05,
+      "loss": 0.962,
+      "step": 5753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895846642431306e-05,
+      "loss": 0.9829,
+      "step": 5754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893114368998745e-05,
+      "loss": 0.9469,
+      "step": 5755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890382028354598e-05,
+      "loss": 0.9182,
+      "step": 5756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887649620704486e-05,
+      "loss": 0.9569,
+      "step": 5757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088491714625404e-05,
+      "loss": 0.9636,
+      "step": 5758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882184605208895e-05,
+      "loss": 0.9429,
+      "step": 5759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879451997774684e-05,
+      "loss": 0.9508,
+      "step": 5760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0876719324157057e-05,
+      "loss": 0.9543,
+      "step": 5761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873986584561655e-05,
+      "loss": 0.9269,
+      "step": 5762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871253779194138e-05,
+      "loss": 0.9383,
+      "step": 5763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868520908260155e-05,
+      "loss": 0.9597,
+      "step": 5764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086578797196538e-05,
+      "loss": 0.9744,
+      "step": 5765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0863054970515474e-05,
+      "loss": 0.9322,
+      "step": 5766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086032190411611e-05,
+      "loss": 0.9553,
+      "step": 5767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857588772972968e-05,
+      "loss": 0.9311,
+      "step": 5768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854855577291733e-05,
+      "loss": 0.9065,
+      "step": 5769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852122317278084e-05,
+      "loss": 0.9623,
+      "step": 5770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084938899313772e-05,
+      "loss": 0.9232,
+      "step": 5771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846655605076337e-05,
+      "loss": 0.9154,
+      "step": 5772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843922153299638e-05,
+      "loss": 1.0094,
+      "step": 5773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841188638013325e-05,
+      "loss": 0.9507,
+      "step": 5774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838455059423113e-05,
+      "loss": 0.9654,
+      "step": 5775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835721417734721e-05,
+      "loss": 0.9378,
+      "step": 5776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832987713153864e-05,
+      "loss": 0.9516,
+      "step": 5777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830253945886273e-05,
+      "loss": 0.9474,
+      "step": 5778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827520116137672e-05,
+      "loss": 0.902,
+      "step": 5779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824786224113805e-05,
+      "loss": 0.9095,
+      "step": 5780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822052270020407e-05,
+      "loss": 0.9429,
+      "step": 5781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081931825406322e-05,
+      "loss": 0.9429,
+      "step": 5782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816584176447997e-05,
+      "loss": 0.9599,
+      "step": 5783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813850037380494e-05,
+      "loss": 0.9398,
+      "step": 5784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081111583706646e-05,
+      "loss": 0.933,
+      "step": 5785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808381575711668e-05,
+      "loss": 0.9414,
+      "step": 5786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080564725352188e-05,
+      "loss": 1.025,
+      "step": 5787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080291287070287e-05,
+      "loss": 0.9206,
+      "step": 5788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800178427460417e-05,
+      "loss": 0.9605,
+      "step": 5789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0797443924000302e-05,
+      "loss": 0.9703,
+      "step": 5790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794709360528303e-05,
+      "loss": 0.9208,
+      "step": 5791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079197473725022e-05,
+      "loss": 0.9153,
+      "step": 5792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789240054371846e-05,
+      "loss": 0.9789,
+      "step": 5793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786505312098976e-05,
+      "loss": 0.8934,
+      "step": 5794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783770510637415e-05,
+      "loss": 0.9186,
+      "step": 5795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781035650192976e-05,
+      "loss": 0.9477,
+      "step": 5796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077830073097147e-05,
+      "loss": 0.9671,
+      "step": 5797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775565753178707e-05,
+      "loss": 0.9348,
+      "step": 5798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772830717020517e-05,
+      "loss": 0.9755,
+      "step": 5799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770095622702726e-05,
+      "loss": 0.9807,
+      "step": 5800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767360470431159e-05,
+      "loss": 0.9186,
+      "step": 5801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764625260411652e-05,
+      "loss": 0.9298,
+      "step": 5802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761889992850048e-05,
+      "loss": 0.9964,
+      "step": 5803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759154667952186e-05,
+      "loss": 0.9434,
+      "step": 5804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756419285923916e-05,
+      "loss": 0.9288,
+      "step": 5805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753683846971089e-05,
+      "loss": 0.9966,
+      "step": 5806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750948351299561e-05,
+      "loss": 0.9704,
+      "step": 5807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748212799115193e-05,
+      "loss": 2.8013,
+      "step": 5808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074547719062385e-05,
+      "loss": 0.9052,
+      "step": 5809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742741526031404e-05,
+      "loss": 0.9816,
+      "step": 5810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740005805543723e-05,
+      "loss": 0.9941,
+      "step": 5811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737270029366682e-05,
+      "loss": 0.9647,
+      "step": 5812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734534197706175e-05,
+      "loss": 0.9654,
+      "step": 5813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731798310768076e-05,
+      "loss": 0.9438,
+      "step": 5814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729062368758278e-05,
+      "loss": 0.8585,
+      "step": 5815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726326371882679e-05,
+      "loss": 0.9597,
+      "step": 5816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723590320347172e-05,
+      "loss": 0.9372,
+      "step": 5817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720854214357666e-05,
+      "loss": 0.9592,
+      "step": 5818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071811805412006e-05,
+      "loss": 0.9171,
+      "step": 5819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715381839840266e-05,
+      "loss": 0.9856,
+      "step": 5820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712645571724205e-05,
+      "loss": 0.9826,
+      "step": 5821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709909249977788e-05,
+      "loss": 0.917,
+      "step": 5822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707172874806942e-05,
+      "loss": 0.9731,
+      "step": 5823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070443644641759e-05,
+      "loss": 0.9071,
+      "step": 5824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701699965015669e-05,
+      "loss": 0.9676,
+      "step": 5825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698963430807106e-05,
+      "loss": 2.683,
+      "step": 5826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696226843997847e-05,
+      "loss": 0.9597,
+      "step": 5827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069349020479383e-05,
+      "loss": 0.9286,
+      "step": 5828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0690753513401004e-05,
+      "loss": 1.0009,
+      "step": 5829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688016770025315e-05,
+      "loss": 0.9655,
+      "step": 5830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685279974872726e-05,
+      "loss": 0.9373,
+      "step": 5831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682543128149186e-05,
+      "loss": 1.011,
+      "step": 5832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679806230060661e-05,
+      "loss": 0.9656,
+      "step": 5833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677069280813121e-05,
+      "loss": 0.9583,
+      "step": 5834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674332280612532e-05,
+      "loss": 0.9429,
+      "step": 5835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671595229664867e-05,
+      "loss": 0.9254,
+      "step": 5836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668858128176103e-05,
+      "loss": 2.8192,
+      "step": 5837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666120976352223e-05,
+      "loss": 0.9463,
+      "step": 5838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663383774399217e-05,
+      "loss": 0.9364,
+      "step": 5839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0660646522523063e-05,
+      "loss": 0.8975,
+      "step": 5840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0657909220929764e-05,
+      "loss": 0.9691,
+      "step": 5841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655171869825311e-05,
+      "loss": 0.9009,
+      "step": 5842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652434469415705e-05,
+      "loss": 2.8896,
+      "step": 5843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649697019906952e-05,
+      "loss": 0.9397,
+      "step": 5844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646959521505056e-05,
+      "loss": 0.9291,
+      "step": 5845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644221974416031e-05,
+      "loss": 0.8582,
+      "step": 5846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641484378845892e-05,
+      "loss": 0.9543,
+      "step": 5847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638746735000656e-05,
+      "loss": 0.9251,
+      "step": 5848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636009043086346e-05,
+      "loss": 0.9822,
+      "step": 5849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633271303308988e-05,
+      "loss": 0.9255,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063053351587461e-05,
+      "loss": 0.926,
+      "step": 5851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627795680989248e-05,
+      "loss": 0.9633,
+      "step": 5852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625057798858933e-05,
+      "loss": 0.9593,
+      "step": 5853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062231986968971e-05,
+      "loss": 0.963,
+      "step": 5854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619581893687624e-05,
+      "loss": 0.9542,
+      "step": 5855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616843871058719e-05,
+      "loss": 0.9845,
+      "step": 5856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614105802009044e-05,
+      "loss": 0.9231,
+      "step": 5857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611367686744658e-05,
+      "loss": 0.9607,
+      "step": 5858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608629525471617e-05,
+      "loss": 0.9293,
+      "step": 5859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060589131839598e-05,
+      "loss": 0.9645,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060315306572381e-05,
+      "loss": 0.9717,
+      "step": 5861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0600414767661182e-05,
+      "loss": 0.959,
+      "step": 5862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597676424414162e-05,
+      "loss": 0.921,
+      "step": 5863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594938036188827e-05,
+      "loss": 0.9076,
+      "step": 5864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592199603191251e-05,
+      "loss": 0.9742,
+      "step": 5865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589461125627523e-05,
+      "loss": 0.8795,
+      "step": 5866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586722603703726e-05,
+      "loss": 0.9889,
+      "step": 5867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058398403762594e-05,
+      "loss": 0.9587,
+      "step": 5868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581245427600264e-05,
+      "loss": 0.9375,
+      "step": 5869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0578506773832796e-05,
+      "loss": 0.9261,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575768076529627e-05,
+      "loss": 0.9462,
+      "step": 5871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573029335896863e-05,
+      "loss": 0.942,
+      "step": 5872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570290552140602e-05,
+      "loss": 0.9153,
+      "step": 5873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567551725466965e-05,
+      "loss": 0.9704,
+      "step": 5874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056481285608205e-05,
+      "loss": 0.9637,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056207394419198e-05,
+      "loss": 0.9383,
+      "step": 5876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559334990002868e-05,
+      "loss": 0.9763,
+      "step": 5877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055659599372083e-05,
+      "loss": 0.9285,
+      "step": 5878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553856955552005e-05,
+      "loss": 0.9626,
+      "step": 5879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551117875702508e-05,
+      "loss": 0.9311,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548378754378472e-05,
+      "loss": 0.9227,
+      "step": 5881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0545639591786033e-05,
+      "loss": 0.9829,
+      "step": 5882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542900388131324e-05,
+      "loss": 0.9806,
+      "step": 5883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540161143620487e-05,
+      "loss": 0.8905,
+      "step": 5884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537421858459661e-05,
+      "loss": 0.9375,
+      "step": 5885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534682532854997e-05,
+      "loss": 0.9121,
+      "step": 5886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053194316701264e-05,
+      "loss": 1.0272,
+      "step": 5887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529203761138743e-05,
+      "loss": 0.9293,
+      "step": 5888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052646431543946e-05,
+      "loss": 0.9408,
+      "step": 5889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052372483012095e-05,
+      "loss": 0.9664,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520985305389374e-05,
+      "loss": 0.91,
+      "step": 5891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518245741450894e-05,
+      "loss": 0.9675,
+      "step": 5892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051550613851168e-05,
+      "loss": 0.941,
+      "step": 5893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512766496777896e-05,
+      "loss": 0.9392,
+      "step": 5894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051002681645572e-05,
+      "loss": 0.9676,
+      "step": 5895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507287097751324e-05,
+      "loss": 0.9103,
+      "step": 5896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504547340870889e-05,
+      "loss": 0.9976,
+      "step": 5897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501807546020595e-05,
+      "loss": 0.8934,
+      "step": 5898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499067713406622e-05,
+      "loss": 0.945,
+      "step": 5899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496327843235169e-05,
+      "loss": 0.9283,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493587935712413e-05,
+      "loss": 0.9194,
+      "step": 5901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049084799104455e-05,
+      "loss": 0.9548,
+      "step": 5902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488108009437776e-05,
+      "loss": 0.9325,
+      "step": 5903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485367991098292e-05,
+      "loss": 0.9722,
+      "step": 5904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482627936232297e-05,
+      "loss": 0.9391,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479887845045991e-05,
+      "loss": 0.9206,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477147717745586e-05,
+      "loss": 0.953,
+      "step": 5907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474407554537288e-05,
+      "loss": 0.9208,
+      "step": 5908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471667355627311e-05,
+      "loss": 0.9759,
+      "step": 5909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468927121221866e-05,
+      "loss": 0.9696,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0466186851527172e-05,
+      "loss": 0.9675,
+      "step": 5911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046344654674945e-05,
+      "loss": 0.9711,
+      "step": 5912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046070620709492e-05,
+      "loss": 0.9004,
+      "step": 5913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045796583276981e-05,
+      "loss": 0.9704,
+      "step": 5914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455225423980344e-05,
+      "loss": 0.9602,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452484980932756e-05,
+      "loss": 0.9272,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449744503833274e-05,
+      "loss": 0.8966,
+      "step": 5917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447003992888141e-05,
+      "loss": 0.9513,
+      "step": 5918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444263448303588e-05,
+      "loss": 0.9412,
+      "step": 5919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0441522870285858e-05,
+      "loss": 0.9333,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438782259041196e-05,
+      "loss": 0.9473,
+      "step": 5921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436041614775844e-05,
+      "loss": 0.9548,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433300937696051e-05,
+      "loss": 0.9367,
+      "step": 5923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430560228008068e-05,
+      "loss": 2.8098,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427819485918149e-05,
+      "loss": 0.993,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042507871163255e-05,
+      "loss": 0.9657,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422337905357523e-05,
+      "loss": 0.9814,
+      "step": 5927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0419597067299334e-05,
+      "loss": 0.9544,
+      "step": 5928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0416856197664245e-05,
+      "loss": 0.9336,
+      "step": 5929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414115296658518e-05,
+      "loss": 0.965,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411374364488422e-05,
+      "loss": 0.9461,
+      "step": 5931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408633401360229e-05,
+      "loss": 0.9239,
+      "step": 5932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405892407480207e-05,
+      "loss": 0.9553,
+      "step": 5933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403151383054633e-05,
+      "loss": 1.0069,
+      "step": 5934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040041032828978e-05,
+      "loss": 0.9796,
+      "step": 5935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397669243391933e-05,
+      "loss": 0.9739,
+      "step": 5936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394928128567368e-05,
+      "loss": 0.9253,
+      "step": 5937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039218698402237e-05,
+      "loss": 0.9452,
+      "step": 5938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389445809963223e-05,
+      "loss": 0.9363,
+      "step": 5939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038670460659622e-05,
+      "loss": 0.9206,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383963374127646e-05,
+      "loss": 0.9591,
+      "step": 5941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381222112763795e-05,
+      "loss": 0.9434,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378480822710961e-05,
+      "loss": 0.958,
+      "step": 5943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375739504175441e-05,
+      "loss": 0.9237,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372998157363536e-05,
+      "loss": 0.9212,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370256782481543e-05,
+      "loss": 0.8729,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367515379735766e-05,
+      "loss": 0.9063,
+      "step": 5947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036477394933251e-05,
+      "loss": 0.9547,
+      "step": 5948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362032491478085e-05,
+      "loss": 0.8899,
+      "step": 5949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359291006378795e-05,
+      "loss": 0.9674,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356549494240956e-05,
+      "loss": 0.9117,
+      "step": 5951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353807955270878e-05,
+      "loss": 0.9536,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035106638967488e-05,
+      "loss": 0.9132,
+      "step": 5953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348324797659278e-05,
+      "loss": 0.9818,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345583179430387e-05,
+      "loss": 0.9178,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342841535194533e-05,
+      "loss": 0.9269,
+      "step": 5956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034009986515804e-05,
+      "loss": 0.956,
+      "step": 5957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033735816952723e-05,
+      "loss": 0.97,
+      "step": 5958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334616448508433e-05,
+      "loss": 0.9133,
+      "step": 5959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331874702307972e-05,
+      "loss": 2.8383,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329132931132187e-05,
+      "loss": 0.9434,
+      "step": 5961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03263911351874e-05,
+      "loss": 0.9385,
+      "step": 5962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323649314679958e-05,
+      "loss": 0.9094,
+      "step": 5963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320907469816188e-05,
+      "loss": 0.9253,
+      "step": 5964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318165600802433e-05,
+      "loss": 0.9186,
+      "step": 5965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315423707845031e-05,
+      "loss": 0.9137,
+      "step": 5966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312681791150328e-05,
+      "loss": 0.9729,
+      "step": 5967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030993985092466e-05,
+      "loss": 0.9,
+      "step": 5968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307197887374376e-05,
+      "loss": 0.9623,
+      "step": 5969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030445590070583e-05,
+      "loss": 0.9458,
+      "step": 5970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301713891125363e-05,
+      "loss": 2.8409,
+      "step": 5971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298971858839324e-05,
+      "loss": 0.9504,
+      "step": 5972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296229804054074e-05,
+      "loss": 0.9335,
+      "step": 5973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0293487726975961e-05,
+      "loss": 0.9691,
+      "step": 5974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0290745627811343e-05,
+      "loss": 0.9568,
+      "step": 5975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288003506766575e-05,
+      "loss": 2.8241,
+      "step": 5976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028526136404802e-05,
+      "loss": 0.9739,
+      "step": 5977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028251919986204e-05,
+      "loss": 0.9604,
+      "step": 5978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027977701441499e-05,
+      "loss": 0.9631,
+      "step": 5979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027703480791324e-05,
+      "loss": 0.9432,
+      "step": 5980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027429258056315e-05,
+      "loss": 0.947,
+      "step": 5981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271550332571097e-05,
+      "loss": 0.8641,
+      "step": 5982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268808064143438e-05,
+      "loss": 0.9304,
+      "step": 5983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266065775486552e-05,
+      "loss": 0.9346,
+      "step": 5984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263323466806805e-05,
+      "loss": 0.8995,
+      "step": 5985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260581138310573e-05,
+      "loss": 0.9741,
+      "step": 5986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257838790204234e-05,
+      "loss": 0.9312,
+      "step": 5987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0255096422694159e-05,
+      "loss": 0.9059,
+      "step": 5988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252354035986726e-05,
+      "loss": 0.932,
+      "step": 5989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249611630288315e-05,
+      "loss": 0.9459,
+      "step": 5990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024686920580531e-05,
+      "loss": 0.935,
+      "step": 5991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244126762744089e-05,
+      "loss": 0.9304,
+      "step": 5992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241384301311032e-05,
+      "loss": 0.978,
+      "step": 5993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238641821712533e-05,
+      "loss": 0.9251,
+      "step": 5994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235899324154974e-05,
+      "loss": 0.9508,
+      "step": 5995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023315680884474e-05,
+      "loss": 0.9396,
+      "step": 5996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230414275988218e-05,
+      "loss": 0.9331,
+      "step": 5997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0227671725791808e-05,
+      "loss": 0.9222,
+      "step": 5998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224929158461894e-05,
+      "loss": 0.8753,
+      "step": 5999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222186574204867e-05,
+      "loss": 0.9833,
+      "step": 6000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219443973227123e-05,
+      "loss": 0.9623,
+      "step": 6001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216701355735058e-05,
+      "loss": 0.9453,
+      "step": 6002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021395872193507e-05,
+      "loss": 0.9057,
+      "step": 6003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211216072033559e-05,
+      "loss": 0.898,
+      "step": 6004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208473406236913e-05,
+      "loss": 0.9312,
+      "step": 6005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205730724751546e-05,
+      "loss": 2.8861,
+      "step": 6006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202988027783847e-05,
+      "loss": 0.953,
+      "step": 6007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200245315540228e-05,
+      "loss": 0.9337,
+      "step": 6008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197502588227086e-05,
+      "loss": 0.9876,
+      "step": 6009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194759846050828e-05,
+      "loss": 0.9064,
+      "step": 6010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192017089217863e-05,
+      "loss": 0.9924,
+      "step": 6011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189274317934594e-05,
+      "loss": 0.9057,
+      "step": 6012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018653153240743e-05,
+      "loss": 0.9558,
+      "step": 6013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0183788732842782e-05,
+      "loss": 0.9378,
+      "step": 6014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181045919447057e-05,
+      "loss": 0.9583,
+      "step": 6015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178303092426669e-05,
+      "loss": 0.9363,
+      "step": 6016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175560251988028e-05,
+      "loss": 0.945,
+      "step": 6017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172817398337547e-05,
+      "loss": 0.9101,
+      "step": 6018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170074531681644e-05,
+      "loss": 0.9278,
+      "step": 6019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0167331652226733e-05,
+      "loss": 0.9528,
+      "step": 6020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164588760179225e-05,
+      "loss": 0.9309,
+      "step": 6021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0161845855745545e-05,
+      "loss": 0.949,
+      "step": 6022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159102939132106e-05,
+      "loss": 0.9699,
+      "step": 6023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156360010545327e-05,
+      "loss": 0.9622,
+      "step": 6024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153617070191632e-05,
+      "loss": 0.9581,
+      "step": 6025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150874118277436e-05,
+      "loss": 0.9568,
+      "step": 6026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148131155009167e-05,
+      "loss": 0.9008,
+      "step": 6027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145388180593243e-05,
+      "loss": 2.9519,
+      "step": 6028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014264519523609e-05,
+      "loss": 2.7738,
+      "step": 6029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139902199144127e-05,
+      "loss": 0.9026,
+      "step": 6030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137159192523782e-05,
+      "loss": 0.9525,
+      "step": 6031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134416175581485e-05,
+      "loss": 0.9307,
+      "step": 6032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131673148523659e-05,
+      "loss": 0.9139,
+      "step": 6033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012893011155673e-05,
+      "loss": 0.92,
+      "step": 6034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0126187064887127e-05,
+      "loss": 0.9819,
+      "step": 6035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123444008721282e-05,
+      "loss": 0.9875,
+      "step": 6036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012070094326562e-05,
+      "loss": 0.9196,
+      "step": 6037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117957868726568e-05,
+      "loss": 0.9173,
+      "step": 6038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115214785310567e-05,
+      "loss": 0.9253,
+      "step": 6039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112471693224043e-05,
+      "loss": 0.9507,
+      "step": 6040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010972859267343e-05,
+      "loss": 0.9362,
+      "step": 6041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106985483865154e-05,
+      "loss": 0.9456,
+      "step": 6042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010424236700566e-05,
+      "loss": 0.9596,
+      "step": 6043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0101499242301373e-05,
+      "loss": 0.9457,
+      "step": 6044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009875610995873e-05,
+      "loss": 0.9372,
+      "step": 6045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096012970184166e-05,
+      "loss": 0.9559,
+      "step": 6046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093269823184122e-05,
+      "loss": 0.93,
+      "step": 6047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090526669165026e-05,
+      "loss": 0.9553,
+      "step": 6048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008778350833332e-05,
+      "loss": 3.1667,
+      "step": 6049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008504034089544e-05,
+      "loss": 3.0035,
+      "step": 6050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082297167057825e-05,
+      "loss": 0.9656,
+      "step": 6051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007955398702691e-05,
+      "loss": 0.9562,
+      "step": 6052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076810801009138e-05,
+      "loss": 0.9879,
+      "step": 6053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074067609210946e-05,
+      "loss": 0.906,
+      "step": 6054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071324411838775e-05,
+      "loss": 0.933,
+      "step": 6055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068581209099062e-05,
+      "loss": 0.942,
+      "step": 6056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065838001198252e-05,
+      "loss": 0.9397,
+      "step": 6057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063094788342784e-05,
+      "loss": 0.955,
+      "step": 6058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060351570739096e-05,
+      "loss": 0.937,
+      "step": 6059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057608348593635e-05,
+      "loss": 0.9247,
+      "step": 6060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054865122112842e-05,
+      "loss": 0.9315,
+      "step": 6061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052121891503157e-05,
+      "loss": 0.9674,
+      "step": 6062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049378656971023e-05,
+      "loss": 0.9051,
+      "step": 6063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046635418722883e-05,
+      "loss": 0.9845,
+      "step": 6064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043892176965185e-05,
+      "loss": 0.9441,
+      "step": 6065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041148931904364e-05,
+      "loss": 0.9234,
+      "step": 6066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038405683746868e-05,
+      "loss": 0.9538,
+      "step": 6067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035662432699143e-05,
+      "loss": 0.9079,
+      "step": 6068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032919178967628e-05,
+      "loss": 0.92,
+      "step": 6069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030175922758773e-05,
+      "loss": 0.9363,
+      "step": 6070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002743266427902e-05,
+      "loss": 0.9292,
+      "step": 6071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024689403734813e-05,
+      "loss": 0.9501,
+      "step": 6072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021946141332596e-05,
+      "loss": 0.9879,
+      "step": 6073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001920287727882e-05,
+      "loss": 0.9179,
+      "step": 6074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016459611779923e-05,
+      "loss": 0.954,
+      "step": 6075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013716345042353e-05,
+      "loss": 0.9269,
+      "step": 6076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010973077272557e-05,
+      "loss": 0.9067,
+      "step": 6077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008229808676979e-05,
+      "loss": 0.9148,
+      "step": 6078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005486539462061e-05,
+      "loss": 0.9047,
+      "step": 6079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002743269834252e-05,
+      "loss": 0.9204,
+      "step": 6080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.9559,
+      "step": 6081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997256730165748e-06,
+      "loss": 3.2529,
+      "step": 6082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994513460537942e-06,
+      "loss": 0.9411,
+      "step": 6083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.991770191323024e-06,
+      "loss": 0.9622,
+      "step": 6084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989026922727448e-06,
+      "loss": 0.9392,
+      "step": 6085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98628365495765e-06,
+      "loss": 0.9695,
+      "step": 6086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983540388220078e-06,
+      "loss": 0.9057,
+      "step": 6087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980797122721184e-06,
+      "loss": 0.9639,
+      "step": 6088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978053858667406e-06,
+      "loss": 0.9954,
+      "step": 6089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975310596265189e-06,
+      "loss": 0.9527,
+      "step": 6090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.972567335720983e-06,
+      "loss": 0.9595,
+      "step": 6091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969824077241229e-06,
+      "loss": 0.9134,
+      "step": 6092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967080821032375e-06,
+      "loss": 3.2639,
+      "step": 6093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964337567300862e-06,
+      "loss": 0.9521,
+      "step": 6094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961594316253134e-06,
+      "loss": 0.9836,
+      "step": 6095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958851068095641e-06,
+      "loss": 0.9287,
+      "step": 6096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956107823034819e-06,
+      "loss": 0.9461,
+      "step": 6097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953364581277116e-06,
+      "loss": 0.935,
+      "step": 6098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95062134302898e-06,
+      "loss": 0.9858,
+      "step": 6099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947878108496847e-06,
+      "loss": 0.9834,
+      "step": 6100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945134877887163e-06,
+      "loss": 0.9019,
+      "step": 6101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942391651406367e-06,
+      "loss": 0.9235,
+      "step": 6102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939648429260906e-06,
+      "loss": 0.9395,
+      "step": 6103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936905211657221e-06,
+      "loss": 0.8874,
+      "step": 6104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93416199880175e-06,
+      "loss": 0.9631,
+      "step": 6105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931418790900938e-06,
+      "loss": 0.9132,
+      "step": 6106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928675588161229e-06,
+      "loss": 3.279,
+      "step": 6107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925932390789055e-06,
+      "loss": 0.9343,
+      "step": 6108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923189198990869e-06,
+      "loss": 0.9322,
+      "step": 6109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920446012973094e-06,
+      "loss": 0.9008,
+      "step": 6110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917702832942179e-06,
+      "loss": 0.9899,
+      "step": 6111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914959659104563e-06,
+      "loss": 0.9356,
+      "step": 6112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912216491666683e-06,
+      "loss": 0.9461,
+      "step": 6113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909473330834976e-06,
+      "loss": 0.9556,
+      "step": 6114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906730176815881e-06,
+      "loss": 0.9446,
+      "step": 6115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903987029815834e-06,
+      "loss": 0.9616,
+      "step": 6116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901243890041272e-06,
+      "loss": 0.9221,
+      "step": 6117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.898500757698629e-06,
+      "loss": 0.9594,
+      "step": 6118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895757632994342e-06,
+      "loss": 0.9499,
+      "step": 6119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893014516134847e-06,
+      "loss": 0.9818,
+      "step": 6120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890271407326574e-06,
+      "loss": 0.9568,
+      "step": 6121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887528306775962e-06,
+      "loss": 0.9786,
+      "step": 6122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884785214689435e-06,
+      "loss": 0.924,
+      "step": 6123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882042131273432e-06,
+      "loss": 0.9363,
+      "step": 6124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879299056734386e-06,
+      "loss": 0.9735,
+      "step": 6125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876555991278723e-06,
+      "loss": 0.9496,
+      "step": 6126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873812935112874e-06,
+      "loss": 0.9419,
+      "step": 6127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871069888443273e-06,
+      "loss": 0.9718,
+      "step": 6128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868326851476343e-06,
+      "loss": 0.9491,
+      "step": 6129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86558382441852e-06,
+      "loss": 0.9474,
+      "step": 6130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86284080747622e-06,
+      "loss": 0.9427,
+      "step": 6131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860097800855875e-06,
+      "loss": 0.9651,
+      "step": 6132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857354804763917e-06,
+      "loss": 0.9351,
+      "step": 6133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854611819406762e-06,
+      "loss": 0.9367,
+      "step": 6134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851868844990834e-06,
+      "loss": 0.9278,
+      "step": 6135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.849125881722565e-06,
+      "loss": 0.9449,
+      "step": 6136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84638292980837e-06,
+      "loss": 0.9089,
+      "step": 6137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843639989454675e-06,
+      "loss": 0.9671,
+      "step": 6138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840897060867897e-06,
+      "loss": 0.9684,
+      "step": 6139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838154144254457e-06,
+      "loss": 0.9228,
+      "step": 6140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83541123982078e-06,
+      "loss": 0.9192,
+      "step": 6141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83266834777327e-06,
+      "loss": 0.9261,
+      "step": 6142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829925468318356e-06,
+      "loss": 0.953,
+      "step": 6143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827182601662456e-06,
+      "loss": 0.89,
+      "step": 6144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824439748011975e-06,
+      "loss": 0.9547,
+      "step": 6145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821696907573337e-06,
+      "loss": 0.963,
+      "step": 6146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818954080552946e-06,
+      "loss": 0.9122,
+      "step": 6147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81621126715722e-06,
+      "loss": 0.9777,
+      "step": 6148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813468467592573e-06,
+      "loss": 0.9548,
+      "step": 6149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810725682065408e-06,
+      "loss": 0.9308,
+      "step": 6150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807982910782142e-06,
+      "loss": 0.8896,
+      "step": 6151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805240153949175e-06,
+      "loss": 0.9296,
+      "step": 6152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802497411772916e-06,
+      "loss": 0.8933,
+      "step": 6153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799754684459777e-06,
+      "loss": 0.9288,
+      "step": 6154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797011972216156e-06,
+      "loss": 0.9023,
+      "step": 6155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794269275248458e-06,
+      "loss": 0.8813,
+      "step": 6156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791526593763088e-06,
+      "loss": 0.8944,
+      "step": 6157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788783927966446e-06,
+      "loss": 0.9491,
+      "step": 6158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786041278064931e-06,
+      "loss": 0.9305,
+      "step": 6159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783298644264943e-06,
+      "loss": 0.9375,
+      "step": 6160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780556026772879e-06,
+      "loss": 0.955,
+      "step": 6161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77781342579514e-06,
+      "loss": 0.9392,
+      "step": 6162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775070841538111e-06,
+      "loss": 0.9439,
+      "step": 6163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.772328274208193e-06,
+      "loss": 0.9228,
+      "step": 6164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.769585724011784e-06,
+      "loss": 0.9366,
+      "step": 6165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766843191155264e-06,
+      "loss": 0.9171,
+      "step": 6166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76410067584503e-06,
+      "loss": 0.9555,
+      "step": 6167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761358178287469e-06,
+      "loss": 0.9601,
+      "step": 6168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758615698688968e-06,
+      "loss": 0.9335,
+      "step": 6169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755873237255916e-06,
+      "loss": 0.978,
+      "step": 6170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753130794194693e-06,
+      "loss": 0.9492,
+      "step": 6171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750388369711685e-06,
+      "loss": 0.9421,
+      "step": 6172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747645964013277e-06,
+      "loss": 0.8905,
+      "step": 6173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.744903577305844e-06,
+      "loss": 0.9675,
+      "step": 6174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742161209795771e-06,
+      "loss": 0.9849,
+      "step": 6175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739418861689429e-06,
+      "loss": 0.8911,
+      "step": 6176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736676533193199e-06,
+      "loss": 0.9386,
+      "step": 6177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733934224513453e-06,
+      "loss": 0.9161,
+      "step": 6178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731191935856566e-06,
+      "loss": 0.9557,
+      "step": 6179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728449667428907e-06,
+      "loss": 0.95,
+      "step": 6180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725707419436851e-06,
+      "loss": 0.9199,
+      "step": 6181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722965192086765e-06,
+      "loss": 0.9819,
+      "step": 6182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720222985585014e-06,
+      "loss": 0.9627,
+      "step": 6183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717480800137964e-06,
+      "loss": 0.9672,
+      "step": 6184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714738635951979e-06,
+      "loss": 0.9372,
+      "step": 6185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711996493233428e-06,
+      "loss": 0.9363,
+      "step": 6186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70925437218866e-06,
+      "loss": 0.9514,
+      "step": 6187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706512273024044e-06,
+      "loss": 0.9023,
+      "step": 6188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703770195945929e-06,
+      "loss": 0.9062,
+      "step": 6189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701028141160677e-06,
+      "loss": 0.9446,
+      "step": 6190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698286108874642e-06,
+      "loss": 0.9255,
+      "step": 6191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695544099294173e-06,
+      "loss": 0.9523,
+      "step": 6192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692802112625623e-06,
+      "loss": 0.9305,
+      "step": 6193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690060149075344e-06,
+      "loss": 0.9954,
+      "step": 6194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687318208849675e-06,
+      "loss": 0.9395,
+      "step": 6195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684576292154974e-06,
+      "loss": 0.9467,
+      "step": 6196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68183439919757e-06,
+      "loss": 0.9145,
+      "step": 6197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679092530183812e-06,
+      "loss": 0.9159,
+      "step": 6198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676350685320045e-06,
+      "loss": 0.9662,
+      "step": 6199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673608864812601e-06,
+      "loss": 0.9205,
+      "step": 6200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670867068867818e-06,
+      "loss": 0.9495,
+      "step": 6201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668125297692031e-06,
+      "loss": 0.9468,
+      "step": 6202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665383551491572e-06,
+      "loss": 0.9579,
+      "step": 6203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662641830472774e-06,
+      "loss": 0.9651,
+      "step": 6204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659900134841964e-06,
+      "loss": 2.9844,
+      "step": 6205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657158464805467e-06,
+      "loss": 0.877,
+      "step": 6206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654416820569618e-06,
+      "loss": 0.8959,
+      "step": 6207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651675202340724e-06,
+      "loss": 0.933,
+      "step": 6208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64893361032512e-06,
+      "loss": 0.9295,
+      "step": 6209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646192044729124e-06,
+      "loss": 0.8981,
+      "step": 6210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.643450505759046e-06,
+      "loss": 0.9181,
+      "step": 6211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640708993621209e-06,
+      "loss": 0.9639,
+      "step": 6212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63796750852192e-06,
+      "loss": 0.9481,
+      "step": 6213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.635226050667492e-06,
+      "loss": 0.9098,
+      "step": 6214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632484620264237e-06,
+      "loss": 0.9416,
+      "step": 6215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629743217518459e-06,
+      "loss": 0.9447,
+      "step": 6216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627001842636466e-06,
+      "loss": 0.976,
+      "step": 6217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624260495824562e-06,
+      "loss": 0.9594,
+      "step": 6218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621519177289039e-06,
+      "loss": 0.9348,
+      "step": 6219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61877788723621e-06,
+      "loss": 0.9447,
+      "step": 6220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616036625872358e-06,
+      "loss": 2.9744,
+      "step": 6221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.613295393403783e-06,
+      "loss": 0.9932,
+      "step": 6222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610554190036778e-06,
+      "loss": 0.924,
+      "step": 6223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607813015977634e-06,
+      "loss": 0.9019,
+      "step": 6224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605071871432636e-06,
+      "loss": 0.9277,
+      "step": 6225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60233075660807e-06,
+      "loss": 0.9063,
+      "step": 6226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599589671710221e-06,
+      "loss": 0.9369,
+      "step": 6227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596848616945374e-06,
+      "loss": 0.9867,
+      "step": 6228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594107592519795e-06,
+      "loss": 0.9431,
+      "step": 6229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591366598639773e-06,
+      "loss": 0.9963,
+      "step": 6230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588625635511582e-06,
+      "loss": 2.7917,
+      "step": 6231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585884703341484e-06,
+      "loss": 0.9346,
+      "step": 6232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58314380233576e-06,
+      "loss": 0.9294,
+      "step": 6233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580402932700668e-06,
+      "loss": 0.9409,
+      "step": 6234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577662094642478e-06,
+      "loss": 0.9272,
+      "step": 6235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574921288367455e-06,
+      "loss": 0.928,
+      "step": 6236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572180514081853e-06,
+      "loss": 0.9216,
+      "step": 6237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569439771991934e-06,
+      "loss": 0.961,
+      "step": 6238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56669906230395e-06,
+      "loss": 0.9387,
+      "step": 6239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563958385224157e-06,
+      "loss": 0.9416,
+      "step": 6240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56121774095881e-06,
+      "loss": 0.9201,
+      "step": 6241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.558477129714145e-06,
+      "loss": 0.935,
+      "step": 6242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555736551696415e-06,
+      "loss": 0.8982,
+      "step": 6243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552996007111864e-06,
+      "loss": 0.9216,
+      "step": 6244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550255496166728e-06,
+      "loss": 0.961,
+      "step": 6245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547515019067248e-06,
+      "loss": 0.9699,
+      "step": 6246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54477457601966e-06,
+      "loss": 0.9076,
+      "step": 6247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542034167230193e-06,
+      "loss": 0.9027,
+      "step": 6248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539293792905083e-06,
+      "loss": 0.9619,
+      "step": 6249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536553453250553e-06,
+      "loss": 0.9001,
+      "step": 6250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53381314847283e-06,
+      "loss": 0.9468,
+      "step": 6251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531072878778139e-06,
+      "loss": 0.9578,
+      "step": 6252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528332644372694e-06,
+      "loss": 0.9027,
+      "step": 6253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525592445462715e-06,
+      "loss": 0.9467,
+      "step": 6254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522852282254416e-06,
+      "loss": 0.8964,
+      "step": 6255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52011215495401e-06,
+      "loss": 2.7239,
+      "step": 6256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517372063767706e-06,
+      "loss": 0.9378,
+      "step": 6257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51463200890171e-06,
+      "loss": 0.9571,
+      "step": 6258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511891990562224e-06,
+      "loss": 0.9135,
+      "step": 6259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509152008955453e-06,
+      "loss": 0.8587,
+      "step": 6260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50641206428759e-06,
+      "loss": 2.817,
+      "step": 6261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503672156764836e-06,
+      "loss": 0.9231,
+      "step": 6262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50093228659338e-06,
+      "loss": 0.9227,
+      "step": 6263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498192453979405e-06,
+      "loss": 0.9554,
+      "step": 6264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495452659129114e-06,
+      "loss": 0.928,
+      "step": 6265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492712902248678e-06,
+      "loss": 0.9453,
+      "step": 6266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489973183544283e-06,
+      "loss": 0.9243,
+      "step": 6267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487233503222106e-06,
+      "loss": 0.9549,
+      "step": 6268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484493861488323e-06,
+      "loss": 0.9643,
+      "step": 6269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481754258549108e-06,
+      "loss": 0.9586,
+      "step": 6270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479014694610629e-06,
+      "loss": 0.9289,
+      "step": 6271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476275169879052e-06,
+      "loss": 0.959,
+      "step": 6272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473535684560545e-06,
+      "loss": 0.9368,
+      "step": 6273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47079623886126e-06,
+      "loss": 2.6306,
+      "step": 6274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468056832987362e-06,
+      "loss": 0.936,
+      "step": 6275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465317467145008e-06,
+      "loss": 0.9049,
+      "step": 6276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462578141540342e-06,
+      "loss": 0.9158,
+      "step": 6277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459838856379518e-06,
+      "loss": 1.0113,
+      "step": 6278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45709961186868e-06,
+      "loss": 0.9709,
+      "step": 6279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454360408213969e-06,
+      "loss": 1.0064,
+      "step": 6280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45162124562153e-06,
+      "loss": 0.9528,
+      "step": 6281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448882124297493e-06,
+      "loss": 0.9552,
+      "step": 6282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446143044447997e-06,
+      "loss": 0.9492,
+      "step": 6283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443404006279171e-06,
+      "loss": 0.9595,
+      "step": 6284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440665009997135e-06,
+      "loss": 0.961,
+      "step": 6285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437926055808027e-06,
+      "loss": 0.9617,
+      "step": 6286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435187143917954e-06,
+      "loss": 0.9662,
+      "step": 6287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432448274533039e-06,
+      "loss": 0.9233,
+      "step": 6288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4297094478594e-06,
+      "loss": 0.9573,
+      "step": 6289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42697066410314e-06,
+      "loss": 0.9291,
+      "step": 6290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424231923470378e-06,
+      "loss": 0.9082,
+      "step": 6291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.421493226167208e-06,
+      "loss": 0.8903,
+      "step": 6292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418754572399735e-06,
+      "loss": 0.9725,
+      "step": 6293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.416015962374066e-06,
+      "loss": 0.953,
+      "step": 6294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41327739629628e-06,
+      "loss": 0.9191,
+      "step": 6295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410538874372477e-06,
+      "loss": 2.6233,
+      "step": 6296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407800396808752e-06,
+      "loss": 0.8731,
+      "step": 6297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.405061963811176e-06,
+      "loss": 0.8912,
+      "step": 6298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402323575585843e-06,
+      "loss": 0.9212,
+      "step": 6299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.399585232338821e-06,
+      "loss": 0.9284,
+      "step": 6300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396846934276191e-06,
+      "loss": 0.9557,
+      "step": 6301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394108681604025e-06,
+      "loss": 0.9476,
+      "step": 6302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391370474528387e-06,
+      "loss": 0.9377,
+      "step": 6303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388632313255343e-06,
+      "loss": 2.7687,
+      "step": 6304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385894197990958e-06,
+      "loss": 0.9148,
+      "step": 6305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383156128941283e-06,
+      "loss": 0.9751,
+      "step": 6306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38041810631238e-06,
+      "loss": 0.9497,
+      "step": 6307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377680130310292e-06,
+      "loss": 0.9469,
+      "step": 6308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374942201141069e-06,
+      "loss": 0.9273,
+      "step": 6309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372204319010759e-06,
+      "loss": 0.9197,
+      "step": 6310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369466484125394e-06,
+      "loss": 0.9028,
+      "step": 6311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366728696691014e-06,
+      "loss": 0.9431,
+      "step": 6312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363990956913657e-06,
+      "loss": 0.9859,
+      "step": 6313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361253264999346e-06,
+      "loss": 0.9107,
+      "step": 6314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358515621154112e-06,
+      "loss": 0.9227,
+      "step": 6315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35577802558397e-06,
+      "loss": 0.9555,
+      "step": 6316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353040478494944e-06,
+      "loss": 0.8789,
+      "step": 6317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350302980093053e-06,
+      "loss": 0.9308,
+      "step": 6318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347565530584299e-06,
+      "loss": 0.9602,
+      "step": 6319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.344828130174689e-06,
+      "loss": 0.9489,
+      "step": 6320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34209077907024e-06,
+      "loss": 0.8842,
+      "step": 6321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339353477476938e-06,
+      "loss": 0.9384,
+      "step": 6322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336616225600788e-06,
+      "loss": 0.9119,
+      "step": 6323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333879023647779e-06,
+      "loss": 0.8341,
+      "step": 6324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331141871823899e-06,
+      "loss": 0.9371,
+      "step": 6325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328404770335138e-06,
+      "loss": 0.9196,
+      "step": 6326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325667719387472e-06,
+      "loss": 0.9403,
+      "step": 6327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322930719186884e-06,
+      "loss": 0.935,
+      "step": 6328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32019376993934e-06,
+      "loss": 0.8937,
+      "step": 6329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317456871850816e-06,
+      "loss": 0.9267,
+      "step": 6330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31472002512728e-06,
+      "loss": 0.9714,
+      "step": 6331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311983229974687e-06,
+      "loss": 0.9169,
+      "step": 6332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.309246486599e-06,
+      "loss": 0.9386,
+      "step": 6333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306509795206174e-06,
+      "loss": 0.9289,
+      "step": 6334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303773156002154e-06,
+      "loss": 0.9242,
+      "step": 6335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301036569192896e-06,
+      "loss": 0.9871,
+      "step": 6336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298300034984335e-06,
+      "loss": 0.8974,
+      "step": 6337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295563553582411e-06,
+      "loss": 0.9539,
+      "step": 6338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292827125193063e-06,
+      "loss": 0.9241,
+      "step": 6339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290090750022213e-06,
+      "loss": 0.9144,
+      "step": 6340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287354428275797e-06,
+      "loss": 0.8993,
+      "step": 6341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284618160159736e-06,
+      "loss": 0.9402,
+      "step": 6342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281881945879944e-06,
+      "loss": 0.9404,
+      "step": 6343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27914578564234e-06,
+      "loss": 0.9271,
+      "step": 6344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27640967965283e-06,
+      "loss": 0.8833,
+      "step": 6345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273673628117323e-06,
+      "loss": 0.9592,
+      "step": 6346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270937631241723e-06,
+      "loss": 0.9802,
+      "step": 6347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268201689231925e-06,
+      "loss": 0.9469,
+      "step": 6348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265465802293826e-06,
+      "loss": 0.9813,
+      "step": 6349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26272997063332e-06,
+      "loss": 0.9096,
+      "step": 6350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25999419445628e-06,
+      "loss": 0.9706,
+      "step": 6351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257258473968602e-06,
+      "loss": 0.929,
+      "step": 6352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.254522809376152e-06,
+      "loss": 0.9624,
+      "step": 6353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251787200884808e-06,
+      "loss": 0.9284,
+      "step": 6354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249051648700442e-06,
+      "loss": 0.9086,
+      "step": 6355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246316153028913e-06,
+      "loss": 0.9524,
+      "step": 6356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243580714076087e-06,
+      "loss": 0.9438,
+      "step": 6357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240845332047816e-06,
+      "loss": 0.9605,
+      "step": 6358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238110007149954e-06,
+      "loss": 0.9437,
+      "step": 6359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.235374739588351e-06,
+      "loss": 0.8954,
+      "step": 6360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232639529568843e-06,
+      "loss": 0.9541,
+      "step": 6361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229904377297276e-06,
+      "loss": 2.6351,
+      "step": 6362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227169282979485e-06,
+      "loss": 0.9277,
+      "step": 6363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224434246821295e-06,
+      "loss": 0.9604,
+      "step": 6364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221699269028537e-06,
+      "loss": 0.9386,
+      "step": 6365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218964349807027e-06,
+      "loss": 0.9595,
+      "step": 6366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216229489362587e-06,
+      "loss": 0.9623,
+      "step": 6367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213494687901028e-06,
+      "loss": 0.9442,
+      "step": 6368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210759945628158e-06,
+      "loss": 0.9409,
+      "step": 6369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20802526274978e-06,
+      "loss": 0.9249,
+      "step": 6370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205290639471698e-06,
+      "loss": 2.6922,
+      "step": 6371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202556075999701e-06,
+      "loss": 0.926,
+      "step": 6372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199821572539587e-06,
+      "loss": 0.9655,
+      "step": 6373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197087129297131e-06,
+      "loss": 0.8863,
+      "step": 6374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194352746478122e-06,
+      "loss": 0.9335,
+      "step": 6375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.191618424288337e-06,
+      "loss": 0.9395,
+      "step": 6376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188884162933542e-06,
+      "loss": 0.9044,
+      "step": 6377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186149962619511e-06,
+      "loss": 2.5538,
+      "step": 6378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183415823552005e-06,
+      "loss": 0.9411,
+      "step": 6379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180681745936781e-06,
+      "loss": 0.8997,
+      "step": 6380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177947729979597e-06,
+      "loss": 0.9168,
+      "step": 6381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175213775886197e-06,
+      "loss": 2.7215,
+      "step": 6382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172479883862326e-06,
+      "loss": 0.9535,
+      "step": 6383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169746054113734e-06,
+      "loss": 2.7174,
+      "step": 6384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16701228684614e-06,
+      "loss": 0.9227,
+      "step": 6385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16427858226528e-06,
+      "loss": 0.9106,
+      "step": 6386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16154494057689e-06,
+      "loss": 0.9535,
+      "step": 6387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158811361986679e-06,
+      "loss": 0.909,
+      "step": 6388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156077846700367e-06,
+      "loss": 0.9254,
+      "step": 6389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153344394923665e-06,
+      "loss": 0.9249,
+      "step": 6390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150611006862282e-06,
+      "loss": 0.8883,
+      "step": 6391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14787768272192e-06,
+      "loss": 0.9202,
+      "step": 6392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14514442270827e-06,
+      "loss": 0.9047,
+      "step": 6393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142411227027037e-06,
+      "loss": 0.9107,
+      "step": 6394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139678095883894e-06,
+      "loss": 0.9393,
+      "step": 6395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136945029484528e-06,
+      "loss": 0.9531,
+      "step": 6396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134212028034625e-06,
+      "loss": 0.9503,
+      "step": 6397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131479091739846e-06,
+      "loss": 0.9105,
+      "step": 6398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128746220805867e-06,
+      "loss": 0.9442,
+      "step": 6399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126013415438348e-06,
+      "loss": 0.9094,
+      "step": 6400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123280675842947e-06,
+      "loss": 0.9415,
+      "step": 6401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120548002225318e-06,
+      "loss": 0.9373,
+      "step": 6402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117815394791107e-06,
+      "loss": 0.9742,
+      "step": 6403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11508285374596e-06,
+      "loss": 0.9024,
+      "step": 6404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112350379295519e-06,
+      "loss": 0.9517,
+      "step": 6405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109617971645405e-06,
+      "loss": 0.9866,
+      "step": 6406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106885631001256e-06,
+      "loss": 0.94,
+      "step": 6407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104153357568695e-06,
+      "loss": 0.8913,
+      "step": 6408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.101421151553335e-06,
+      "loss": 0.8867,
+      "step": 6409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098689013160793e-06,
+      "loss": 0.981,
+      "step": 6410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.095956942596673e-06,
+      "loss": 0.9561,
+      "step": 6411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.093224940066583e-06,
+      "loss": 0.919,
+      "step": 6412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090493005776117e-06,
+      "loss": 0.951,
+      "step": 6413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087761139930866e-06,
+      "loss": 0.9307,
+      "step": 6414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085029342736422e-06,
+      "loss": 2.6672,
+      "step": 6415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082297614398367e-06,
+      "loss": 0.9223,
+      "step": 6416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07956595512227e-06,
+      "loss": 0.919,
+      "step": 6417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076834365113715e-06,
+      "loss": 0.9267,
+      "step": 6418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07410284457826e-06,
+      "loss": 0.9174,
+      "step": 6419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071371393721468e-06,
+      "loss": 0.9443,
+      "step": 6420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068640012748897e-06,
+      "loss": 0.9039,
+      "step": 6421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065908701866097e-06,
+      "loss": 0.9353,
+      "step": 6422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06317746127861e-06,
+      "loss": 0.9073,
+      "step": 6423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060446291191985e-06,
+      "loss": 0.9792,
+      "step": 6424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057715191811748e-06,
+      "loss": 0.9693,
+      "step": 6425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054984163343438e-06,
+      "loss": 0.9153,
+      "step": 6426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052253205992566e-06,
+      "loss": 0.8933,
+      "step": 6427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04952231996466e-06,
+      "loss": 0.9477,
+      "step": 6428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04679150546524e-06,
+      "loss": 0.9265,
+      "step": 6429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0440607626998e-06,
+      "loss": 0.9362,
+      "step": 6430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041330091873852e-06,
+      "loss": 0.9519,
+      "step": 6431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03859949319289e-06,
+      "loss": 0.9433,
+      "step": 6432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035868966862405e-06,
+      "loss": 0.9416,
+      "step": 6433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03313851308789e-06,
+      "loss": 0.9379,
+      "step": 6434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030408132074818e-06,
+      "loss": 0.9212,
+      "step": 6435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02767782402867e-06,
+      "loss": 0.9293,
+      "step": 6436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024947589154913e-06,
+      "loss": 0.9373,
+      "step": 6437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022217427659013e-06,
+      "loss": 0.9709,
+      "step": 6438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019487339746435e-06,
+      "loss": 0.9532,
+      "step": 6439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01675732562262e-06,
+      "loss": 0.9054,
+      "step": 6440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014027385493025e-06,
+      "loss": 0.9246,
+      "step": 6441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011297519563091e-06,
+      "loss": 0.892,
+      "step": 6442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008567728038252e-06,
+      "loss": 0.9377,
+      "step": 6443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005838011123941e-06,
+      "loss": 0.9122,
+      "step": 6444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003108369025588e-06,
+      "loss": 0.9306,
+      "step": 6445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000378801948603e-06,
+      "loss": 0.9582,
+      "step": 6446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997649310098412e-06,
+      "loss": 0.8972,
+      "step": 6447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994919893680415e-06,
+      "loss": 0.9132,
+      "step": 6448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992190552900019e-06,
+      "loss": 0.9788,
+      "step": 6449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989461287962622e-06,
+      "loss": 0.9143,
+      "step": 6450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986732099073613e-06,
+      "loss": 0.979,
+      "step": 6451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984002986438376e-06,
+      "loss": 0.9436,
+      "step": 6452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9812739502623e-06,
+      "loss": 0.9208,
+      "step": 6453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978544990750751e-06,
+      "loss": 0.9369,
+      "step": 6454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9758161081091e-06,
+      "loss": 0.9538,
+      "step": 6455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97308730254271e-06,
+      "loss": 0.9057,
+      "step": 6456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970358574256937e-06,
+      "loss": 0.9182,
+      "step": 6457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967629923457137e-06,
+      "loss": 0.9461,
+      "step": 6458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964901350348648e-06,
+      "loss": 0.9589,
+      "step": 6459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962172855136815e-06,
+      "loss": 0.9309,
+      "step": 6460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959444438026973e-06,
+      "loss": 0.9365,
+      "step": 6461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956716099224442e-06,
+      "loss": 0.8622,
+      "step": 6462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953987838934555e-06,
+      "loss": 0.9292,
+      "step": 6463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951259657362617e-06,
+      "loss": 0.9498,
+      "step": 6464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.948531554713944e-06,
+      "loss": 0.9238,
+      "step": 6465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945803531193841e-06,
+      "loss": 0.9629,
+      "step": 6466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943075587007603e-06,
+      "loss": 0.9351,
+      "step": 6467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940347722360526e-06,
+      "loss": 0.9349,
+      "step": 6468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937619937457891e-06,
+      "loss": 0.9154,
+      "step": 6469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934892232504982e-06,
+      "loss": 0.9539,
+      "step": 6470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932164607707075e-06,
+      "loss": 0.9228,
+      "step": 6471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92943706326943e-06,
+      "loss": 0.9067,
+      "step": 6472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926709599397319e-06,
+      "loss": 0.9409,
+      "step": 6473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923982216295996e-06,
+      "loss": 0.9261,
+      "step": 6474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921254914170706e-06,
+      "loss": 0.9056,
+      "step": 6475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918527693226699e-06,
+      "loss": 0.957,
+      "step": 6476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915800553669207e-06,
+      "loss": 0.9232,
+      "step": 6477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913073495703463e-06,
+      "loss": 0.9432,
+      "step": 6478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910346519534696e-06,
+      "loss": 2.7352,
+      "step": 6479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907619625368123e-06,
+      "loss": 0.9248,
+      "step": 6480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904892813408956e-06,
+      "loss": 0.9483,
+      "step": 6481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902166083862409e-06,
+      "loss": 0.9325,
+      "step": 6482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.899439436933672e-06,
+      "loss": 0.9262,
+      "step": 6483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89671287282795e-06,
+      "loss": 0.9234,
+      "step": 6484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893986391750422e-06,
+      "loss": 0.9683,
+      "step": 6485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891259993906275e-06,
+      "loss": 0.946,
+      "step": 6486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888533679500688e-06,
+      "loss": 0.9029,
+      "step": 6487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885807448738824e-06,
+      "loss": 0.9188,
+      "step": 6488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883081301825848e-06,
+      "loss": 0.9427,
+      "step": 6489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.9104,
+      "step": 6490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877629260367192e-06,
+      "loss": 0.9272,
+      "step": 6491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874903366231807e-06,
+      "loss": 0.9504,
+      "step": 6492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872177556765893e-06,
+      "loss": 0.9412,
+      "step": 6493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869451832174593e-06,
+      "loss": 0.9407,
+      "step": 6494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866726192663035e-06,
+      "loss": 0.9374,
+      "step": 6495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864000638436327e-06,
+      "loss": 0.9433,
+      "step": 6496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861275169699588e-06,
+      "loss": 0.9244,
+      "step": 6497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85854978665792e-06,
+      "loss": 2.7073,
+      "step": 6498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.855824489516425e-06,
+      "loss": 0.9785,
+      "step": 6499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853099278480198e-06,
+      "loss": 0.9353,
+      "step": 6500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850374153754322e-06,
+      "loss": 0.9251,
+      "step": 6501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847649115543877e-06,
+      "loss": 2.6847,
+      "step": 6502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844924164053942e-06,
+      "loss": 0.9508,
+      "step": 6503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842199299489576e-06,
+      "loss": 0.9572,
+      "step": 6504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83947452205585e-06,
+      "loss": 0.894,
+      "step": 6505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836749831957806e-06,
+      "loss": 0.9429,
+      "step": 6506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834025229400497e-06,
+      "loss": 0.9532,
+      "step": 6507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831300714588964e-06,
+      "loss": 0.9125,
+      "step": 6508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82857628772824e-06,
+      "loss": 0.9626,
+      "step": 6509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825851949023355e-06,
+      "loss": 0.9424,
+      "step": 6510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823127698679329e-06,
+      "loss": 0.9723,
+      "step": 6511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820403536901172e-06,
+      "loss": 0.9101,
+      "step": 6512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8176794638939e-06,
+      "loss": 0.9719,
+      "step": 6513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814955479862506e-06,
+      "loss": 0.973,
+      "step": 6514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812231585011987e-06,
+      "loss": 0.9363,
+      "step": 6515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.809507779547334e-06,
+      "loss": 0.9915,
+      "step": 6516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806784063673521e-06,
+      "loss": 0.9215,
+      "step": 6517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804060437595525e-06,
+      "loss": 0.9382,
+      "step": 6518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801336901518318e-06,
+      "loss": 0.9425,
+      "step": 6519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798613455646852e-06,
+      "loss": 0.9231,
+      "step": 6520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795890100186089e-06,
+      "loss": 0.8897,
+      "step": 6521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79316683534097e-06,
+      "loss": 0.9223,
+      "step": 6522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790443661316434e-06,
+      "loss": 0.9563,
+      "step": 6523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787720578317422e-06,
+      "loss": 0.949,
+      "step": 6524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784997586548854e-06,
+      "loss": 0.9798,
+      "step": 6525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78227468621565e-06,
+      "loss": 0.9121,
+      "step": 6526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779551877522729e-06,
+      "loss": 0.9,
+      "step": 6527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776829160674987e-06,
+      "loss": 0.8751,
+      "step": 6528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774106535877331e-06,
+      "loss": 0.9366,
+      "step": 6529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771384003334648e-06,
+      "loss": 0.9096,
+      "step": 6530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768661563251822e-06,
+      "loss": 0.8864,
+      "step": 6531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765939215833738e-06,
+      "loss": 0.93,
+      "step": 6532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763216961285259e-06,
+      "loss": 0.8951,
+      "step": 6533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760494799811257e-06,
+      "loss": 0.9405,
+      "step": 6534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757772731616581e-06,
+      "loss": 0.8831,
+      "step": 6535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755050756906085e-06,
+      "loss": 0.96,
+      "step": 6536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752328875884617e-06,
+      "loss": 0.8888,
+      "step": 6537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749607088756999e-06,
+      "loss": 0.9049,
+      "step": 6538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746885395728074e-06,
+      "loss": 0.9677,
+      "step": 6539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74416379700266e-06,
+      "loss": 0.8837,
+      "step": 6540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741442292785567e-06,
+      "loss": 0.884,
+      "step": 6541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738720883281608e-06,
+      "loss": 0.9231,
+      "step": 6542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735999568695581e-06,
+      "loss": 0.9333,
+      "step": 6543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733278349232277e-06,
+      "loss": 0.946,
+      "step": 6544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730557225096488e-06,
+      "loss": 0.9328,
+      "step": 6545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727836196492987e-06,
+      "loss": 0.9606,
+      "step": 6546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72511526362655e-06,
+      "loss": 0.915,
+      "step": 6547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.722394426701943e-06,
+      "loss": 0.9361,
+      "step": 6548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719673685923915e-06,
+      "loss": 0.9041,
+      "step": 6549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71695304149723e-06,
+      "loss": 0.9139,
+      "step": 6550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714232493626615e-06,
+      "loss": 0.9507,
+      "step": 6551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711512042516816e-06,
+      "loss": 0.9441,
+      "step": 6552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.70879168837256e-06,
+      "loss": 0.9323,
+      "step": 6553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706071431398564e-06,
+      "loss": 0.9149,
+      "step": 6554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703351271799545e-06,
+      "loss": 0.9472,
+      "step": 6555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70063120978021e-06,
+      "loss": 0.9142,
+      "step": 6556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697911245545257e-06,
+      "loss": 0.923,
+      "step": 6557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69519137929938e-06,
+      "loss": 2.7428,
+      "step": 6558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692471611247255e-06,
+      "loss": 0.9098,
+      "step": 6559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689751941593568e-06,
+      "loss": 0.9042,
+      "step": 6560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687032370542988e-06,
+      "loss": 0.9306,
+      "step": 6561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684312898300172e-06,
+      "loss": 0.9105,
+      "step": 6562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681593525069775e-06,
+      "loss": 0.9262,
+      "step": 6563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.678874251056452e-06,
+      "loss": 0.9326,
+      "step": 6564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676155076464834e-06,
+      "loss": 0.8904,
+      "step": 6565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673436001499557e-06,
+      "loss": 0.9348,
+      "step": 6566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670717026365245e-06,
+      "loss": 2.5737,
+      "step": 6567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667998151266515e-06,
+      "loss": 0.9398,
+      "step": 6568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66527937640798e-06,
+      "loss": 2.6857,
+      "step": 6569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662560701994237e-06,
+      "loss": 0.9254,
+      "step": 6570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659842128229886e-06,
+      "loss": 0.9445,
+      "step": 6571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65712365531951e-06,
+      "loss": 0.9275,
+      "step": 6572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654405283467689e-06,
+      "loss": 0.9245,
+      "step": 6573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651687012878997e-06,
+      "loss": 0.9427,
+      "step": 6574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648968843757995e-06,
+      "loss": 0.9707,
+      "step": 6575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646250776309242e-06,
+      "loss": 0.8871,
+      "step": 6576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643532810737288e-06,
+      "loss": 0.9443,
+      "step": 6577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640814947246673e-06,
+      "loss": 0.939,
+      "step": 6578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63809718604193e-06,
+      "loss": 0.9244,
+      "step": 6579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635379527327583e-06,
+      "loss": 0.9637,
+      "step": 6580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632661971308154e-06,
+      "loss": 0.9841,
+      "step": 6581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629944518188156e-06,
+      "loss": 0.9352,
+      "step": 6582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627227168172083e-06,
+      "loss": 0.9337,
+      "step": 6583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624509921464432e-06,
+      "loss": 0.9067,
+      "step": 6584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6217927782697e-06,
+      "loss": 0.9384,
+      "step": 6585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619075738792357e-06,
+      "loss": 0.9421,
+      "step": 6586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616358803236878e-06,
+      "loss": 0.8961,
+      "step": 6587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613641971807723e-06,
+      "loss": 0.9506,
+      "step": 6588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610925244709349e-06,
+      "loss": 0.9191,
+      "step": 6589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60820862214621e-06,
+      "loss": 0.8728,
+      "step": 6590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60549210432274e-06,
+      "loss": 0.9435,
+      "step": 6591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602775691443374e-06,
+      "loss": 0.9456,
+      "step": 6592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600059383712539e-06,
+      "loss": 0.8875,
+      "step": 6593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597343181334644e-06,
+      "loss": 0.93,
+      "step": 6594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594627084514106e-06,
+      "loss": 0.9087,
+      "step": 6595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591911093455317e-06,
+      "loss": 0.9298,
+      "step": 6596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589195208362677e-06,
+      "loss": 0.9096,
+      "step": 6597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586479429440569e-06,
+      "loss": 0.9129,
+      "step": 6598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583763756893366e-06,
+      "loss": 0.912,
+      "step": 6599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581048190925442e-06,
+      "loss": 0.9064,
+      "step": 6600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578332731741155e-06,
+      "loss": 0.9426,
+      "step": 6601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575617379544857e-06,
+      "loss": 0.9341,
+      "step": 6602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572902134540897e-06,
+      "loss": 0.9182,
+      "step": 6603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570186996933601e-06,
+      "loss": 0.9303,
+      "step": 6604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56747196692731e-06,
+      "loss": 0.9347,
+      "step": 6605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56475704472634e-06,
+      "loss": 0.9263,
+      "step": 6606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562042230535001e-06,
+      "loss": 0.9227,
+      "step": 6607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559327524557599e-06,
+      "loss": 0.9611,
+      "step": 6608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556612926998427e-06,
+      "loss": 0.9478,
+      "step": 6609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553898438061777e-06,
+      "loss": 0.8834,
+      "step": 6610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551184057951927e-06,
+      "loss": 0.9148,
+      "step": 6611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548469786873149e-06,
+      "loss": 0.9339,
+      "step": 6612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545755625029704e-06,
+      "loss": 0.953,
+      "step": 6613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543041572625851e-06,
+      "loss": 0.907,
+      "step": 6614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540327629865832e-06,
+      "loss": 0.9331,
+      "step": 6615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537613796953893e-06,
+      "loss": 0.9054,
+      "step": 6616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534900074094256e-06,
+      "loss": 0.9649,
+      "step": 6617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532186461491145e-06,
+      "loss": 0.9182,
+      "step": 6618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52947295934878e-06,
+      "loss": 0.9437,
+      "step": 6619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526759567871356e-06,
+      "loss": 0.9274,
+      "step": 6620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524046287263076e-06,
+      "loss": 0.9566,
+      "step": 6621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521333117728133e-06,
+      "loss": 0.9188,
+      "step": 6622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5186200594707e-06,
+      "loss": 0.8946,
+      "step": 6623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515907112694955e-06,
+      "loss": 0.9098,
+      "step": 6624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513194277605054e-06,
+      "loss": 0.9779,
+      "step": 6625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510481554405157e-06,
+      "loss": 2.7023,
+      "step": 6626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507768943299415e-06,
+      "loss": 0.9285,
+      "step": 6627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505056444491958e-06,
+      "loss": 0.9362,
+      "step": 6628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502344058186917e-06,
+      "loss": 0.9505,
+      "step": 6629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499631784588422e-06,
+      "loss": 0.9247,
+      "step": 6630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496919623900576e-06,
+      "loss": 0.9355,
+      "step": 6631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49420757632749e-06,
+      "loss": 0.9196,
+      "step": 6632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491495642073256e-06,
+      "loss": 0.8893,
+      "step": 6633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488783821341961e-06,
+      "loss": 0.9233,
+      "step": 6634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48607211433769e-06,
+      "loss": 0.8957,
+      "step": 6635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483360521264506e-06,
+      "loss": 0.958,
+      "step": 6636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480649042326477e-06,
+      "loss": 0.9606,
+      "step": 6637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47793767772765e-06,
+      "loss": 0.9086,
+      "step": 6638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475226427672072e-06,
+      "loss": 0.9199,
+      "step": 6639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472515292363783e-06,
+      "loss": 0.913,
+      "step": 6640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.469804272006802e-06,
+      "loss": 0.9406,
+      "step": 6641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467093366805153e-06,
+      "loss": 0.8604,
+      "step": 6642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464382576962849e-06,
+      "loss": 0.9661,
+      "step": 6643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461671902683884e-06,
+      "loss": 0.9044,
+      "step": 6644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458961344172256e-06,
+      "loss": 0.9366,
+      "step": 6645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456250901631945e-06,
+      "loss": 0.9081,
+      "step": 6646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453540575266928e-06,
+      "loss": 0.8638,
+      "step": 6647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450830365281177e-06,
+      "loss": 0.9306,
+      "step": 6648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448120271878639e-06,
+      "loss": 0.9355,
+      "step": 6649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445410295263267e-06,
+      "loss": 0.9529,
+      "step": 6650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442700435639009e-06,
+      "loss": 0.8916,
+      "step": 6651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439990693209783e-06,
+      "loss": 0.9435,
+      "step": 6652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437281068179521e-06,
+      "loss": 0.9638,
+      "step": 6653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434571560752133e-06,
+      "loss": 0.8865,
+      "step": 6654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431862171131524e-06,
+      "loss": 0.9306,
+      "step": 6655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429152899521591e-06,
+      "loss": 0.9252,
+      "step": 6656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426443746126218e-06,
+      "loss": 0.9264,
+      "step": 6657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.423734711149288e-06,
+      "loss": 0.9263,
+      "step": 6658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42102579479467e-06,
+      "loss": 0.9508,
+      "step": 6659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418316997266215e-06,
+      "loss": 0.9415,
+      "step": 6660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41560831876779e-06,
+      "loss": 0.9382,
+      "step": 6661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412899759503223e-06,
+      "loss": 0.9533,
+      "step": 6662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410191319676353e-06,
+      "loss": 0.8633,
+      "step": 6663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407482999491007e-06,
+      "loss": 0.888,
+      "step": 6664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404774799150996e-06,
+      "loss": 0.9193,
+      "step": 6665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402066718860129e-06,
+      "loss": 0.9461,
+      "step": 6666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399358758822204e-06,
+      "loss": 0.9401,
+      "step": 6667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396650919241007e-06,
+      "loss": 0.9059,
+      "step": 6668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393943200320323e-06,
+      "loss": 0.9224,
+      "step": 6669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39123560226391e-06,
+      "loss": 0.9203,
+      "step": 6670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388528125275539e-06,
+      "loss": 0.9776,
+      "step": 6671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385820769558965e-06,
+      "loss": 0.9528,
+      "step": 6672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383113535317921e-06,
+      "loss": 0.9717,
+      "step": 6673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380406422756146e-06,
+      "loss": 0.9361,
+      "step": 6674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377699432077365e-06,
+      "loss": 0.9126,
+      "step": 6675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374992563485289e-06,
+      "loss": 0.9437,
+      "step": 6676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37228581718363e-06,
+      "loss": 0.9415,
+      "step": 6677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369579193376081e-06,
+      "loss": 0.9244,
+      "step": 6678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366872692266332e-06,
+      "loss": 0.9255,
+      "step": 6679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364166314058064e-06,
+      "loss": 0.9401,
+      "step": 6680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361460058954937e-06,
+      "loss": 0.9583,
+      "step": 6681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358753927160623e-06,
+      "loss": 2.8384,
+      "step": 6682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356047918878762e-06,
+      "loss": 0.9867,
+      "step": 6683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353342034313001e-06,
+      "loss": 0.8931,
+      "step": 6684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350636273666975e-06,
+      "loss": 0.9141,
+      "step": 6685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347930637144299e-06,
+      "loss": 0.9253,
+      "step": 6686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34522512494859e-06,
+      "loss": 0.9472,
+      "step": 6687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342519737283458e-06,
+      "loss": 0.9463,
+      "step": 6688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339814474352488e-06,
+      "loss": 0.9131,
+      "step": 6689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337109336359275e-06,
+      "loss": 0.9215,
+      "step": 6690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334404323507384e-06,
+      "loss": 0.9127,
+      "step": 6691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331699436000388e-06,
+      "loss": 2.7502,
+      "step": 6692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328994674041848e-06,
+      "loss": 0.912,
+      "step": 6693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326290037835304e-06,
+      "loss": 0.9279,
+      "step": 6694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323585527584295e-06,
+      "loss": 0.9418,
+      "step": 6695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320881143492358e-06,
+      "loss": 0.9876,
+      "step": 6696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318176885763002e-06,
+      "loss": 0.8978,
+      "step": 6697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315472754599742e-06,
+      "loss": 0.8959,
+      "step": 6698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312768750206077e-06,
+      "loss": 0.8993,
+      "step": 6699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310064872785498e-06,
+      "loss": 0.8842,
+      "step": 6700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307361122541487e-06,
+      "loss": 0.9451,
+      "step": 6701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304657499677513e-06,
+      "loss": 0.9077,
+      "step": 6702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301954004397043e-06,
+      "loss": 2.6219,
+      "step": 6703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299250636903521e-06,
+      "loss": 0.8851,
+      "step": 6704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296547397400396e-06,
+      "loss": 0.9634,
+      "step": 6705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293844286091101e-06,
+      "loss": 2.691,
+      "step": 6706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.291141303179057e-06,
+      "loss": 0.9067,
+      "step": 6707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288438448867678e-06,
+      "loss": 0.9156,
+      "step": 6708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28573572336037e-06,
+      "loss": 0.96,
+      "step": 6709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283033126860525e-06,
+      "loss": 0.9543,
+      "step": 6710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280330659571532e-06,
+      "loss": 0.883,
+      "step": 6711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277628321696759e-06,
+      "loss": 0.9426,
+      "step": 6712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274926113439576e-06,
+      "loss": 0.9136,
+      "step": 6713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272224035003343e-06,
+      "loss": 0.9639,
+      "step": 6714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269522086591393e-06,
+      "loss": 0.9268,
+      "step": 6715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26682026840707e-06,
+      "loss": 0.9372,
+      "step": 6716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264118580653705e-06,
+      "loss": 0.9389,
+      "step": 6717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261417023534604e-06,
+      "loss": 0.957,
+      "step": 6718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25871559725308e-06,
+      "loss": 0.9395,
+      "step": 6719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256014302012427e-06,
+      "loss": 0.9625,
+      "step": 6720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253313138015933e-06,
+      "loss": 0.962,
+      "step": 6721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250612105466876e-06,
+      "loss": 0.9449,
+      "step": 6722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247911204568519e-06,
+      "loss": 0.9671,
+      "step": 6723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245210435524122e-06,
+      "loss": 0.949,
+      "step": 6724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242509798536935e-06,
+      "loss": 0.9098,
+      "step": 6725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239809293810186e-06,
+      "loss": 0.9393,
+      "step": 6726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237108921547116e-06,
+      "loss": 0.9519,
+      "step": 6727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234408681950929e-06,
+      "loss": 0.8864,
+      "step": 6728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231708575224838e-06,
+      "loss": 0.9312,
+      "step": 6729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229008601572043e-06,
+      "loss": 0.9058,
+      "step": 6730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226308761195726e-06,
+      "loss": 0.9185,
+      "step": 6731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223609054299067e-06,
+      "loss": 0.9602,
+      "step": 6732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220909481085235e-06,
+      "loss": 0.951,
+      "step": 6733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218210041757384e-06,
+      "loss": 0.9397,
+      "step": 6734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215510736518667e-06,
+      "loss": 0.9339,
+      "step": 6735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.212811565572209e-06,
+      "loss": 0.9633,
+      "step": 6736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210112529121146e-06,
+      "loss": 0.8585,
+      "step": 6737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207413627368597e-06,
+      "loss": 0.903,
+      "step": 6738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204714860517662e-06,
+      "loss": 0.845,
+      "step": 6739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202016228771442e-06,
+      "loss": 0.9047,
+      "step": 6740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199317732333019e-06,
+      "loss": 0.9226,
+      "step": 6741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196619371405473e-06,
+      "loss": 0.9414,
+      "step": 6742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193921146191868e-06,
+      "loss": 0.9486,
+      "step": 6743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19122305689526e-06,
+      "loss": 0.9301,
+      "step": 6744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188525103718694e-06,
+      "loss": 0.9332,
+      "step": 6745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185827286865212e-06,
+      "loss": 0.9561,
+      "step": 6746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183129606537826e-06,
+      "loss": 0.8981,
+      "step": 6747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180432062939565e-06,
+      "loss": 0.9547,
+      "step": 6748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17773465627342e-06,
+      "loss": 0.9985,
+      "step": 6749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175037386742394e-06,
+      "loss": 0.9084,
+      "step": 6750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17234025454947e-06,
+      "loss": 0.9249,
+      "step": 6751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169643259897614e-06,
+      "loss": 0.8838,
+      "step": 6752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1669464029898e-06,
+      "loss": 0.9124,
+      "step": 6753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164249684028976e-06,
+      "loss": 0.9344,
+      "step": 6754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161553103218081e-06,
+      "loss": 0.9869,
+      "step": 6755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158856660760054e-06,
+      "loss": 0.9638,
+      "step": 6756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156160356857807e-06,
+      "loss": 0.8902,
+      "step": 6757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153464191714258e-06,
+      "loss": 0.9212,
+      "step": 6758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150768165532312e-06,
+      "loss": 0.9154,
+      "step": 6759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148072278514849e-06,
+      "loss": 0.9161,
+      "step": 6760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145376530864751e-06,
+      "loss": 0.9276,
+      "step": 6761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142680922784897e-06,
+      "loss": 0.9443,
+      "step": 6762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139985454478134e-06,
+      "loss": 0.8988,
+      "step": 6763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137290126147316e-06,
+      "loss": 0.8937,
+      "step": 6764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134594937995279e-06,
+      "loss": 0.9361,
+      "step": 6765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131899890224848e-06,
+      "loss": 0.9022,
+      "step": 6766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129204983038847e-06,
+      "loss": 0.924,
+      "step": 6767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126510216640074e-06,
+      "loss": 0.9091,
+      "step": 6768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123815591231328e-06,
+      "loss": 0.9421,
+      "step": 6769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121121107015398e-06,
+      "loss": 0.946,
+      "step": 6770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118426764195049e-06,
+      "loss": 0.909,
+      "step": 6771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115732562973054e-06,
+      "loss": 0.9144,
+      "step": 6772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113038503552157e-06,
+      "loss": 0.9345,
+      "step": 6773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.110344586135103e-06,
+      "loss": 0.9058,
+      "step": 6774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10765081092463e-06,
+      "loss": 0.9525,
+      "step": 6775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104957178123451e-06,
+      "loss": 0.9185,
+      "step": 6776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102263687934282e-06,
+      "loss": 0.9032,
+      "step": 6777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099570340559816e-06,
+      "loss": 0.9242,
+      "step": 6778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096877136202746e-06,
+      "loss": 0.8863,
+      "step": 6779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094184075065751e-06,
+      "loss": 2.6803,
+      "step": 6780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091491157351494e-06,
+      "loss": 0.8836,
+      "step": 6781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088798383262631e-06,
+      "loss": 0.9769,
+      "step": 6782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086105753001816e-06,
+      "loss": 2.639,
+      "step": 6783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083413266771675e-06,
+      "loss": 2.7164,
+      "step": 6784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080720924774836e-06,
+      "loss": 0.9308,
+      "step": 6785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078028727213907e-06,
+      "loss": 0.8967,
+      "step": 6786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075336674291495e-06,
+      "loss": 0.9469,
+      "step": 6787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07264476621019e-06,
+      "loss": 0.8979,
+      "step": 6788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069953003172572e-06,
+      "loss": 0.9436,
+      "step": 6789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06726138538121e-06,
+      "loss": 0.8989,
+      "step": 6790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064569913038666e-06,
+      "loss": 0.8856,
+      "step": 6791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06187858634748e-06,
+      "loss": 0.9449,
+      "step": 6792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0591874055102e-06,
+      "loss": 0.9477,
+      "step": 6793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056496370729339e-06,
+      "loss": 0.8902,
+      "step": 6794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053805482207418e-06,
+      "loss": 0.9427,
+      "step": 6795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051114740146942e-06,
+      "loss": 0.9293,
+      "step": 6796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0484241447504e-06,
+      "loss": 2.6546,
+      "step": 6797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045733696220274e-06,
+      "loss": 0.9045,
+      "step": 6798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043043394759041e-06,
+      "loss": 0.9302,
+      "step": 6799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04035324056915e-06,
+      "loss": 0.9534,
+      "step": 6800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037663233853059e-06,
+      "loss": 0.9196,
+      "step": 6801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034973374813195e-06,
+      "loss": 0.9682,
+      "step": 6802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032283663651992e-06,
+      "loss": 0.9191,
+      "step": 6803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029594100571866e-06,
+      "loss": 0.9484,
+      "step": 6804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026904685775215e-06,
+      "loss": 0.9524,
+      "step": 6805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024215419464437e-06,
+      "loss": 0.9088,
+      "step": 6806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021526301841907e-06,
+      "loss": 0.9273,
+      "step": 6807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01883733311e-06,
+      "loss": 0.9612,
+      "step": 6808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016148513471077e-06,
+      "loss": 0.9624,
+      "step": 6809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013459843127482e-06,
+      "loss": 0.9155,
+      "step": 6810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01077132228155e-06,
+      "loss": 0.9812,
+      "step": 6811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008082951135617e-06,
+      "loss": 0.9399,
+      "step": 6812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005394729891982e-06,
+      "loss": 0.9085,
+      "step": 6813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002706658752961e-06,
+      "loss": 0.9375,
+      "step": 6814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000018737920837e-06,
+      "loss": 0.9531,
+      "step": 6815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997330967597893e-06,
+      "loss": 0.9236,
+      "step": 6816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9946433479864e-06,
+      "loss": 0.8905,
+      "step": 6817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991955879288613e-06,
+      "loss": 0.9406,
+      "step": 6818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989268561706778e-06,
+      "loss": 0.9071,
+      "step": 6819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986581395443134e-06,
+      "loss": 0.9435,
+      "step": 6820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983894380699899e-06,
+      "loss": 0.8856,
+      "step": 6821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981207517679293e-06,
+      "loss": 0.9168,
+      "step": 6822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978520806583503e-06,
+      "loss": 0.8952,
+      "step": 6823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97583424761473e-06,
+      "loss": 0.9289,
+      "step": 6824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973147840975152e-06,
+      "loss": 0.8913,
+      "step": 6825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970461586866926e-06,
+      "loss": 0.9004,
+      "step": 6826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967775485492213e-06,
+      "loss": 0.8893,
+      "step": 6827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96508953705316e-06,
+      "loss": 0.903,
+      "step": 6828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962403741751893e-06,
+      "loss": 0.9039,
+      "step": 6829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959718099790534e-06,
+      "loss": 0.9286,
+      "step": 6830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957032611371191e-06,
+      "loss": 0.9379,
+      "step": 6831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954347276695961e-06,
+      "loss": 0.9376,
+      "step": 6832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951662095966934e-06,
+      "loss": 0.9487,
+      "step": 6833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948977069386176e-06,
+      "loss": 0.9418,
+      "step": 6834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946292197155757e-06,
+      "loss": 0.9396,
+      "step": 6835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943607479477728e-06,
+      "loss": 0.9329,
+      "step": 6836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940922916554122e-06,
+      "loss": 0.9248,
+      "step": 6837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938238508586971e-06,
+      "loss": 0.9187,
+      "step": 6838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935554255778288e-06,
+      "loss": 0.9188,
+      "step": 6839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93287015833008e-06,
+      "loss": 0.9379,
+      "step": 6840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93018621644434e-06,
+      "loss": 0.8998,
+      "step": 6841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927502430323044e-06,
+      "loss": 0.9017,
+      "step": 6842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924818800168168e-06,
+      "loss": 0.9048,
+      "step": 6843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922135326181664e-06,
+      "loss": 0.8913,
+      "step": 6844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919452008565478e-06,
+      "loss": 2.7535,
+      "step": 6845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91676884752155e-06,
+      "loss": 0.9192,
+      "step": 6846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914085843251796e-06,
+      "loss": 0.8814,
+      "step": 6847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911402995958121e-06,
+      "loss": 0.9463,
+      "step": 6848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90872030584244e-06,
+      "loss": 0.9131,
+      "step": 6849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906037773106625e-06,
+      "loss": 0.9364,
+      "step": 6850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903355397952557e-06,
+      "loss": 0.9423,
+      "step": 6851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900673180582096e-06,
+      "loss": 0.9426,
+      "step": 6852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897991121197094e-06,
+      "loss": 0.9222,
+      "step": 6853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895309219999393e-06,
+      "loss": 0.9228,
+      "step": 6854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892627477190815e-06,
+      "loss": 0.9163,
+      "step": 6855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88994589297318e-06,
+      "loss": 0.9185,
+      "step": 6856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887264467548295e-06,
+      "loss": 0.9611,
+      "step": 6857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884583201117939e-06,
+      "loss": 0.9149,
+      "step": 6858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881902093883904e-06,
+      "loss": 2.7152,
+      "step": 6859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879221146047948e-06,
+      "loss": 0.954,
+      "step": 6860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876540357811832e-06,
+      "loss": 0.9235,
+      "step": 6861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8738597293773e-06,
+      "loss": 0.9548,
+      "step": 6862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87117926094608e-06,
+      "loss": 0.8839,
+      "step": 6863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868498952719893e-06,
+      "loss": 0.9273,
+      "step": 6864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865818804900449e-06,
+      "loss": 0.9227,
+      "step": 6865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863138817689439e-06,
+      "loss": 0.8943,
+      "step": 6866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860458991288552e-06,
+      "loss": 0.9621,
+      "step": 6867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857779325899449e-06,
+      "loss": 0.9503,
+      "step": 6868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855099821723797e-06,
+      "loss": 0.9021,
+      "step": 6869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852420478963246e-06,
+      "loss": 0.9372,
+      "step": 6870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849741297819423e-06,
+      "loss": 0.908,
+      "step": 6871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847062278493951e-06,
+      "loss": 0.9248,
+      "step": 6872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844383421188447e-06,
+      "loss": 0.9097,
+      "step": 6873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8417047261045e-06,
+      "loss": 0.9044,
+      "step": 6874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839026193443706e-06,
+      "loss": 0.9228,
+      "step": 6875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836347823407631e-06,
+      "loss": 0.9122,
+      "step": 6876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833669616197839e-06,
+      "loss": 0.9243,
+      "step": 6877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830991572015885e-06,
+      "loss": 0.8751,
+      "step": 6878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828313691063294e-06,
+      "loss": 0.9031,
+      "step": 6879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825635973541601e-06,
+      "loss": 0.9525,
+      "step": 6880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822958419652314e-06,
+      "loss": 0.9173,
+      "step": 6881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820281029596932e-06,
+      "loss": 0.9357,
+      "step": 6882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817603803576948e-06,
+      "loss": 0.8814,
+      "step": 6883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814926741793831e-06,
+      "loss": 0.9428,
+      "step": 6884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812249844449048e-06,
+      "loss": 0.9516,
+      "step": 6885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80957311174405e-06,
+      "loss": 2.6498,
+      "step": 6886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806896543880274e-06,
+      "loss": 0.962,
+      "step": 6887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804220141059148e-06,
+      "loss": 0.887,
+      "step": 6888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801543903482078e-06,
+      "loss": 0.8724,
+      "step": 6889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798867831350473e-06,
+      "loss": 0.9339,
+      "step": 6890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796191924865724e-06,
+      "loss": 2.6916,
+      "step": 6891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793516184229198e-06,
+      "loss": 0.9195,
+      "step": 6892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790840609642261e-06,
+      "loss": 0.8575,
+      "step": 6893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788165201306272e-06,
+      "loss": 0.9031,
+      "step": 6894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78548995942256e-06,
+      "loss": 0.9188,
+      "step": 6895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782814884192457e-06,
+      "loss": 0.9439,
+      "step": 6896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780139975817272e-06,
+      "loss": 0.8873,
+      "step": 6897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777465234498307e-06,
+      "loss": 0.9574,
+      "step": 6898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.9219,
+      "step": 6899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77211625383419e-06,
+      "loss": 0.9143,
+      "step": 6900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769442014891571e-06,
+      "loss": 0.936,
+      "step": 6901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766767943810258e-06,
+      "loss": 0.9565,
+      "step": 6902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764094040791476e-06,
+      "loss": 0.906,
+      "step": 6903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76142030603646e-06,
+      "loss": 0.9246,
+      "step": 6904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75874673974642e-06,
+      "loss": 0.882,
+      "step": 6905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756073342122552e-06,
+      "loss": 0.9481,
+      "step": 6906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753400113366052e-06,
+      "loss": 0.9137,
+      "step": 6907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750727053678086e-06,
+      "loss": 2.6665,
+      "step": 6908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748054163259824e-06,
+      "loss": 0.8832,
+      "step": 6909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745381442312406e-06,
+      "loss": 0.92,
+      "step": 6910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742708891036974e-06,
+      "loss": 0.9475,
+      "step": 6911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740036509634656e-06,
+      "loss": 0.9327,
+      "step": 6912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737364298306554e-06,
+      "loss": 0.956,
+      "step": 6913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734692257253766e-06,
+      "loss": 0.9069,
+      "step": 6914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732020386677389e-06,
+      "loss": 0.9353,
+      "step": 6915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729348686778483e-06,
+      "loss": 0.9093,
+      "step": 6916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726677157758113e-06,
+      "loss": 0.8952,
+      "step": 6917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724005799817323e-06,
+      "loss": 0.9413,
+      "step": 6918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721334613157149e-06,
+      "loss": 0.9591,
+      "step": 6919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718663597978613e-06,
+      "loss": 0.9341,
+      "step": 6920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715992754482719e-06,
+      "loss": 0.9357,
+      "step": 6921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713322082870466e-06,
+      "loss": 0.9731,
+      "step": 6922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710651583342839e-06,
+      "loss": 2.6233,
+      "step": 6923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707981256100796e-06,
+      "loss": 0.9515,
+      "step": 6924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705311101345306e-06,
+      "loss": 0.9026,
+      "step": 6925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702641119277303e-06,
+      "loss": 0.8876,
+      "step": 6926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69997131009772e-06,
+      "loss": 0.9407,
+      "step": 6927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697301674007478e-06,
+      "loss": 0.9122,
+      "step": 6928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694632211207477e-06,
+      "loss": 0.9401,
+      "step": 6929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691962921898608e-06,
+      "loss": 0.8389,
+      "step": 6930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689293806281755e-06,
+      "loss": 0.9395,
+      "step": 6931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686624864557775e-06,
+      "loss": 0.8946,
+      "step": 6932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683956096927528e-06,
+      "loss": 0.9489,
+      "step": 6933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681287503591843e-06,
+      "loss": 0.9297,
+      "step": 6934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678619084751554e-06,
+      "loss": 0.8976,
+      "step": 6935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675950840607474e-06,
+      "loss": 0.9371,
+      "step": 6936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673282771360398e-06,
+      "loss": 0.9124,
+      "step": 6937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67061487721111e-06,
+      "loss": 0.9175,
+      "step": 6938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667947158360394e-06,
+      "loss": 0.916,
+      "step": 6939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665279615008998e-06,
+      "loss": 0.9366,
+      "step": 6940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662612247357678e-06,
+      "loss": 0.9175,
+      "step": 6941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659945055607158e-06,
+      "loss": 0.9111,
+      "step": 6942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657278039958166e-06,
+      "loss": 0.9098,
+      "step": 6943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654611200611413e-06,
+      "loss": 0.8972,
+      "step": 6944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651944537767577e-06,
+      "loss": 0.9005,
+      "step": 6945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649278051627356e-06,
+      "loss": 0.9827,
+      "step": 6946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646611742391404e-06,
+      "loss": 0.9432,
+      "step": 6947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643945610260382e-06,
+      "loss": 0.9204,
+      "step": 6948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641279655434928e-06,
+      "loss": 0.9233,
+      "step": 6949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63861387811567e-06,
+      "loss": 0.9062,
+      "step": 6950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63594827850322e-06,
+      "loss": 0.9283,
+      "step": 6951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633282856798186e-06,
+      "loss": 0.9488,
+      "step": 6952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630617613201143e-06,
+      "loss": 0.8984,
+      "step": 6953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627952547912678e-06,
+      "loss": 0.9578,
+      "step": 6954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625287661133337e-06,
+      "loss": 0.9084,
+      "step": 6955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622622953063677e-06,
+      "loss": 0.9307,
+      "step": 6956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619958423904233e-06,
+      "loss": 0.891,
+      "step": 6957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617294073855514e-06,
+      "loss": 0.8744,
+      "step": 6958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614629903118034e-06,
+      "loss": 0.921,
+      "step": 6959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61196591189229e-06,
+      "loss": 0.9532,
+      "step": 6960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6093021003787505e-06,
+      "loss": 0.9345,
+      "step": 6961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606638468777892e-06,
+      "loss": 0.9598,
+      "step": 6962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603975017290159e-06,
+      "loss": 0.9075,
+      "step": 6963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601311746115992e-06,
+      "loss": 0.8665,
+      "step": 6964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598648655455821e-06,
+      "loss": 0.8802,
+      "step": 6965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595985745510053e-06,
+      "loss": 0.8901,
+      "step": 6966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593323016479087e-06,
+      "loss": 0.9218,
+      "step": 6967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590660468563313e-06,
+      "loss": 0.9501,
+      "step": 6968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587998101963091e-06,
+      "loss": 0.9244,
+      "step": 6969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585335916878788e-06,
+      "loss": 0.9052,
+      "step": 6970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58267391351074e-06,
+      "loss": 0.8869,
+      "step": 6971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.58001209205928e-06,
+      "loss": 2.6734,
+      "step": 6972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577350452724728e-06,
+      "loss": 0.8955,
+      "step": 6973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57468899570738e-06,
+      "loss": 0.9294,
+      "step": 6974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572027721207528e-06,
+      "loss": 0.8848,
+      "step": 6975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569366629425448e-06,
+      "loss": 0.9144,
+      "step": 6976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566705720561399e-06,
+      "loss": 0.9271,
+      "step": 6977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564044994815634e-06,
+      "loss": 0.9055,
+      "step": 6978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5613844523883765e-06,
+      "loss": 0.953,
+      "step": 6979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55872409347985e-06,
+      "loss": 0.9103,
+      "step": 6980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55606391829027e-06,
+      "loss": 0.9319,
+      "step": 6981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553403927019817e-06,
+      "loss": 0.9087,
+      "step": 6982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550744119868675e-06,
+      "loss": 0.9076,
+      "step": 6983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548084497037006e-06,
+      "loss": 0.9413,
+      "step": 6984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545425058724963e-06,
+      "loss": 0.9519,
+      "step": 6985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542765805132682e-06,
+      "loss": 0.9481,
+      "step": 6986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.540106736460285e-06,
+      "loss": 0.9317,
+      "step": 6987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537447852907882e-06,
+      "loss": 0.9503,
+      "step": 6988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534789154675571e-06,
+      "loss": 0.9231,
+      "step": 6989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5321306419634245e-06,
+      "loss": 0.9139,
+      "step": 6990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529472314971522e-06,
+      "loss": 0.8613,
+      "step": 6991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.526814173899905e-06,
+      "loss": 0.9503,
+      "step": 6992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524156218948617e-06,
+      "loss": 0.927,
+      "step": 6993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5214984503176865e-06,
+      "loss": 0.9468,
+      "step": 6994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51884086820712e-06,
+      "loss": 0.9514,
+      "step": 6995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516183472816916e-06,
+      "loss": 0.9631,
+      "step": 6996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5135262643470604e-06,
+      "loss": 0.9073,
+      "step": 6997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510869242997518e-06,
+      "loss": 0.9464,
+      "step": 6998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508212408968251e-06,
+      "loss": 0.9108,
+      "step": 6999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505555762459187e-06,
+      "loss": 0.9497,
+      "step": 7000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5028993036702626e-06,
+      "loss": 0.9379,
+      "step": 7001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500243032801393e-06,
+      "loss": 0.9395,
+      "step": 7002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497586950052468e-06,
+      "loss": 0.9045,
+      "step": 7003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494931055623375e-06,
+      "loss": 2.5619,
+      "step": 7004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492275349713989e-06,
+      "loss": 0.9016,
+      "step": 7005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489619832524158e-06,
+      "loss": 0.9675,
+      "step": 7006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486964504253731e-06,
+      "loss": 0.8803,
+      "step": 7007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484309365102529e-06,
+      "loss": 0.9542,
+      "step": 7008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48165441527037e-06,
+      "loss": 0.9077,
+      "step": 7009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478999654957055e-06,
+      "loss": 0.9509,
+      "step": 7010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476345084362358e-06,
+      "loss": 0.9238,
+      "step": 7011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473690703686059e-06,
+      "loss": 0.9244,
+      "step": 7012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471036513127918e-06,
+      "loss": 0.9397,
+      "step": 7013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4683825128876656e-06,
+      "loss": 0.9432,
+      "step": 7014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465728703165036e-06,
+      "loss": 0.9592,
+      "step": 7015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46307508415974e-06,
+      "loss": 0.9089,
+      "step": 7016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460421656071477e-06,
+      "loss": 0.9478,
+      "step": 7017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457768419099935e-06,
+      "loss": 0.9397,
+      "step": 7018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455115373444779e-06,
+      "loss": 0.9755,
+      "step": 7019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45246251930567e-06,
+      "loss": 0.9433,
+      "step": 7020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449809856882241e-06,
+      "loss": 0.9212,
+      "step": 7021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447157386374127e-06,
+      "loss": 0.9208,
+      "step": 7022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444505107980939e-06,
+      "loss": 0.9597,
+      "step": 7023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441853021902273e-06,
+      "loss": 0.9421,
+      "step": 7024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4392011283377075e-06,
+      "loss": 0.9018,
+      "step": 7025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436549427486825e-06,
+      "loss": 0.943,
+      "step": 7026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433897919549169e-06,
+      "loss": 0.9078,
+      "step": 7027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4312466047242845e-06,
+      "loss": 0.9096,
+      "step": 7028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428595483211692e-06,
+      "loss": 0.945,
+      "step": 7029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.425944555210906e-06,
+      "loss": 0.9628,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423293820921426e-06,
+      "loss": 0.906,
+      "step": 7031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420643280542726e-06,
+      "loss": 0.892,
+      "step": 7032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4179929342742785e-06,
+      "loss": 0.9229,
+      "step": 7033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41534278231554e-06,
+      "loss": 0.9805,
+      "step": 7034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.412692824865939e-06,
+      "loss": 0.9005,
+      "step": 7035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410043062124904e-06,
+      "loss": 0.9106,
+      "step": 7036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4073934942918426e-06,
+      "loss": 0.9731,
+      "step": 7037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404744121566148e-06,
+      "loss": 0.8915,
+      "step": 7038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402094944147204e-06,
+      "loss": 0.8693,
+      "step": 7039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399445962234369e-06,
+      "loss": 0.9407,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3967971760269975e-06,
+      "loss": 0.9083,
+      "step": 7041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394148585724424e-06,
+      "loss": 0.9407,
+      "step": 7042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391500191525966e-06,
+      "loss": 0.9224,
+      "step": 7043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388851993630936e-06,
+      "loss": 0.8776,
+      "step": 7044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386203992238614e-06,
+      "loss": 0.872,
+      "step": 7045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383556187548282e-06,
+      "loss": 0.9441,
+      "step": 7046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380908579759207e-06,
+      "loss": 0.9076,
+      "step": 7047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3782611690706275e-06,
+      "loss": 2.6738,
+      "step": 7048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3756139556817794e-06,
+      "loss": 0.9672,
+      "step": 7049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372966939791875e-06,
+      "loss": 0.9434,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370320121600119e-06,
+      "loss": 0.901,
+      "step": 7051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367673501305701e-06,
+      "loss": 0.884,
+      "step": 7052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3650270791077895e-06,
+      "loss": 0.8962,
+      "step": 7053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362380855205542e-06,
+      "loss": 0.9603,
+      "step": 7054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3597348297981075e-06,
+      "loss": 0.9094,
+      "step": 7055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357089003084601e-06,
+      "loss": 0.9266,
+      "step": 7056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354443375264149e-06,
+      "loss": 0.9414,
+      "step": 7057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351797946535837e-06,
+      "loss": 0.9105,
+      "step": 7058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349152717098751e-06,
+      "loss": 0.9313,
+      "step": 7059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346507687151965e-06,
+      "loss": 0.889,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.343862856894521e-06,
+      "loss": 0.938,
+      "step": 7061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341218226525464e-06,
+      "loss": 0.9166,
+      "step": 7062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3385737962438165e-06,
+      "loss": 0.9398,
+      "step": 7063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335929566248581e-06,
+      "loss": 0.9684,
+      "step": 7064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3332855367387565e-06,
+      "loss": 0.9088,
+      "step": 7065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330641707913311e-06,
+      "loss": 0.8986,
+      "step": 7066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.327998079971213e-06,
+      "loss": 0.9104,
+      "step": 7067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3253546531114136e-06,
+      "loss": 0.9094,
+      "step": 7068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322711427532835e-06,
+      "loss": 0.9223,
+      "step": 7069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320068403434399e-06,
+      "loss": 0.9234,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317425581015008e-06,
+      "loss": 0.8907,
+      "step": 7071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314782960473543e-06,
+      "loss": 0.9031,
+      "step": 7072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312140542008883e-06,
+      "loss": 0.9403,
+      "step": 7073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309498325819879e-06,
+      "loss": 0.8945,
+      "step": 7074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30685631210537e-06,
+      "loss": 0.9225,
+      "step": 7075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304214501064191e-06,
+      "loss": 0.9437,
+      "step": 7076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.301572892895139e-06,
+      "loss": 0.9332,
+      "step": 7077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2989314877970174e-06,
+      "loss": 0.9009,
+      "step": 7078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296290285968608e-06,
+      "loss": 0.9235,
+      "step": 7079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293649287608668e-06,
+      "loss": 0.9247,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291008492915952e-06,
+      "loss": 0.9586,
+      "step": 7081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2883679020891885e-06,
+      "loss": 2.591,
+      "step": 7082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2857275153271e-06,
+      "loss": 0.8768,
+      "step": 7083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2830873328283914e-06,
+      "loss": 0.9435,
+      "step": 7084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280447354791744e-06,
+      "loss": 0.9018,
+      "step": 7085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.277807581415839e-06,
+      "loss": 0.8798,
+      "step": 7086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275168012899323e-06,
+      "loss": 0.951,
+      "step": 7087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272528649440843e-06,
+      "loss": 0.9531,
+      "step": 7088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269889491239031e-06,
+      "loss": 0.9089,
+      "step": 7089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267250538492485e-06,
+      "loss": 0.8661,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264611791399805e-06,
+      "loss": 0.9374,
+      "step": 7091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2619732501595796e-06,
+      "loss": 0.9049,
+      "step": 7092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25933491497036e-06,
+      "loss": 0.9087,
+      "step": 7093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2566967860307036e-06,
+      "loss": 0.9897,
+      "step": 7094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.254058863539138e-06,
+      "loss": 0.929,
+      "step": 7095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2514211476941845e-06,
+      "loss": 0.956,
+      "step": 7096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248783638694345e-06,
+      "loss": 0.922,
+      "step": 7097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246146336738102e-06,
+      "loss": 0.8673,
+      "step": 7098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243509242023932e-06,
+      "loss": 0.9496,
+      "step": 7099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24087235475029e-06,
+      "loss": 0.9107,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23823567511561e-06,
+      "loss": 0.9134,
+      "step": 7101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235599203318322e-06,
+      "loss": 0.9442,
+      "step": 7102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232962939556831e-06,
+      "loss": 0.9765,
+      "step": 7103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230326884029531e-06,
+      "loss": 0.8726,
+      "step": 7104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227691036934799e-06,
+      "loss": 0.9545,
+      "step": 7105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225055398470995e-06,
+      "loss": 0.9288,
+      "step": 7106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222419968836467e-06,
+      "loss": 0.9151,
+      "step": 7107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219784748229547e-06,
+      "loss": 0.9237,
+      "step": 7108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217149736848543e-06,
+      "loss": 0.905,
+      "step": 7109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214514934891763e-06,
+      "loss": 2.652,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211880342557479e-06,
+      "loss": 2.7288,
+      "step": 7111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20924596004396e-06,
+      "loss": 0.9027,
+      "step": 7112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206611787549469e-06,
+      "loss": 0.9272,
+      "step": 7113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203977825272226e-06,
+      "loss": 0.9557,
+      "step": 7114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.201344073410459e-06,
+      "loss": 0.9062,
+      "step": 7115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198710532162372e-06,
+      "loss": 0.9166,
+      "step": 7116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196077201726147e-06,
+      "loss": 0.9397,
+      "step": 7117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193444082299964e-06,
+      "loss": 0.9267,
+      "step": 7118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.190811174081973e-06,
+      "loss": 2.7234,
+      "step": 7119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188178477270316e-06,
+      "loss": 0.9656,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185545992063121e-06,
+      "loss": 0.9052,
+      "step": 7121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182913718658489e-06,
+      "loss": 0.8821,
+      "step": 7122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.180281657254522e-06,
+      "loss": 0.9275,
+      "step": 7123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177649808049287e-06,
+      "loss": 0.8855,
+      "step": 7124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17501817124085e-06,
+      "loss": 0.9482,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172386747027257e-06,
+      "loss": 0.9003,
+      "step": 7126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16975553560653e-06,
+      "loss": 0.893,
+      "step": 7127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1671245371766885e-06,
+      "loss": 0.933,
+      "step": 7128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164493751935727e-06,
+      "loss": 0.8992,
+      "step": 7129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161863180081623e-06,
+      "loss": 0.9427,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159232821812348e-06,
+      "loss": 2.6771,
+      "step": 7131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1566026773258405e-06,
+      "loss": 0.946,
+      "step": 7132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153972746820039e-06,
+      "loss": 0.8755,
+      "step": 7133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151343030492863e-06,
+      "loss": 0.9518,
+      "step": 7134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148713528542204e-06,
+      "loss": 0.9148,
+      "step": 7135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146084241165949e-06,
+      "loss": 0.9319,
+      "step": 7136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143455168561971e-06,
+      "loss": 0.8893,
+      "step": 7137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140826310928115e-06,
+      "loss": 0.9157,
+      "step": 7138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138197668462221e-06,
+      "loss": 0.8827,
+      "step": 7139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135569241362106e-06,
+      "loss": 0.919,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132941029825572e-06,
+      "loss": 0.9305,
+      "step": 7141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130313034050411e-06,
+      "loss": 0.9034,
+      "step": 7142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127685254234385e-06,
+      "loss": 0.9185,
+      "step": 7143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.125057690575254e-06,
+      "loss": 0.9115,
+      "step": 7144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122430343270761e-06,
+      "loss": 0.9297,
+      "step": 7145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1198032125186165e-06,
+      "loss": 0.8709,
+      "step": 7146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117176298516534e-06,
+      "loss": 0.9549,
+      "step": 7147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114549601462199e-06,
+      "loss": 0.9123,
+      "step": 7148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111923121553284e-06,
+      "loss": 0.9272,
+      "step": 7149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109296858987451e-06,
+      "loss": 0.9434,
+      "step": 7150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106670813962333e-06,
+      "loss": 0.9314,
+      "step": 7151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1040449866755626e-06,
+      "loss": 0.9207,
+      "step": 7152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101419377324734e-06,
+      "loss": 0.9377,
+      "step": 7153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098793986107451e-06,
+      "loss": 0.9183,
+      "step": 7154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096168813221285e-06,
+      "loss": 0.9382,
+      "step": 7155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093543858863788e-06,
+      "loss": 0.9225,
+      "step": 7156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090919123232505e-06,
+      "loss": 0.9055,
+      "step": 7157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088294606524967e-06,
+      "loss": 2.4944,
+      "step": 7158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085670308938674e-06,
+      "loss": 0.9184,
+      "step": 7159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083046230671127e-06,
+      "loss": 0.8842,
+      "step": 7160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080422371919793e-06,
+      "loss": 0.9269,
+      "step": 7161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077798732882135e-06,
+      "loss": 0.9306,
+      "step": 7162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0751753137556e-06,
+      "loss": 0.9529,
+      "step": 7163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072552114737606e-06,
+      "loss": 0.9631,
+      "step": 7164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069929136025568e-06,
+      "loss": 0.9623,
+      "step": 7165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067306377816883e-06,
+      "loss": 0.9234,
+      "step": 7166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0646838403089164e-06,
+      "loss": 0.951,
+      "step": 7167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0620615236990385e-06,
+      "loss": 0.9195,
+      "step": 7168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0594394281845846e-06,
+      "loss": 0.9641,
+      "step": 7169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0568175539628846e-06,
+      "loss": 0.9555,
+      "step": 7170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0541959012312514e-06,
+      "loss": 0.9319,
+      "step": 7171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051574470186973e-06,
+      "loss": 0.9053,
+      "step": 7172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048953261027328e-06,
+      "loss": 0.9597,
+      "step": 7173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046332273949578e-06,
+      "loss": 0.8957,
+      "step": 7174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043711509150964e-06,
+      "loss": 0.9481,
+      "step": 7175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041090966828717e-06,
+      "loss": 0.9211,
+      "step": 7176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03847064718004e-06,
+      "loss": 0.9462,
+      "step": 7177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035850550402126e-06,
+      "loss": 0.8912,
+      "step": 7178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033230676692159e-06,
+      "loss": 0.9016,
+      "step": 7179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03061102624729e-06,
+      "loss": 0.9034,
+      "step": 7180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027991599264666e-06,
+      "loss": 0.9414,
+      "step": 7181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025372395941412e-06,
+      "loss": 0.9655,
+      "step": 7182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022753416474634e-06,
+      "loss": 0.9161,
+      "step": 7183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02013466106143e-06,
+      "loss": 2.5705,
+      "step": 7184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017516129898871e-06,
+      "loss": 0.9419,
+      "step": 7185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014897823184015e-06,
+      "loss": 0.9407,
+      "step": 7186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012279741113909e-06,
+      "loss": 0.9318,
+      "step": 7187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009661883885566e-06,
+      "loss": 0.9544,
+      "step": 7188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007044251696009e-06,
+      "loss": 0.9002,
+      "step": 7189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004426844742215e-06,
+      "loss": 0.8717,
+      "step": 7190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001809663221163e-06,
+      "loss": 0.9574,
+      "step": 7191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999192707329812e-06,
+      "loss": 2.7619,
+      "step": 7192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996575977265099e-06,
+      "loss": 0.9095,
+      "step": 7193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993959473223947e-06,
+      "loss": 0.8859,
+      "step": 7194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991343195403264e-06,
+      "loss": 0.8755,
+      "step": 7195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988727143999936e-06,
+      "loss": 0.9471,
+      "step": 7196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.98611131921084e-06,
+      "loss": 0.9238,
+      "step": 7197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983495721232819e-06,
+      "loss": 0.933,
+      "step": 7198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980880350262721e-06,
+      "loss": 0.9345,
+      "step": 7199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9782652064973675e-06,
+      "loss": 0.9168,
+      "step": 7200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975650290133554e-06,
+      "loss": 0.9553,
+      "step": 7201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.973035601368071e-06,
+      "loss": 0.8927,
+      "step": 7202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97042114039769e-06,
+      "loss": 0.9045,
+      "step": 7203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967806907419158e-06,
+      "loss": 0.9384,
+      "step": 7204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965192902629214e-06,
+      "loss": 0.9369,
+      "step": 7205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962579126224574e-06,
+      "loss": 0.8976,
+      "step": 7206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959965578401937e-06,
+      "loss": 0.8811,
+      "step": 7207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957352259357992e-06,
+      "loss": 0.8953,
+      "step": 7208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954739169289395e-06,
+      "loss": 0.894,
+      "step": 7209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952126308392804e-06,
+      "loss": 0.9151,
+      "step": 7210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9495136768648505e-06,
+      "loss": 0.95,
+      "step": 7211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946901274902142e-06,
+      "loss": 0.9286,
+      "step": 7212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9442891027012825e-06,
+      "loss": 0.9621,
+      "step": 7213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941677160458846e-06,
+      "loss": 0.905,
+      "step": 7214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939065448371398e-06,
+      "loss": 0.8948,
+      "step": 7215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9364539666354844e-06,
+      "loss": 0.9247,
+      "step": 7216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.933842715447632e-06,
+      "loss": 0.9421,
+      "step": 7217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9312316950043505e-06,
+      "loss": 0.8768,
+      "step": 7218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928620905502135e-06,
+      "loss": 0.9299,
+      "step": 7219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926010347137459e-06,
+      "loss": 0.9345,
+      "step": 7220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923400020106786e-06,
+      "loss": 0.9091,
+      "step": 7221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920789924606548e-06,
+      "loss": 0.8743,
+      "step": 7222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918180060833171e-06,
+      "loss": 0.8879,
+      "step": 7223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915570428983071e-06,
+      "loss": 0.9122,
+      "step": 7224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912961029252624e-06,
+      "loss": 0.9182,
+      "step": 7225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910351861838209e-06,
+      "loss": 0.9804,
+      "step": 7226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907742926936174e-06,
+      "loss": 0.9208,
+      "step": 7227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905134224742858e-06,
+      "loss": 0.9417,
+      "step": 7228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.902525755454582e-06,
+      "loss": 0.9168,
+      "step": 7229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899917519267642e-06,
+      "loss": 0.9215,
+      "step": 7230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897309516378326e-06,
+      "loss": 0.9193,
+      "step": 7231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894701746982902e-06,
+      "loss": 0.9255,
+      "step": 7232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8920942112776116e-06,
+      "loss": 0.9201,
+      "step": 7233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.889486909458689e-06,
+      "loss": 0.9201,
+      "step": 7234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886879841722347e-06,
+      "loss": 0.8896,
+      "step": 7235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88427300826478e-06,
+      "loss": 0.9308,
+      "step": 7236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8816664092821705e-06,
+      "loss": 0.9339,
+      "step": 7237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.879060044970675e-06,
+      "loss": 0.9007,
+      "step": 7238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876453915526435e-06,
+      "loss": 0.956,
+      "step": 7239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873848021145582e-06,
+      "loss": 0.9343,
+      "step": 7240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871242362024217e-06,
+      "loss": 0.9118,
+      "step": 7241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868636938358435e-06,
+      "loss": 0.918,
+      "step": 7242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866031750344302e-06,
+      "loss": 0.9543,
+      "step": 7243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863426798177871e-06,
+      "loss": 0.892,
+      "step": 7244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86082208205519e-06,
+      "loss": 0.8796,
+      "step": 7245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858217602172264e-06,
+      "loss": 0.8936,
+      "step": 7246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855613358725101e-06,
+      "loss": 0.9311,
+      "step": 7247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853009351909687e-06,
+      "loss": 2.5493,
+      "step": 7248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85040558192198e-06,
+      "loss": 0.9446,
+      "step": 7249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847802048957932e-06,
+      "loss": 0.895,
+      "step": 7250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84519875321347e-06,
+      "loss": 0.9045,
+      "step": 7251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842595694884508e-06,
+      "loss": 0.9223,
+      "step": 7252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839992874166943e-06,
+      "loss": 0.9357,
+      "step": 7253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83739029125664e-06,
+      "loss": 0.8964,
+      "step": 7254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834787946349472e-06,
+      "loss": 0.9515,
+      "step": 7255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832185839641266e-06,
+      "loss": 0.8939,
+      "step": 7256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829583971327851e-06,
+      "loss": 0.9464,
+      "step": 7257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826982341605033e-06,
+      "loss": 0.8501,
+      "step": 7258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824380950668592e-06,
+      "loss": 0.9286,
+      "step": 7259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821779798714301e-06,
+      "loss": 0.8932,
+      "step": 7260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819178885937911e-06,
+      "loss": 0.9489,
+      "step": 7261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816578212535153e-06,
+      "loss": 0.9438,
+      "step": 7262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813977778701745e-06,
+      "loss": 0.904,
+      "step": 7263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811377584633372e-06,
+      "loss": 0.9225,
+      "step": 7264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808777630525725e-06,
+      "loss": 0.9011,
+      "step": 7265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806177916574461e-06,
+      "loss": 0.9321,
+      "step": 7266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803578442975219e-06,
+      "loss": 0.8685,
+      "step": 7267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800979209923623e-06,
+      "loss": 2.5574,
+      "step": 7268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7983802176152844e-06,
+      "loss": 0.9107,
+      "step": 7269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795781466245786e-06,
+      "loss": 0.8625,
+      "step": 7270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7931829560107e-06,
+      "loss": 0.9504,
+      "step": 7271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790584687105577e-06,
+      "loss": 0.8819,
+      "step": 7272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7879866597259505e-06,
+      "loss": 0.9153,
+      "step": 7273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78538887406734e-06,
+      "loss": 0.8891,
+      "step": 7274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782791330325234e-06,
+      "loss": 0.8814,
+      "step": 7275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7801940286951166e-06,
+      "loss": 0.9235,
+      "step": 7276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777596969372453e-06,
+      "loss": 2.7405,
+      "step": 7277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.775000152552676e-06,
+      "loss": 0.9398,
+      "step": 7278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772403578431218e-06,
+      "loss": 0.8652,
+      "step": 7279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769807247203479e-06,
+      "loss": 0.9194,
+      "step": 7280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76721115906485e-06,
+      "loss": 0.9103,
+      "step": 7281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764615314210701e-06,
+      "loss": 0.9386,
+      "step": 7282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762019712836381e-06,
+      "loss": 0.9109,
+      "step": 7283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7594243551372214e-06,
+      "loss": 0.8721,
+      "step": 7284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756829241308543e-06,
+      "loss": 0.8925,
+      "step": 7285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.754234371545634e-06,
+      "loss": 0.8962,
+      "step": 7286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75163974604378e-06,
+      "loss": 0.8812,
+      "step": 7287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7490453649982325e-06,
+      "loss": 0.9036,
+      "step": 7288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746451228604234e-06,
+      "loss": 0.8755,
+      "step": 7289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7438573370570134e-06,
+      "loss": 0.9096,
+      "step": 7290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741263690551767e-06,
+      "loss": 0.9125,
+      "step": 7291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738670289283686e-06,
+      "loss": 0.9166,
+      "step": 7292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736077133447934e-06,
+      "loss": 0.9354,
+      "step": 7293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73348422323966e-06,
+      "loss": 0.9581,
+      "step": 7294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730891558853998e-06,
+      "loss": 0.8662,
+      "step": 7295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728299140486054e-06,
+      "loss": 0.8757,
+      "step": 7296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725706968330925e-06,
+      "loss": 0.8979,
+      "step": 7297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723115042583689e-06,
+      "loss": 0.9227,
+      "step": 7298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720523363439393e-06,
+      "loss": 0.9178,
+      "step": 7299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7179319310930825e-06,
+      "loss": 0.95,
+      "step": 7300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715340745739771e-06,
+      "loss": 0.8959,
+      "step": 7301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712749807574461e-06,
+      "loss": 0.9291,
+      "step": 7302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7101591167921385e-06,
+      "loss": 0.9016,
+      "step": 7303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707568673587759e-06,
+      "loss": 0.9182,
+      "step": 7304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704978478156272e-06,
+      "loss": 0.8949,
+      "step": 7305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702388530692605e-06,
+      "loss": 0.8985,
+      "step": 7306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69979883139166e-06,
+      "loss": 0.9294,
+      "step": 7307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.9239,
+      "step": 7308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694620178057485e-06,
+      "loss": 0.8848,
+      "step": 7309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692031224413968e-06,
+      "loss": 0.9355,
+      "step": 7310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689442519712625e-06,
+      "loss": 0.9234,
+      "step": 7311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686854064148258e-06,
+      "loss": 0.8952,
+      "step": 7312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.684265857915669e-06,
+      "loss": 0.8931,
+      "step": 7313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6816779012096314e-06,
+      "loss": 0.9238,
+      "step": 7314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6790901942249024e-06,
+      "loss": 0.8951,
+      "step": 7315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676502737156223e-06,
+      "loss": 0.87,
+      "step": 7316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673915530198308e-06,
+      "loss": 0.8904,
+      "step": 7317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671328573545863e-06,
+      "loss": 0.9085,
+      "step": 7318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6687418673935736e-06,
+      "loss": 0.8527,
+      "step": 7319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666155411936091e-06,
+      "loss": 0.9273,
+      "step": 7320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66356920736807e-06,
+      "loss": 0.9464,
+      "step": 7321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660983253884138e-06,
+      "loss": 0.9507,
+      "step": 7322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6583975516788924e-06,
+      "loss": 0.895,
+      "step": 7323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655812100946928e-06,
+      "loss": 0.9881,
+      "step": 7324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653226901882809e-06,
+      "loss": 0.8905,
+      "step": 7325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650641954681087e-06,
+      "loss": 0.8955,
+      "step": 7326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648057259536297e-06,
+      "loss": 0.9449,
+      "step": 7327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6454728166429435e-06,
+      "loss": 0.9178,
+      "step": 7328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6428886261955295e-06,
+      "loss": 0.904,
+      "step": 7329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640304688388517e-06,
+      "loss": 0.9072,
+      "step": 7330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637721003416368e-06,
+      "loss": 0.9151,
+      "step": 7331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635137571473523e-06,
+      "loss": 2.5758,
+      "step": 7332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632554392754389e-06,
+      "loss": 0.9603,
+      "step": 7333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6299714674533684e-06,
+      "loss": 0.9189,
+      "step": 7334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627388795764842e-06,
+      "loss": 0.9271,
+      "step": 7335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624806377883166e-06,
+      "loss": 0.896,
+      "step": 7336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622224214002684e-06,
+      "loss": 0.9129,
+      "step": 7337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619642304317714e-06,
+      "loss": 2.4968,
+      "step": 7338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6170606490225595e-06,
+      "loss": 0.9443,
+      "step": 7339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614479248311511e-06,
+      "loss": 0.9366,
+      "step": 7340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611898102378819e-06,
+      "loss": 0.9381,
+      "step": 7341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609317211418738e-06,
+      "loss": 2.7047,
+      "step": 7342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606736575625494e-06,
+      "loss": 0.901,
+      "step": 7343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604156195193288e-06,
+      "loss": 0.9266,
+      "step": 7344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601576070316312e-06,
+      "loss": 0.9348,
+      "step": 7345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59899620118873e-06,
+      "loss": 0.8821,
+      "step": 7346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596416588004694e-06,
+      "loss": 0.8615,
+      "step": 7347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593837230958333e-06,
+      "loss": 0.9236,
+      "step": 7348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591258130243756e-06,
+      "loss": 0.9582,
+      "step": 7349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588679286055055e-06,
+      "loss": 0.896,
+      "step": 7350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586100698586303e-06,
+      "loss": 0.9181,
+      "step": 7351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583522368031548e-06,
+      "loss": 0.9153,
+      "step": 7352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580944294584832e-06,
+      "loss": 0.9316,
+      "step": 7353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578366478440157e-06,
+      "loss": 0.9113,
+      "step": 7354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575788919791522e-06,
+      "loss": 0.9053,
+      "step": 7355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5732116188329074e-06,
+      "loss": 0.9567,
+      "step": 7356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.570634575758261e-06,
+      "loss": 0.9242,
+      "step": 7357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5680577907615235e-06,
+      "loss": 0.9465,
+      "step": 7358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56548126403661e-06,
+      "loss": 0.8948,
+      "step": 7359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.562904995777417e-06,
+      "loss": 0.9606,
+      "step": 7360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560328986177824e-06,
+      "loss": 0.9668,
+      "step": 7361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557753235431689e-06,
+      "loss": 0.9227,
+      "step": 7362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55517774373285e-06,
+      "loss": 0.9305,
+      "step": 7363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55260251127513e-06,
+      "loss": 0.8806,
+      "step": 7364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5500275382523215e-06,
+      "loss": 0.8596,
+      "step": 7365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547452824858213e-06,
+      "loss": 0.8768,
+      "step": 7366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544878371286558e-06,
+      "loss": 0.8662,
+      "step": 7367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542304177731102e-06,
+      "loss": 0.9422,
+      "step": 7368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.539730244385568e-06,
+      "loss": 0.9415,
+      "step": 7369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537156571443655e-06,
+      "loss": 0.8889,
+      "step": 7370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534583159099045e-06,
+      "loss": 0.8627,
+      "step": 7371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532010007545405e-06,
+      "loss": 0.8966,
+      "step": 7372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529437116976374e-06,
+      "loss": 0.9494,
+      "step": 7373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526864487585582e-06,
+      "loss": 0.9365,
+      "step": 7374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524292119566625e-06,
+      "loss": 0.8962,
+      "step": 7375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521720013113087e-06,
+      "loss": 0.8939,
+      "step": 7376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519148168418542e-06,
+      "loss": 2.6686,
+      "step": 7377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5165765856765266e-06,
+      "loss": 0.9073,
+      "step": 7378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.514005265080568e-06,
+      "loss": 0.8912,
+      "step": 7379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5114342068241756e-06,
+      "loss": 0.9683,
+      "step": 7380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508863411100828e-06,
+      "loss": 0.8884,
+      "step": 7381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506292878103999e-06,
+      "loss": 0.9389,
+      "step": 7382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503722608027129e-06,
+      "loss": 0.928,
+      "step": 7383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501152601063645e-06,
+      "loss": 0.8846,
+      "step": 7384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498582857406961e-06,
+      "loss": 0.9189,
+      "step": 7385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49601337725045e-06,
+      "loss": 0.9095,
+      "step": 7386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4934441607874895e-06,
+      "loss": 0.9316,
+      "step": 7387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490875208211428e-06,
+      "loss": 0.9447,
+      "step": 7388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488306519715585e-06,
+      "loss": 2.5618,
+      "step": 7389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485738095493272e-06,
+      "loss": 0.8811,
+      "step": 7390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483169935737777e-06,
+      "loss": 0.9422,
+      "step": 7391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480602040642365e-06,
+      "loss": 0.8911,
+      "step": 7392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.478034410400289e-06,
+      "loss": 0.9464,
+      "step": 7393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475467045204769e-06,
+      "loss": 0.9363,
+      "step": 7394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472899945249024e-06,
+      "loss": 0.9035,
+      "step": 7395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470333110726227e-06,
+      "loss": 0.9209,
+      "step": 7396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467766541829557e-06,
+      "loss": 0.9385,
+      "step": 7397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465200238752162e-06,
+      "loss": 0.9227,
+      "step": 7398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462634201687163e-06,
+      "loss": 0.9111,
+      "step": 7399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460068430827671e-06,
+      "loss": 0.9466,
+      "step": 7400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4575029263667765e-06,
+      "loss": 0.8867,
+      "step": 7401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454937688497545e-06,
+      "loss": 0.9199,
+      "step": 7402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4523727174130246e-06,
+      "loss": 0.9305,
+      "step": 7403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44980801330624e-06,
+      "loss": 0.8679,
+      "step": 7404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447243576370204e-06,
+      "loss": 0.9045,
+      "step": 7405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444679406797904e-06,
+      "loss": 0.9265,
+      "step": 7406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4421155047822985e-06,
+      "loss": 0.93,
+      "step": 7407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439551870516345e-06,
+      "loss": 2.6806,
+      "step": 7408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4369885041929695e-06,
+      "loss": 0.9263,
+      "step": 7409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434425406005072e-06,
+      "loss": 0.8485,
+      "step": 7410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.431862576145546e-06,
+      "loss": 0.9117,
+      "step": 7411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429300014807253e-06,
+      "loss": 0.9239,
+      "step": 7412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4267377221830405e-06,
+      "loss": 0.9353,
+      "step": 7413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42417569846574e-06,
+      "loss": 0.9412,
+      "step": 7414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421613943848148e-06,
+      "loss": 0.914,
+      "step": 7415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419052458523058e-06,
+      "loss": 0.9368,
+      "step": 7416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416491242683233e-06,
+      "loss": 0.9247,
+      "step": 7417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413930296521417e-06,
+      "loss": 0.9168,
+      "step": 7418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411369620230339e-06,
+      "loss": 0.9115,
+      "step": 7419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4088092140026945e-06,
+      "loss": 0.9527,
+      "step": 7420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406249078031172e-06,
+      "loss": 0.9299,
+      "step": 7421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403689212508441e-06,
+      "loss": 0.879,
+      "step": 7422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401129617627138e-06,
+      "loss": 0.9208,
+      "step": 7423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.398570293579887e-06,
+      "loss": 0.9347,
+      "step": 7424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396011240559294e-06,
+      "loss": 0.9086,
+      "step": 7425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393452458757936e-06,
+      "loss": 0.8778,
+      "step": 7426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390893948368382e-06,
+      "loss": 0.982,
+      "step": 7427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388335709583166e-06,
+      "loss": 0.9485,
+      "step": 7428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385777742594814e-06,
+      "loss": 0.9171,
+      "step": 7429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383220047595828e-06,
+      "loss": 0.8828,
+      "step": 7430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380662624778682e-06,
+      "loss": 0.9166,
+      "step": 7431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37810547433584e-06,
+      "loss": 0.8875,
+      "step": 7432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375548596459739e-06,
+      "loss": 0.9125,
+      "step": 7433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372991991342798e-06,
+      "loss": 2.4411,
+      "step": 7434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370435659177416e-06,
+      "loss": 0.9097,
+      "step": 7435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36787960015597e-06,
+      "loss": 0.8812,
+      "step": 7436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3653238144708165e-06,
+      "loss": 0.9679,
+      "step": 7437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362768302314295e-06,
+      "loss": 0.9226,
+      "step": 7438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3602130638787155e-06,
+      "loss": 0.9425,
+      "step": 7439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3576580993563805e-06,
+      "loss": 0.9491,
+      "step": 7440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355103408939558e-06,
+      "loss": 0.898,
+      "step": 7441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352548992820502e-06,
+      "loss": 0.9354,
+      "step": 7442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349994851191454e-06,
+      "loss": 2.7034,
+      "step": 7443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34744098424462e-06,
+      "loss": 0.9484,
+      "step": 7444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34488739217219e-06,
+      "loss": 0.9246,
+      "step": 7445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342334075166341e-06,
+      "loss": 0.922,
+      "step": 7446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339781033419219e-06,
+      "loss": 0.933,
+      "step": 7447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337228267122959e-06,
+      "loss": 0.9215,
+      "step": 7448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334675776469664e-06,
+      "loss": 0.8998,
+      "step": 7449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332123561651427e-06,
+      "loss": 0.9437,
+      "step": 7450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329571622860318e-06,
+      "loss": 0.8925,
+      "step": 7451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327019960288372e-06,
+      "loss": 0.9203,
+      "step": 7452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324468574127627e-06,
+      "loss": 0.9276,
+      "step": 7453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321917464570089e-06,
+      "loss": 0.927,
+      "step": 7454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3193666318077355e-06,
+      "loss": 0.8749,
+      "step": 7455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3168160760325325e-06,
+      "loss": 0.8906,
+      "step": 7456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31426579743642e-06,
+      "loss": 0.9207,
+      "step": 7457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3117157962113264e-06,
+      "loss": 0.9179,
+      "step": 7458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309166072549151e-06,
+      "loss": 0.9374,
+      "step": 7459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306616626641768e-06,
+      "loss": 0.9036,
+      "step": 7460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304067458681048e-06,
+      "loss": 0.9231,
+      "step": 7461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301518568858815e-06,
+      "loss": 2.5028,
+      "step": 7462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298969957366897e-06,
+      "loss": 0.9113,
+      "step": 7463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296421624397091e-06,
+      "loss": 0.9245,
+      "step": 7464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293873570141166e-06,
+      "loss": 0.9428,
+      "step": 7465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291325794790879e-06,
+      "loss": 0.8903,
+      "step": 7466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288778298537966e-06,
+      "loss": 0.8708,
+      "step": 7467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286231081574136e-06,
+      "loss": 0.9042,
+      "step": 7468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2836841440910854e-06,
+      "loss": 0.9135,
+      "step": 7469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281137486280479e-06,
+      "loss": 0.8883,
+      "step": 7470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278591108333969e-06,
+      "loss": 0.9541,
+      "step": 7471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276045010443188e-06,
+      "loss": 0.9067,
+      "step": 7472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2734991927997335e-06,
+      "loss": 0.8741,
+      "step": 7473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2709536555952e-06,
+      "loss": 0.9101,
+      "step": 7474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2684083990211534e-06,
+      "loss": 2.6116,
+      "step": 7475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26586342326913e-06,
+      "loss": 0.9166,
+      "step": 7476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263318728530661e-06,
+      "loss": 0.8947,
+      "step": 7477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26077431499724e-06,
+      "loss": 0.8964,
+      "step": 7478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2582301828603535e-06,
+      "loss": 0.9419,
+      "step": 7479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25568633231146e-06,
+      "loss": 0.917,
+      "step": 7480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.253142763541996e-06,
+      "loss": 0.9101,
+      "step": 7481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25059947674338e-06,
+      "loss": 0.9322,
+      "step": 7482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248056472107009e-06,
+      "loss": 2.7042,
+      "step": 7483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245513749824253e-06,
+      "loss": 0.9496,
+      "step": 7484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242971310086474e-06,
+      "loss": 0.9443,
+      "step": 7485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240429153084993e-06,
+      "loss": 0.9298,
+      "step": 7486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237887279011126e-06,
+      "loss": 0.8887,
+      "step": 7487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235345688056167e-06,
+      "loss": 0.9125,
+      "step": 7488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232804380411374e-06,
+      "loss": 0.9027,
+      "step": 7489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230263356268003e-06,
+      "loss": 0.9344,
+      "step": 7490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227722615817277e-06,
+      "loss": 0.9201,
+      "step": 7491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225182159250396e-06,
+      "loss": 0.9344,
+      "step": 7492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222641986758549e-06,
+      "loss": 0.9081,
+      "step": 7493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220102098532893e-06,
+      "loss": 0.9919,
+      "step": 7494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217562494764569e-06,
+      "loss": 0.9169,
+      "step": 7495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215023175644701e-06,
+      "loss": 0.8751,
+      "step": 7496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212484141364377e-06,
+      "loss": 2.6653,
+      "step": 7497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209945392114679e-06,
+      "loss": 0.9418,
+      "step": 7498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207406928086658e-06,
+      "loss": 0.8967,
+      "step": 7499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204868749471348e-06,
+      "loss": 0.9073,
+      "step": 7500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202330856459763e-06,
+      "loss": 0.9395,
+      "step": 7501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199793249242888e-06,
+      "loss": 0.9062,
+      "step": 7502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197255928011695e-06,
+      "loss": 0.898,
+      "step": 7503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194718892957131e-06,
+      "loss": 0.8835,
+      "step": 7504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192182144270119e-06,
+      "loss": 2.6151,
+      "step": 7505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1896456821415675e-06,
+      "loss": 0.9265,
+      "step": 7506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1871095067623525e-06,
+      "loss": 0.9176,
+      "step": 7507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184573618323333e-06,
+      "loss": 0.8896,
+      "step": 7508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18203801701536e-06,
+      "loss": 0.866,
+      "step": 7509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179502703029239e-06,
+      "loss": 0.9343,
+      "step": 7510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176967676555769e-06,
+      "loss": 0.9132,
+      "step": 7511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174432937785729e-06,
+      "loss": 0.932,
+      "step": 7512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171898486909865e-06,
+      "loss": 0.8866,
+      "step": 7513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169364324118913e-06,
+      "loss": 0.8838,
+      "step": 7514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166830449603577e-06,
+      "loss": 2.4857,
+      "step": 7515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164296863554547e-06,
+      "loss": 0.9384,
+      "step": 7516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161763566162494e-06,
+      "loss": 0.8852,
+      "step": 7517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159230557618051e-06,
+      "loss": 0.883,
+      "step": 7518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156697838111848e-06,
+      "loss": 0.9589,
+      "step": 7519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.154165407834488e-06,
+      "loss": 0.8859,
+      "step": 7520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1516332669765425e-06,
+      "loss": 0.9325,
+      "step": 7521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1491014157285736e-06,
+      "loss": 0.9184,
+      "step": 7522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146569854281111e-06,
+      "loss": 2.5229,
+      "step": 7523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144038582824673e-06,
+      "loss": 0.9031,
+      "step": 7524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141507601549752e-06,
+      "loss": 0.8918,
+      "step": 7525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1389769106468125e-06,
+      "loss": 0.8903,
+      "step": 7526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136446510306306e-06,
+      "loss": 0.9115,
+      "step": 7527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133916400718662e-06,
+      "loss": 0.8919,
+      "step": 7528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131386582074276e-06,
+      "loss": 0.9111,
+      "step": 7529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128857054563541e-06,
+      "loss": 0.9599,
+      "step": 7530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126327818376807e-06,
+      "loss": 0.9338,
+      "step": 7531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1237988737044165e-06,
+      "loss": 0.9311,
+      "step": 7532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121270220736688e-06,
+      "loss": 0.9153,
+      "step": 7533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118741859663911e-06,
+      "loss": 0.9118,
+      "step": 7534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.116213790676365e-06,
+      "loss": 2.6374,
+      "step": 7535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113686013964295e-06,
+      "loss": 0.8815,
+      "step": 7536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111158529717931e-06,
+      "loss": 0.9562,
+      "step": 7537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108631338127485e-06,
+      "loss": 0.9181,
+      "step": 7538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106104439383128e-06,
+      "loss": 0.9365,
+      "step": 7539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103577833675034e-06,
+      "loss": 0.9183,
+      "step": 7540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101051521193346e-06,
+      "loss": 0.896,
+      "step": 7541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098525502128172e-06,
+      "loss": 0.9395,
+      "step": 7542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095999776669614e-06,
+      "loss": 0.9173,
+      "step": 7543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093474345007746e-06,
+      "loss": 0.9109,
+      "step": 7544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090949207332618e-06,
+      "loss": 0.9248,
+      "step": 7545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088424363834264e-06,
+      "loss": 0.8854,
+      "step": 7546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085899814702687e-06,
+      "loss": 0.9353,
+      "step": 7547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083375560127875e-06,
+      "loss": 0.9081,
+      "step": 7548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.080851600299795e-06,
+      "loss": 0.914,
+      "step": 7549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078327935408381e-06,
+      "loss": 0.9307,
+      "step": 7550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075804565643562e-06,
+      "loss": 0.9325,
+      "step": 7551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073281491195224e-06,
+      "loss": 0.9045,
+      "step": 7552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070758712253244e-06,
+      "loss": 0.9615,
+      "step": 7553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068236229007485e-06,
+      "loss": 0.9014,
+      "step": 7554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065714041647765e-06,
+      "loss": 0.9434,
+      "step": 7555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063192150363895e-06,
+      "loss": 0.928,
+      "step": 7556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060670555345664e-06,
+      "loss": 0.9306,
+      "step": 7557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058149256782833e-06,
+      "loss": 0.941,
+      "step": 7558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055628254865145e-06,
+      "loss": 0.9079,
+      "step": 7559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053107549782315e-06,
+      "loss": 0.9246,
+      "step": 7560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050587141724043e-06,
+      "loss": 0.9276,
+      "step": 7561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048067030880005e-06,
+      "loss": 0.909,
+      "step": 7562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045547217439844e-06,
+      "loss": 0.8938,
+      "step": 7563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043027701593194e-06,
+      "loss": 0.9838,
+      "step": 7564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040508483529668e-06,
+      "loss": 0.8971,
+      "step": 7565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037989563438842e-06,
+      "loss": 0.8939,
+      "step": 7566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.035470941510282e-06,
+      "loss": 0.93,
+      "step": 7567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032952617933525e-06,
+      "loss": 2.7074,
+      "step": 7568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.030434592898091e-06,
+      "loss": 0.9497,
+      "step": 7569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027916866593475e-06,
+      "loss": 0.8973,
+      "step": 7570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025399439209146e-06,
+      "loss": 0.8936,
+      "step": 7571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02288231093456e-06,
+      "loss": 0.8979,
+      "step": 7572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0203654819591365e-06,
+      "loss": 0.9316,
+      "step": 7573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017848952472281e-06,
+      "loss": 0.9178,
+      "step": 7574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015332722663384e-06,
+      "loss": 0.8876,
+      "step": 7575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012816792721796e-06,
+      "loss": 0.931,
+      "step": 7576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010301162836857e-06,
+      "loss": 0.9269,
+      "step": 7577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007785833197883e-06,
+      "loss": 0.9233,
+      "step": 7578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005270803994165e-06,
+      "loss": 0.9052,
+      "step": 7579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002756075414973e-06,
+      "loss": 0.9078,
+      "step": 7580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00024164764955e-06,
+      "loss": 0.9531,
+      "step": 7581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.997727520887125e-06,
+      "loss": 0.9333,
+      "step": 7582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9952136953169e-06,
+      "loss": 0.8666,
+      "step": 7583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992700171128043e-06,
+      "loss": 0.9293,
+      "step": 7584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.990186948509722e-06,
+      "loss": 2.4643,
+      "step": 7585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9876740276510695e-06,
+      "loss": 0.9028,
+      "step": 7586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985161408741188e-06,
+      "loss": 0.9345,
+      "step": 7587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982649091969173e-06,
+      "loss": 0.9129,
+      "step": 7588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980137077524084e-06,
+      "loss": 0.9006,
+      "step": 7589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977625365594966e-06,
+      "loss": 2.722,
+      "step": 7590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.975113956370841e-06,
+      "loss": 0.9182,
+      "step": 7591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972602850040702e-06,
+      "loss": 0.8895,
+      "step": 7592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970092046793523e-06,
+      "loss": 0.9292,
+      "step": 7593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967581546818261e-06,
+      "loss": 0.9328,
+      "step": 7594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965071350303837e-06,
+      "loss": 0.9058,
+      "step": 7595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962561457439164e-06,
+      "loss": 0.9345,
+      "step": 7596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960051868413117e-06,
+      "loss": 0.9222,
+      "step": 7597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957542583414559e-06,
+      "loss": 0.9335,
+      "step": 7598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95503360263233e-06,
+      "loss": 0.909,
+      "step": 7599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952524926255238e-06,
+      "loss": 0.9101,
+      "step": 7600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950016554472082e-06,
+      "loss": 0.8879,
+      "step": 7601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947508487471623e-06,
+      "loss": 0.8776,
+      "step": 7602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945000725442609e-06,
+      "loss": 0.8989,
+      "step": 7603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9424932685737666e-06,
+      "loss": 0.9087,
+      "step": 7604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939986117053787e-06,
+      "loss": 0.9523,
+      "step": 7605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93747927107135e-06,
+      "loss": 0.8815,
+      "step": 7606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934972730815115e-06,
+      "loss": 0.9244,
+      "step": 7607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932466496473703e-06,
+      "loss": 0.9145,
+      "step": 7608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9299605682357285e-06,
+      "loss": 0.9207,
+      "step": 7609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927454946289773e-06,
+      "loss": 0.9187,
+      "step": 7610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924949630824395e-06,
+      "loss": 0.8968,
+      "step": 7611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922444622028138e-06,
+      "loss": 0.8411,
+      "step": 7612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919939920089513e-06,
+      "loss": 0.8914,
+      "step": 7613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.917435525197015e-06,
+      "loss": 0.9168,
+      "step": 7614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914931437539113e-06,
+      "loss": 0.9537,
+      "step": 7615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91242765730425e-06,
+      "loss": 0.9172,
+      "step": 7616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909924184680856e-06,
+      "loss": 0.8824,
+      "step": 7617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907421019857319e-06,
+      "loss": 0.9467,
+      "step": 7618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90491816302202e-06,
+      "loss": 0.9416,
+      "step": 7619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902415614363319e-06,
+      "loss": 0.9578,
+      "step": 7620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899913374069539e-06,
+      "loss": 0.8989,
+      "step": 7621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897411442328986e-06,
+      "loss": 0.8947,
+      "step": 7622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89490981932995e-06,
+      "loss": 0.9009,
+      "step": 7623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892408505260685e-06,
+      "loss": 0.8674,
+      "step": 7624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889907500309433e-06,
+      "loss": 0.9685,
+      "step": 7625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887406804664404e-06,
+      "loss": 0.8772,
+      "step": 7626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884906418513789e-06,
+      "loss": 0.9389,
+      "step": 7627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882406342045761e-06,
+      "loss": 0.8881,
+      "step": 7628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.879906575448455e-06,
+      "loss": 0.877,
+      "step": 7629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877407118909994e-06,
+      "loss": 0.9382,
+      "step": 7630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874907972618484e-06,
+      "loss": 0.9365,
+      "step": 7631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872409136761989e-06,
+      "loss": 0.9519,
+      "step": 7632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869910611528565e-06,
+      "loss": 0.8931,
+      "step": 7633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867412397106235e-06,
+      "loss": 0.9039,
+      "step": 7634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8649144936830045e-06,
+      "loss": 0.8965,
+      "step": 7635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.862416901446857e-06,
+      "loss": 0.9159,
+      "step": 7636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859919620585745e-06,
+      "loss": 0.8664,
+      "step": 7637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85742265128761e-06,
+      "loss": 0.92,
+      "step": 7638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85492599374035e-06,
+      "loss": 0.8809,
+      "step": 7639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8524296481318585e-06,
+      "loss": 0.9163,
+      "step": 7640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849933614650003e-06,
+      "loss": 0.8966,
+      "step": 7641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847437893482615e-06,
+      "loss": 0.884,
+      "step": 7642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844942484817515e-06,
+      "loss": 0.957,
+      "step": 7643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842447388842496e-06,
+      "loss": 0.9527,
+      "step": 7644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8399526057453245e-06,
+      "loss": 0.9233,
+      "step": 7645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837458135713751e-06,
+      "loss": 0.9478,
+      "step": 7646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8349639789354924e-06,
+      "loss": 0.9139,
+      "step": 7647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8324701355982495e-06,
+      "loss": 0.9021,
+      "step": 7648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829976605889695e-06,
+      "loss": 0.889,
+      "step": 7649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827483389997483e-06,
+      "loss": 0.878,
+      "step": 7650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824990488109241e-06,
+      "loss": 0.9008,
+      "step": 7651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822497900412577e-06,
+      "loss": 0.9337,
+      "step": 7652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820005627095062e-06,
+      "loss": 0.9133,
+      "step": 7653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817513668344261e-06,
+      "loss": 0.8974,
+      "step": 7654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815022024347697e-06,
+      "loss": 0.891,
+      "step": 7655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81253069529289e-06,
+      "loss": 0.9606,
+      "step": 7656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810039681367326e-06,
+      "loss": 2.4784,
+      "step": 7657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807548982758458e-06,
+      "loss": 0.9118,
+      "step": 7658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80505859965373e-06,
+      "loss": 0.9118,
+      "step": 7659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802568532240559e-06,
+      "loss": 0.9108,
+      "step": 7660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800078780706328e-06,
+      "loss": 0.8834,
+      "step": 7661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79758934523841e-06,
+      "loss": 0.8973,
+      "step": 7662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795100226024145e-06,
+      "loss": 0.8991,
+      "step": 7663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792611423250855e-06,
+      "loss": 0.8907,
+      "step": 7664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790122937105838e-06,
+      "loss": 0.8982,
+      "step": 7665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787634767776358e-06,
+      "loss": 0.9081,
+      "step": 7666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7851469154496685e-06,
+      "loss": 0.9384,
+      "step": 7667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782659380312996e-06,
+      "loss": 0.9616,
+      "step": 7668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780172162553532e-06,
+      "loss": 0.9035,
+      "step": 7669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777685262358459e-06,
+      "loss": 0.9032,
+      "step": 7670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.775198679914927e-06,
+      "loss": 0.909,
+      "step": 7671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772712415410067e-06,
+      "loss": 0.9223,
+      "step": 7672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770226469030985e-06,
+      "loss": 0.9105,
+      "step": 7673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767740840964755e-06,
+      "loss": 0.9454,
+      "step": 7674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765255531398444e-06,
+      "loss": 0.8933,
+      "step": 7675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762770540519067e-06,
+      "loss": 0.9337,
+      "step": 7676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76028586851365e-06,
+      "loss": 0.9288,
+      "step": 7677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757801515569175e-06,
+      "loss": 0.9973,
+      "step": 7678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7553174818725955e-06,
+      "loss": 0.9572,
+      "step": 7679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752833767610853e-06,
+      "loss": 0.9141,
+      "step": 7680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750350372970863e-06,
+      "loss": 0.8779,
+      "step": 7681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747867298139506e-06,
+      "loss": 0.9376,
+      "step": 7682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745384543303651e-06,
+      "loss": 0.9077,
+      "step": 7683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742902108650138e-06,
+      "loss": 2.5452,
+      "step": 7684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740419994365785e-06,
+      "loss": 0.9267,
+      "step": 7685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737938200637385e-06,
+      "loss": 0.9228,
+      "step": 7686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7354567276517e-06,
+      "loss": 0.9063,
+      "step": 7687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732975575595481e-06,
+      "loss": 0.9742,
+      "step": 7688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730494744655446e-06,
+      "loss": 2.5371,
+      "step": 7689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728014235018287e-06,
+      "loss": 0.933,
+      "step": 7690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7255340468706776e-06,
+      "loss": 0.9446,
+      "step": 7691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723054180399266e-06,
+      "loss": 0.9322,
+      "step": 7692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720574635790674e-06,
+      "loss": 0.9346,
+      "step": 7693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718095413231506e-06,
+      "loss": 0.8682,
+      "step": 7694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715616512908327e-06,
+      "loss": 0.9203,
+      "step": 7695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713137935007693e-06,
+      "loss": 0.8672,
+      "step": 7696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710659679716131e-06,
+      "loss": 0.9542,
+      "step": 7697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7081817472201386e-06,
+      "loss": 0.9061,
+      "step": 7698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7057041377062005e-06,
+      "loss": 0.9186,
+      "step": 7699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703226851360762e-06,
+      "loss": 0.9159,
+      "step": 7700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700749888370256e-06,
+      "loss": 0.88,
+      "step": 7701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698273248921089e-06,
+      "loss": 0.9253,
+      "step": 7702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695796933199635e-06,
+      "loss": 0.8706,
+      "step": 7703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.693320941392255e-06,
+      "loss": 0.9436,
+      "step": 7704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690845273685279e-06,
+      "loss": 0.8854,
+      "step": 7705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688369930265013e-06,
+      "loss": 0.9105,
+      "step": 7706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6858949113177465e-06,
+      "loss": 0.9089,
+      "step": 7707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683420217029727e-06,
+      "loss": 0.9102,
+      "step": 7708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680945847587194e-06,
+      "loss": 0.8906,
+      "step": 7709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67847180317636e-06,
+      "loss": 0.9374,
+      "step": 7710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675998083983399e-06,
+      "loss": 0.906,
+      "step": 7711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673524690194485e-06,
+      "loss": 0.9626,
+      "step": 7712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671051621995744e-06,
+      "loss": 0.9472,
+      "step": 7713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668578879573294e-06,
+      "loss": 0.8901,
+      "step": 7714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666106463113221e-06,
+      "loss": 0.9251,
+      "step": 7715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663634372801583e-06,
+      "loss": 0.8864,
+      "step": 7716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.9239,
+      "step": 7717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658691171367746e-06,
+      "loss": 0.924,
+      "step": 7718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65622006061755e-06,
+      "loss": 0.953,
+      "step": 7719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653749276759799e-06,
+      "loss": 2.5536,
+      "step": 7720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651278819980426e-06,
+      "loss": 0.9233,
+      "step": 7721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64880869046535e-06,
+      "loss": 0.8719,
+      "step": 7722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646338888400464e-06,
+      "loss": 0.9492,
+      "step": 7723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643869413971626e-06,
+      "loss": 0.9278,
+      "step": 7724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641400267364682e-06,
+      "loss": 0.9141,
+      "step": 7725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638931448765447e-06,
+      "loss": 0.918,
+      "step": 7726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636462958359714e-06,
+      "loss": 0.8996,
+      "step": 7727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.633994796333253e-06,
+      "loss": 0.8743,
+      "step": 7728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6315269628718e-06,
+      "loss": 0.922,
+      "step": 7729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629059458161074e-06,
+      "loss": 0.9362,
+      "step": 7730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626592282386769e-06,
+      "loss": 0.9094,
+      "step": 7731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624125435734553e-06,
+      "loss": 0.9148,
+      "step": 7732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621658918390069e-06,
+      "loss": 0.8579,
+      "step": 7733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.619192730538941e-06,
+      "loss": 0.9291,
+      "step": 7734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616726872366753e-06,
+      "loss": 0.9483,
+      "step": 7735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614261344059082e-06,
+      "loss": 0.9157,
+      "step": 7736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611796145801464e-06,
+      "loss": 0.9285,
+      "step": 7737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609331277779424e-06,
+      "loss": 0.9203,
+      "step": 7738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606866740178455e-06,
+      "loss": 0.9378,
+      "step": 7739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604402533184025e-06,
+      "loss": 0.9144,
+      "step": 7740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601938656981586e-06,
+      "loss": 2.659,
+      "step": 7741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599475111756547e-06,
+      "loss": 0.9362,
+      "step": 7742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597011897694308e-06,
+      "loss": 0.9063,
+      "step": 7743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594549014980242e-06,
+      "loss": 0.8827,
+      "step": 7744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.592086463799687e-06,
+      "loss": 0.8949,
+      "step": 7745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.589624244337964e-06,
+      "loss": 0.9399,
+      "step": 7746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5871623567803756e-06,
+      "loss": 0.9106,
+      "step": 7747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584700801312185e-06,
+      "loss": 0.9092,
+      "step": 7748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.582239578118642e-06,
+      "loss": 0.9125,
+      "step": 7749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579778687384963e-06,
+      "loss": 0.9031,
+      "step": 7750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57731812929634e-06,
+      "loss": 0.9161,
+      "step": 7751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.574857904037949e-06,
+      "loss": 0.9263,
+      "step": 7752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572398011794933e-06,
+      "loss": 0.8766,
+      "step": 7753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569938452752412e-06,
+      "loss": 0.9039,
+      "step": 7754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567479227095483e-06,
+      "loss": 0.9094,
+      "step": 7755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565020335009211e-06,
+      "loss": 0.9408,
+      "step": 7756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562561776678648e-06,
+      "loss": 0.8714,
+      "step": 7757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560103552288802e-06,
+      "loss": 0.9284,
+      "step": 7758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5576456620246755e-06,
+      "loss": 0.8746,
+      "step": 7759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555188106071237e-06,
+      "loss": 0.8742,
+      "step": 7760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552730884613429e-06,
+      "loss": 0.9775,
+      "step": 7761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550273997836171e-06,
+      "loss": 0.8955,
+      "step": 7762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547817445924361e-06,
+      "loss": 0.9411,
+      "step": 7763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545361229062862e-06,
+      "loss": 0.9425,
+      "step": 7764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54290534743652e-06,
+      "loss": 0.8865,
+      "step": 7765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54044980123015e-06,
+      "loss": 0.9328,
+      "step": 7766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537994590628542e-06,
+      "loss": 0.8934,
+      "step": 7767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535539715816477e-06,
+      "loss": 0.8904,
+      "step": 7768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533085176978684e-06,
+      "loss": 0.9251,
+      "step": 7769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530630974299887e-06,
+      "loss": 0.9203,
+      "step": 7770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528177107964776e-06,
+      "loss": 0.9076,
+      "step": 7771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525723578158016e-06,
+      "loss": 0.9024,
+      "step": 7772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523270385064248e-06,
+      "loss": 0.9008,
+      "step": 7773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.520817528868089e-06,
+      "loss": 0.9354,
+      "step": 7774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5183650097541295e-06,
+      "loss": 0.8777,
+      "step": 7775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5159128279069385e-06,
+      "loss": 0.831,
+      "step": 7776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5134609835110474e-06,
+      "loss": 0.8862,
+      "step": 7777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.511009476750979e-06,
+      "loss": 0.8756,
+      "step": 7778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508558307811213e-06,
+      "loss": 0.8935,
+      "step": 7779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506107476876218e-06,
+      "loss": 0.906,
+      "step": 7780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503656984130432e-06,
+      "loss": 0.8932,
+      "step": 7781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5012068297582665e-06,
+      "loss": 0.9273,
+      "step": 7782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498757013944109e-06,
+      "loss": 0.9532,
+      "step": 7783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4963075368723275e-06,
+      "loss": 0.9232,
+      "step": 7784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4938583987272465e-06,
+      "loss": 0.8724,
+      "step": 7785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4914095996931875e-06,
+      "loss": 0.9489,
+      "step": 7786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488961139954421e-06,
+      "loss": 0.9268,
+      "step": 7787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486513019695219e-06,
+      "loss": 0.9092,
+      "step": 7788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484065239099817e-06,
+      "loss": 0.9601,
+      "step": 7789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4816177983524146e-06,
+      "loss": 0.929,
+      "step": 7790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479170697637199e-06,
+      "loss": 0.9096,
+      "step": 7791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476723937138333e-06,
+      "loss": 0.8847,
+      "step": 7792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474277517039937e-06,
+      "loss": 0.8707,
+      "step": 7793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471831437526123e-06,
+      "loss": 0.9128,
+      "step": 7794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46938569878097e-06,
+      "loss": 0.904,
+      "step": 7795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.466940300988535e-06,
+      "loss": 0.8791,
+      "step": 7796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46449524433285e-06,
+      "loss": 0.9208,
+      "step": 7797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4620505289979106e-06,
+      "loss": 0.9245,
+      "step": 7798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459606155167698e-06,
+      "loss": 0.9583,
+      "step": 7799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45716212302617e-06,
+      "loss": 0.8948,
+      "step": 7800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454718432757242e-06,
+      "loss": 0.8751,
+      "step": 7801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452275084544821e-06,
+      "loss": 0.9159,
+      "step": 7802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449832078572781e-06,
+      "loss": 0.8971,
+      "step": 7803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447389415024973e-06,
+      "loss": 0.9045,
+      "step": 7804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.44494709408522e-06,
+      "loss": 0.856,
+      "step": 7805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4425051159373165e-06,
+      "loss": 0.9151,
+      "step": 7806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440063480765039e-06,
+      "loss": 0.9182,
+      "step": 7807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437622188752123e-06,
+      "loss": 0.9124,
+      "step": 7808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4351812400823e-06,
+      "loss": 0.9351,
+      "step": 7809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432740634939264e-06,
+      "loss": 0.9299,
+      "step": 7810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430300373506678e-06,
+      "loss": 0.957,
+      "step": 7811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427860455968186e-06,
+      "loss": 0.9525,
+      "step": 7812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425420882507408e-06,
+      "loss": 0.9303,
+      "step": 7813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4229816533079285e-06,
+      "loss": 0.9162,
+      "step": 7814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420542768553316e-06,
+      "loss": 0.9313,
+      "step": 7815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41810422842711e-06,
+      "loss": 0.9049,
+      "step": 7816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415666033112823e-06,
+      "loss": 0.9129,
+      "step": 7817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413228182793946e-06,
+      "loss": 0.9309,
+      "step": 7818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410790677653932e-06,
+      "loss": 0.9301,
+      "step": 7819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408353517876221e-06,
+      "loss": 0.8875,
+      "step": 7820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405916703644227e-06,
+      "loss": 0.9507,
+      "step": 7821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403480235141323e-06,
+      "loss": 0.8943,
+      "step": 7822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401044112550871e-06,
+      "loss": 0.8629,
+      "step": 7823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3986083360562015e-06,
+      "loss": 0.837,
+      "step": 7824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39617290584062e-06,
+      "loss": 0.8803,
+      "step": 7825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39373782208741e-06,
+      "loss": 0.9241,
+      "step": 7826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3913030849798165e-06,
+      "loss": 0.8818,
+      "step": 7827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38886869470107e-06,
+      "loss": 0.9066,
+      "step": 7828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3864346514343714e-06,
+      "loss": 2.5991,
+      "step": 7829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384000955362896e-06,
+      "loss": 2.625,
+      "step": 7830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3815676066697946e-06,
+      "loss": 0.9385,
+      "step": 7831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379134605538182e-06,
+      "loss": 0.8949,
+      "step": 7832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37670195215116e-06,
+      "loss": 0.9597,
+      "step": 7833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.374269646691801e-06,
+      "loss": 0.9236,
+      "step": 7834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371837689343142e-06,
+      "loss": 0.9303,
+      "step": 7835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3694060802882045e-06,
+      "loss": 0.9204,
+      "step": 7836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3669748197099795e-06,
+      "loss": 0.9161,
+      "step": 7837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364543907791432e-06,
+      "loss": 0.9365,
+      "step": 7838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362113344715505e-06,
+      "loss": 0.9425,
+      "step": 7839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359683130665103e-06,
+      "loss": 0.8979,
+      "step": 7840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357253265823119e-06,
+      "loss": 0.9069,
+      "step": 7841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354823750372414e-06,
+      "loss": 0.8666,
+      "step": 7842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352394584495811e-06,
+      "loss": 0.9491,
+      "step": 7843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349965768376135e-06,
+      "loss": 0.9182,
+      "step": 7844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3475373021961525e-06,
+      "loss": 0.8817,
+      "step": 7845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345109186138625e-06,
+      "loss": 0.9131,
+      "step": 7846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342681420386284e-06,
+      "loss": 0.9076,
+      "step": 7847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340254005121823e-06,
+      "loss": 0.8736,
+      "step": 7848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337826940527923e-06,
+      "loss": 0.8787,
+      "step": 7849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335400226787233e-06,
+      "loss": 0.9289,
+      "step": 7850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332973864082377e-06,
+      "loss": 0.9518,
+      "step": 7851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330547852595954e-06,
+      "loss": 0.9301,
+      "step": 7852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328122192510528e-06,
+      "loss": 0.9007,
+      "step": 7853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325696884008645e-06,
+      "loss": 0.9322,
+      "step": 7854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323271927272826e-06,
+      "loss": 0.9074,
+      "step": 7855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320847322485556e-06,
+      "loss": 0.9182,
+      "step": 7856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318423069829302e-06,
+      "loss": 0.9194,
+      "step": 7857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315999169486503e-06,
+      "loss": 0.8772,
+      "step": 7858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313575621639568e-06,
+      "loss": 0.907,
+      "step": 7859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311152426470889e-06,
+      "loss": 0.8949,
+      "step": 7860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3087295841628135e-06,
+      "loss": 0.9139,
+      "step": 7861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306307094897679e-06,
+      "loss": 0.9332,
+      "step": 7862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303884958857789e-06,
+      "loss": 0.9187,
+      "step": 7863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301463176225423e-06,
+      "loss": 0.9282,
+      "step": 7864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299041747182834e-06,
+      "loss": 0.9096,
+      "step": 7865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296620671912249e-06,
+      "loss": 0.9433,
+      "step": 7866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294199950595861e-06,
+      "loss": 0.8988,
+      "step": 7867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291779583415848e-06,
+      "loss": 0.8795,
+      "step": 7868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2893595705543486e-06,
+      "loss": 0.8927,
+      "step": 7869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.286939912193487e-06,
+      "loss": 0.9012,
+      "step": 7870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284520608515354e-06,
+      "loss": 0.8904,
+      "step": 7871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282101659702012e-06,
+      "loss": 2.6407,
+      "step": 7872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.279683065935505e-06,
+      "loss": 0.9137,
+      "step": 7873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277264827397845e-06,
+      "loss": 0.9042,
+      "step": 7874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274846944271012e-06,
+      "loss": 0.9027,
+      "step": 7875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272429416736971e-06,
+      "loss": 0.9017,
+      "step": 7876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270012244977647e-06,
+      "loss": 0.9407,
+      "step": 7877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.267595429174943e-06,
+      "loss": 0.9129,
+      "step": 7878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265178969510749e-06,
+      "loss": 0.9262,
+      "step": 7879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.262762866166908e-06,
+      "loss": 0.9025,
+      "step": 7880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2603471193252486e-06,
+      "loss": 0.904,
+      "step": 7881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257931729167562e-06,
+      "loss": 0.8974,
+      "step": 7882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2555166958756246e-06,
+      "loss": 2.6199,
+      "step": 7883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253102019631178e-06,
+      "loss": 2.5689,
+      "step": 7884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2506877006159414e-06,
+      "loss": 0.9702,
+      "step": 7885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2482737390116034e-06,
+      "loss": 0.8807,
+      "step": 7886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245860134999831e-06,
+      "loss": 0.9008,
+      "step": 7887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2434468887622554e-06,
+      "loss": 0.9095,
+      "step": 7888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241034000480493e-06,
+      "loss": 0.9331,
+      "step": 7889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238621470336117e-06,
+      "loss": 0.9125,
+      "step": 7890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2362092985106884e-06,
+      "loss": 0.9482,
+      "step": 7891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2337974851857355e-06,
+      "loss": 0.9382,
+      "step": 7892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23138603054276e-06,
+      "loss": 0.8212,
+      "step": 7893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228974934763237e-06,
+      "loss": 0.9469,
+      "step": 7894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226564198028616e-06,
+      "loss": 0.9136,
+      "step": 7895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224153820520314e-06,
+      "loss": 0.9384,
+      "step": 7896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22174380241973e-06,
+      "loss": 0.9148,
+      "step": 7897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219334143908223e-06,
+      "loss": 0.9076,
+      "step": 7898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216924845167132e-06,
+      "loss": 0.9646,
+      "step": 7899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2145159063777815e-06,
+      "loss": 0.9055,
+      "step": 7900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2121073277214455e-06,
+      "loss": 0.9304,
+      "step": 7901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209699109379387e-06,
+      "loss": 0.8818,
+      "step": 7902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20729125153284e-06,
+      "loss": 2.5942,
+      "step": 7903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204883754363e-06,
+      "loss": 0.8835,
+      "step": 7904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20247661805105e-06,
+      "loss": 0.943,
+      "step": 7905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200069842778138e-06,
+      "loss": 0.9667,
+      "step": 7906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197663428725388e-06,
+      "loss": 0.9059,
+      "step": 7907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195257376073898e-06,
+      "loss": 0.8925,
+      "step": 7908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192851685004728e-06,
+      "loss": 0.9184,
+      "step": 7909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190446355698928e-06,
+      "loss": 0.929,
+      "step": 7910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188041388337506e-06,
+      "loss": 0.928,
+      "step": 7911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185636783101447e-06,
+      "loss": 0.8952,
+      "step": 7912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183232540171714e-06,
+      "loss": 0.9255,
+      "step": 7913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18082865972924e-06,
+      "loss": 0.9078,
+      "step": 7914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178425141954926e-06,
+      "loss": 0.8885,
+      "step": 7915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176021987029656e-06,
+      "loss": 0.9185,
+      "step": 7916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173619195134272e-06,
+      "loss": 0.9115,
+      "step": 7917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171216766449605e-06,
+      "loss": 0.8925,
+      "step": 7918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168814701156437e-06,
+      "loss": 0.9368,
+      "step": 7919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166412999435551e-06,
+      "loss": 0.9132,
+      "step": 7920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164011661467682e-06,
+      "loss": 0.8929,
+      "step": 7921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161610687433542e-06,
+      "loss": 0.9373,
+      "step": 7922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159210077513818e-06,
+      "loss": 0.9091,
+      "step": 7923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.156809831889174e-06,
+      "loss": 0.8895,
+      "step": 7924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15440995074023e-06,
+      "loss": 0.8915,
+      "step": 7925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1520104342475965e-06,
+      "loss": 0.9067,
+      "step": 7926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149611282591849e-06,
+      "loss": 0.9096,
+      "step": 7927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147212495953538e-06,
+      "loss": 0.8906,
+      "step": 7928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1448140745131845e-06,
+      "loss": 0.9049,
+      "step": 7929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142416018451279e-06,
+      "loss": 0.8954,
+      "step": 7930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140018327948291e-06,
+      "loss": 0.8888,
+      "step": 7931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1376210031846614e-06,
+      "loss": 0.9173,
+      "step": 7932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135224044340796e-06,
+      "loss": 0.9101,
+      "step": 7933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.13282745159708e-06,
+      "loss": 0.8886,
+      "step": 7934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130431225133872e-06,
+      "loss": 0.9034,
+      "step": 7935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1280353651315e-06,
+      "loss": 0.9544,
+      "step": 7936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12563987177027e-06,
+      "loss": 0.9502,
+      "step": 7937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123244745230445e-06,
+      "loss": 0.8916,
+      "step": 7938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.120849985692276e-06,
+      "loss": 0.9054,
+      "step": 7939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.118455593335984e-06,
+      "loss": 0.9194,
+      "step": 7940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116061568341757e-06,
+      "loss": 0.8901,
+      "step": 7941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113667910889761e-06,
+      "loss": 0.9043,
+      "step": 7942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111274621160127e-06,
+      "loss": 0.9019,
+      "step": 7943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108881699332964e-06,
+      "loss": 0.9232,
+      "step": 7944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106489145588356e-06,
+      "loss": 0.9273,
+      "step": 7945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104096960106349e-06,
+      "loss": 0.898,
+      "step": 7946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101705143066972e-06,
+      "loss": 0.8839,
+      "step": 7947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0993136946502184e-06,
+      "loss": 0.8818,
+      "step": 7948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096922615036061e-06,
+      "loss": 0.8702,
+      "step": 7949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094531904404444e-06,
+      "loss": 0.8718,
+      "step": 7950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092141562935273e-06,
+      "loss": 0.9171,
+      "step": 7951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089751590808436e-06,
+      "loss": 0.894,
+      "step": 7952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087361988203798e-06,
+      "loss": 0.872,
+      "step": 7953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08497275530118e-06,
+      "loss": 0.9034,
+      "step": 7954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08258389228039e-06,
+      "loss": 0.9464,
+      "step": 7955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080195399321199e-06,
+      "loss": 0.8262,
+      "step": 7956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077807276603357e-06,
+      "loss": 0.904,
+      "step": 7957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075419524306584e-06,
+      "loss": 0.9065,
+      "step": 7958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073032142610565e-06,
+      "loss": 0.914,
+      "step": 7959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070645131694969e-06,
+      "loss": 0.9151,
+      "step": 7960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068258491739427e-06,
+      "loss": 0.9088,
+      "step": 7961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065872222923549e-06,
+      "loss": 0.9005,
+      "step": 7962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0634863254269175e-06,
+      "loss": 0.9493,
+      "step": 7963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061100799429076e-06,
+      "loss": 0.9148,
+      "step": 7964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058715645109553e-06,
+      "loss": 0.9238,
+      "step": 7965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056330862647844e-06,
+      "loss": 0.9228,
+      "step": 7966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053946452223415e-06,
+      "loss": 0.9299,
+      "step": 7967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051562414015705e-06,
+      "loss": 0.893,
+      "step": 7968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049178748204127e-06,
+      "loss": 0.9232,
+      "step": 7969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.046795454968063e-06,
+      "loss": 0.9159,
+      "step": 7970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044412534486873e-06,
+      "loss": 0.8958,
+      "step": 7971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042029986939878e-06,
+      "loss": 0.8914,
+      "step": 7972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039647812506381e-06,
+      "loss": 0.9035,
+      "step": 7973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037266011365655e-06,
+      "loss": 0.9249,
+      "step": 7974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034884583696933e-06,
+      "loss": 0.9097,
+      "step": 7975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0325035296794425e-06,
+      "loss": 0.8903,
+      "step": 7976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030122849492367e-06,
+      "loss": 0.9181,
+      "step": 7977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027742543314862e-06,
+      "loss": 0.8777,
+      "step": 7978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025362611326063e-06,
+      "loss": 0.9207,
+      "step": 7979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022983053705066e-06,
+      "loss": 0.8927,
+      "step": 7980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.020603870630949e-06,
+      "loss": 0.8882,
+      "step": 7981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0182250622827575e-06,
+      "loss": 0.8769,
+      "step": 7982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015846628839509e-06,
+      "loss": 0.9306,
+      "step": 7983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0134685704801996e-06,
+      "loss": 0.878,
+      "step": 7984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01109088738378e-06,
+      "loss": 0.916,
+      "step": 7985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00871357972919e-06,
+      "loss": 0.9351,
+      "step": 7986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006336647695336e-06,
+      "loss": 0.8584,
+      "step": 7987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003960091461088e-06,
+      "loss": 0.897,
+      "step": 7988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001583911205299e-06,
+      "loss": 0.949,
+      "step": 7989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999208107106787e-06,
+      "loss": 0.9345,
+      "step": 7990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9968326793443475e-06,
+      "loss": 0.9302,
+      "step": 7991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994457628096745e-06,
+      "loss": 0.9033,
+      "step": 7992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992082953542707e-06,
+      "loss": 0.8752,
+      "step": 7993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989708655860947e-06,
+      "loss": 0.88,
+      "step": 7994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98733473523014e-06,
+      "loss": 0.8684,
+      "step": 7995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984961191828936e-06,
+      "loss": 0.9217,
+      "step": 7996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98258802583596e-06,
+      "loss": 0.9295,
+      "step": 7997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980215237429808e-06,
+      "loss": 0.9643,
+      "step": 7998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977842826789034e-06,
+      "loss": 0.9195,
+      "step": 7999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9754707940921875e-06,
+      "loss": 0.9431,
+      "step": 8000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973099139517764e-06,
+      "loss": 0.9275,
+      "step": 8001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97072786324425e-06,
+      "loss": 0.9096,
+      "step": 8002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.968356965450095e-06,
+      "loss": 2.5179,
+      "step": 8003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965986446313723e-06,
+      "loss": 0.9068,
+      "step": 8004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963616306013527e-06,
+      "loss": 0.9204,
+      "step": 8005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961246544727877e-06,
+      "loss": 0.9216,
+      "step": 8006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9588771626351e-06,
+      "loss": 0.9187,
+      "step": 8007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956508159913517e-06,
+      "loss": 0.9196,
+      "step": 8008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954139536741397e-06,
+      "loss": 0.8955,
+      "step": 8009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951771293296992e-06,
+      "loss": 0.8802,
+      "step": 8010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949403429758536e-06,
+      "loss": 0.8825,
+      "step": 8011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947035946304213e-06,
+      "loss": 0.9045,
+      "step": 8012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944668843112195e-06,
+      "loss": 0.9299,
+      "step": 8013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942302120360612e-06,
+      "loss": 0.9163,
+      "step": 8014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939935778227576e-06,
+      "loss": 0.9309,
+      "step": 8015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.937569816891168e-06,
+      "loss": 0.921,
+      "step": 8016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935204236529437e-06,
+      "loss": 0.9251,
+      "step": 8017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932839037320407e-06,
+      "loss": 0.8561,
+      "step": 8018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930474219442075e-06,
+      "loss": 0.9388,
+      "step": 8019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928109783072399e-06,
+      "loss": 0.8745,
+      "step": 8020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925745728389322e-06,
+      "loss": 0.8744,
+      "step": 8021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923382055570744e-06,
+      "loss": 0.9049,
+      "step": 8022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921018764794551e-06,
+      "loss": 0.8777,
+      "step": 8023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918655856238588e-06,
+      "loss": 2.7661,
+      "step": 8024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916293330080679e-06,
+      "loss": 0.9028,
+      "step": 8025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913931186498617e-06,
+      "loss": 0.9198,
+      "step": 8026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911569425670168e-06,
+      "loss": 0.9334,
+      "step": 8027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909208047773062e-06,
+      "loss": 0.8956,
+      "step": 8028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9068470529850106e-06,
+      "loss": 0.9183,
+      "step": 8029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904486441483684e-06,
+      "loss": 0.9153,
+      "step": 8030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902126213446732e-06,
+      "loss": 0.9073,
+      "step": 8031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899766369051783e-06,
+      "loss": 0.9147,
+      "step": 8032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897406908476419e-06,
+      "loss": 0.8842,
+      "step": 8033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895047831898203e-06,
+      "loss": 0.8727,
+      "step": 8034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892689139494675e-06,
+      "loss": 0.9017,
+      "step": 8035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8903308314433296e-06,
+      "loss": 0.8998,
+      "step": 8036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887972907921646e-06,
+      "loss": 0.9413,
+      "step": 8037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8856153691070705e-06,
+      "loss": 0.9071,
+      "step": 8038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8832582151770215e-06,
+      "loss": 0.9436,
+      "step": 8039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880901446308889e-06,
+      "loss": 0.9432,
+      "step": 8040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878545062680025e-06,
+      "loss": 0.8828,
+      "step": 8041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876189064467766e-06,
+      "loss": 0.9021,
+      "step": 8042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873833451849415e-06,
+      "loss": 0.8638,
+      "step": 8043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871478225002236e-06,
+      "loss": 0.9413,
+      "step": 8044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869123384103479e-06,
+      "loss": 2.5177,
+      "step": 8045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866768929330356e-06,
+      "loss": 0.8908,
+      "step": 8046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864414860860054e-06,
+      "loss": 0.9061,
+      "step": 8047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86206117886973e-06,
+      "loss": 0.9312,
+      "step": 8048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859707883536507e-06,
+      "loss": 0.9263,
+      "step": 8049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857354975037487e-06,
+      "loss": 0.9461,
+      "step": 8050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85500245354973e-06,
+      "loss": 0.8877,
+      "step": 8051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852650319250285e-06,
+      "loss": 0.9148,
+      "step": 8052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8502985723161655e-06,
+      "loss": 0.8967,
+      "step": 8053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847947212924342e-06,
+      "loss": 0.8807,
+      "step": 8054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845596241251773e-06,
+      "loss": 0.8985,
+      "step": 8055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843245657475385e-06,
+      "loss": 0.9511,
+      "step": 8056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840895461772063e-06,
+      "loss": 0.9391,
+      "step": 8057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838545654318676e-06,
+      "loss": 0.8785,
+      "step": 8058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836196235292061e-06,
+      "loss": 0.9208,
+      "step": 8059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.833847204869022e-06,
+      "loss": 0.9066,
+      "step": 8060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83149856322634e-06,
+      "loss": 0.9279,
+      "step": 8061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829150310540757e-06,
+      "loss": 0.9595,
+      "step": 8062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826802446988994e-06,
+      "loss": 0.9201,
+      "step": 8063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8244549727477444e-06,
+      "loss": 0.9207,
+      "step": 8064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822107887993659e-06,
+      "loss": 0.9189,
+      "step": 8065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819761192903374e-06,
+      "loss": 0.9269,
+      "step": 8066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8174148876534894e-06,
+      "loss": 0.9213,
+      "step": 8067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815068972420578e-06,
+      "loss": 0.8971,
+      "step": 8068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.812723447381186e-06,
+      "loss": 0.8816,
+      "step": 8069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8103783127118175e-06,
+      "loss": 0.8843,
+      "step": 8070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8080335685889625e-06,
+      "loss": 0.9574,
+      "step": 8071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805689215189073e-06,
+      "loss": 0.9812,
+      "step": 8072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803345252688577e-06,
+      "loss": 0.9166,
+      "step": 8073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801001681263872e-06,
+      "loss": 0.8929,
+      "step": 8074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798658501091315e-06,
+      "loss": 0.9314,
+      "step": 8075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796315712347249e-06,
+      "loss": 0.9302,
+      "step": 8076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793973315207985e-06,
+      "loss": 0.9434,
+      "step": 8077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791631309849792e-06,
+      "loss": 0.9567,
+      "step": 8078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.789289696448922e-06,
+      "loss": 0.8853,
+      "step": 8079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786948475181594e-06,
+      "loss": 0.917,
+      "step": 8080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784607646223999e-06,
+      "loss": 0.8985,
+      "step": 8081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782267209752298e-06,
+      "loss": 0.913,
+      "step": 8082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779927165942616e-06,
+      "loss": 0.8701,
+      "step": 8083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777587514971056e-06,
+      "loss": 0.9238,
+      "step": 8084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775248257013693e-06,
+      "loss": 0.9138,
+      "step": 8085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772909392246561e-06,
+      "loss": 0.8803,
+      "step": 8086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770570920845676e-06,
+      "loss": 0.9008,
+      "step": 8087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768232842987022e-06,
+      "loss": 0.8773,
+      "step": 8088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7658951588465484e-06,
+      "loss": 0.9225,
+      "step": 8089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7635578686001835e-06,
+      "loss": 0.9434,
+      "step": 8090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761220972423814e-06,
+      "loss": 0.9016,
+      "step": 8091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758884470493308e-06,
+      "loss": 0.9033,
+      "step": 8092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756548362984498e-06,
+      "loss": 2.4723,
+      "step": 8093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754212650073189e-06,
+      "loss": 0.9183,
+      "step": 8094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7518773319351585e-06,
+      "loss": 0.859,
+      "step": 8095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749542408746146e-06,
+      "loss": 0.9197,
+      "step": 8096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74720788068187e-06,
+      "loss": 0.8941,
+      "step": 8097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744873747918019e-06,
+      "loss": 0.8259,
+      "step": 8098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742540010630243e-06,
+      "loss": 0.9511,
+      "step": 8099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740206668994169e-06,
+      "loss": 0.9355,
+      "step": 8100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737873723185397e-06,
+      "loss": 0.917,
+      "step": 8101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735541173379491e-06,
+      "loss": 0.9649,
+      "step": 8102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733209019751992e-06,
+      "loss": 0.9169,
+      "step": 8103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7308772624784e-06,
+      "loss": 0.9112,
+      "step": 8104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728545901734194e-06,
+      "loss": 0.9324,
+      "step": 8105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726214937694827e-06,
+      "loss": 0.9169,
+      "step": 8106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723884370535704e-06,
+      "loss": 0.8866,
+      "step": 8107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721554200432225e-06,
+      "loss": 0.9063,
+      "step": 8108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719224427559747e-06,
+      "loss": 0.8958,
+      "step": 8109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7168950520935895e-06,
+      "loss": 0.9304,
+      "step": 8110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714566074209058e-06,
+      "loss": 0.8907,
+      "step": 8111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7122374940814145e-06,
+      "loss": 0.8995,
+      "step": 8112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709909311885898e-06,
+      "loss": 0.9048,
+      "step": 8113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70758152779772e-06,
+      "loss": 0.8952,
+      "step": 8114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705254141992056e-06,
+      "loss": 0.9091,
+      "step": 8115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7029271546440546e-06,
+      "loss": 0.8866,
+      "step": 8116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700600565928839e-06,
+      "loss": 0.8922,
+      "step": 8117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698274376021487e-06,
+      "loss": 0.9626,
+      "step": 8118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695948585097066e-06,
+      "loss": 0.8655,
+      "step": 8119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693623193330598e-06,
+      "loss": 0.9566,
+      "step": 8120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691298200897083e-06,
+      "loss": 0.8814,
+      "step": 8121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688973607971488e-06,
+      "loss": 0.9103,
+      "step": 8122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686649414728753e-06,
+      "loss": 0.9142,
+      "step": 8123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684325621343789e-06,
+      "loss": 0.9117,
+      "step": 8124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6820022279914656e-06,
+      "loss": 0.9392,
+      "step": 8125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.9372,
+      "step": 8126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677356642084114e-06,
+      "loss": 0.8999,
+      "step": 8127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675034449878692e-06,
+      "loss": 0.8984,
+      "step": 8128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.672712658405123e-06,
+      "loss": 0.8933,
+      "step": 8129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67039126783814e-06,
+      "loss": 0.8621,
+      "step": 8130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668070278352432e-06,
+      "loss": 0.8753,
+      "step": 8131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665749690122673e-06,
+      "loss": 0.8777,
+      "step": 8132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663429503323492e-06,
+      "loss": 0.8979,
+      "step": 8133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6611097181295e-06,
+      "loss": 0.9022,
+      "step": 8134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658790334715273e-06,
+      "loss": 0.9335,
+      "step": 8135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.656471353255355e-06,
+      "loss": 0.8961,
+      "step": 8136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6541527739242645e-06,
+      "loss": 0.9251,
+      "step": 8137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651834596896488e-06,
+      "loss": 2.6289,
+      "step": 8138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649516822346474e-06,
+      "loss": 0.9433,
+      "step": 8139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647199450448656e-06,
+      "loss": 0.8685,
+      "step": 8140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64488248137742e-06,
+      "loss": 0.9128,
+      "step": 8141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642565915307129e-06,
+      "loss": 0.8823,
+      "step": 8142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640249752412131e-06,
+      "loss": 0.9079,
+      "step": 8143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637933992866714e-06,
+      "loss": 0.9389,
+      "step": 8144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63561863684516e-06,
+      "loss": 0.936,
+      "step": 8145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633303684521712e-06,
+      "loss": 2.4741,
+      "step": 8146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630989136070576e-06,
+      "loss": 0.9209,
+      "step": 8147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628674991665938e-06,
+      "loss": 0.9185,
+      "step": 8148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62636125148195e-06,
+      "loss": 0.8897,
+      "step": 8149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624047915692732e-06,
+      "loss": 0.8717,
+      "step": 8150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621734984472379e-06,
+      "loss": 2.7547,
+      "step": 8151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619422457994944e-06,
+      "loss": 0.9414,
+      "step": 8152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617110336434464e-06,
+      "loss": 0.9299,
+      "step": 8153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614798619964932e-06,
+      "loss": 0.8769,
+      "step": 8154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612487308760319e-06,
+      "loss": 0.9217,
+      "step": 8155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610176402994565e-06,
+      "loss": 0.9394,
+      "step": 8156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.607865902841576e-06,
+      "loss": 0.8357,
+      "step": 8157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6055558084752314e-06,
+      "loss": 0.8981,
+      "step": 8158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603246120069381e-06,
+      "loss": 0.8382,
+      "step": 8159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600936837797834e-06,
+      "loss": 0.925,
+      "step": 8160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598627961834382e-06,
+      "loss": 0.9026,
+      "step": 8161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5963194923527735e-06,
+      "loss": 0.8878,
+      "step": 8162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5940114295267345e-06,
+      "loss": 0.8404,
+      "step": 8163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591703773529968e-06,
+      "loss": 0.9446,
+      "step": 8164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589396524536127e-06,
+      "loss": 0.8914,
+      "step": 8165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587089682718849e-06,
+      "loss": 0.9116,
+      "step": 8166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584783248251738e-06,
+      "loss": 0.9306,
+      "step": 8167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582477221308359e-06,
+      "loss": 0.9223,
+      "step": 8168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580171602062257e-06,
+      "loss": 0.895,
+      "step": 8169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57786639068694e-06,
+      "loss": 0.9069,
+      "step": 8170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57556158735589e-06,
+      "loss": 0.8961,
+      "step": 8171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573257192242558e-06,
+      "loss": 0.9061,
+      "step": 8172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570953205520353e-06,
+      "loss": 0.9616,
+      "step": 8173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568649627362668e-06,
+      "loss": 0.9316,
+      "step": 8174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566346457942863e-06,
+      "loss": 0.9154,
+      "step": 8175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564043697434256e-06,
+      "loss": 0.8751,
+      "step": 8176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5617413460101455e-06,
+      "loss": 0.9145,
+      "step": 8177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559439403843795e-06,
+      "loss": 0.8693,
+      "step": 8178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5571378711084395e-06,
+      "loss": 0.9048,
+      "step": 8179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554836747977283e-06,
+      "loss": 0.9243,
+      "step": 8180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552536034623492e-06,
+      "loss": 0.931,
+      "step": 8181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.550235731220209e-06,
+      "loss": 0.9087,
+      "step": 8182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547935837940547e-06,
+      "loss": 0.911,
+      "step": 8183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545636354957582e-06,
+      "loss": 0.968,
+      "step": 8184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543337282444368e-06,
+      "loss": 0.9087,
+      "step": 8185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541038620573913e-06,
+      "loss": 0.9178,
+      "step": 8186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538740369519209e-06,
+      "loss": 0.8755,
+      "step": 8187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536442529453214e-06,
+      "loss": 0.8735,
+      "step": 8188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534145100548847e-06,
+      "loss": 0.8952,
+      "step": 8189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531848082979003e-06,
+      "loss": 0.8847,
+      "step": 8190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529551476916547e-06,
+      "loss": 0.9349,
+      "step": 8191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527255282534309e-06,
+      "loss": 0.9198,
+      "step": 8192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524959500005095e-06,
+      "loss": 0.9275,
+      "step": 8193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522664129501667e-06,
+      "loss": 0.9291,
+      "step": 8194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520369171196766e-06,
+      "loss": 2.5999,
+      "step": 8195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518074625263105e-06,
+      "loss": 0.8866,
+      "step": 8196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5157804918733525e-06,
+      "loss": 0.8646,
+      "step": 8197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5134867712001594e-06,
+      "loss": 0.8703,
+      "step": 8198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511193463416138e-06,
+      "loss": 0.9018,
+      "step": 8199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508900568693875e-06,
+      "loss": 0.876,
+      "step": 8200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506608087205923e-06,
+      "loss": 0.9115,
+      "step": 8201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504316019124797e-06,
+      "loss": 0.883,
+      "step": 8202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502024364622994e-06,
+      "loss": 0.8951,
+      "step": 8203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499733123872968e-06,
+      "loss": 0.8826,
+      "step": 8204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497442297047151e-06,
+      "loss": 0.9338,
+      "step": 8205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495151884317943e-06,
+      "loss": 0.8857,
+      "step": 8206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4928618858577e-06,
+      "loss": 0.8767,
+      "step": 8207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4905723018387626e-06,
+      "loss": 0.9074,
+      "step": 8208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488283132433438e-06,
+      "loss": 0.8644,
+      "step": 8209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485994377813988e-06,
+      "loss": 0.8942,
+      "step": 8210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48370603815266e-06,
+      "loss": 0.91,
+      "step": 8211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481418113621663e-06,
+      "loss": 0.9156,
+      "step": 8212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4791306043931746e-06,
+      "loss": 0.9198,
+      "step": 8213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476843510639347e-06,
+      "loss": 0.9164,
+      "step": 8214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474556832532289e-06,
+      "loss": 0.9092,
+      "step": 8215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472270570244086e-06,
+      "loss": 0.8856,
+      "step": 8216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4699847239467975e-06,
+      "loss": 0.9065,
+      "step": 8217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467699293812438e-06,
+      "loss": 0.8654,
+      "step": 8218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4654142800129975e-06,
+      "loss": 0.8165,
+      "step": 8219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463129682720448e-06,
+      "loss": 0.9082,
+      "step": 8220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460845502106703e-06,
+      "loss": 2.581,
+      "step": 8221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458561738343671e-06,
+      "loss": 0.9511,
+      "step": 8222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456278391603207e-06,
+      "loss": 0.907,
+      "step": 8223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453995462057149e-06,
+      "loss": 2.5611,
+      "step": 8224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4517129498773e-06,
+      "loss": 0.9306,
+      "step": 8225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449430855235433e-06,
+      "loss": 0.9069,
+      "step": 8226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447149178303287e-06,
+      "loss": 0.9164,
+      "step": 8227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444867919252566e-06,
+      "loss": 0.9087,
+      "step": 8228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4425870782549505e-06,
+      "loss": 0.8885,
+      "step": 8229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440306655482089e-06,
+      "loss": 0.8537,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438026651105588e-06,
+      "loss": 0.9068,
+      "step": 8231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435747065297034e-06,
+      "loss": 0.9344,
+      "step": 8232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433467898227976e-06,
+      "loss": 0.9298,
+      "step": 8233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431189150069936e-06,
+      "loss": 0.8285,
+      "step": 8234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428910820994403e-06,
+      "loss": 0.8829,
+      "step": 8235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426632911172829e-06,
+      "loss": 0.9175,
+      "step": 8236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42435542077664e-06,
+      "loss": 0.9129,
+      "step": 8237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4220783499772335e-06,
+      "loss": 0.8699,
+      "step": 8238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419801698945961e-06,
+      "loss": 0.8962,
+      "step": 8239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417525467854162e-06,
+      "loss": 2.5345,
+      "step": 8240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415249656873136e-06,
+      "loss": 0.9411,
+      "step": 8241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412974266174144e-06,
+      "loss": 0.9044,
+      "step": 8242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410699295928425e-06,
+      "loss": 0.9038,
+      "step": 8243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408424746307177e-06,
+      "loss": 0.8941,
+      "step": 8244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406150617481577e-06,
+      "loss": 0.8748,
+      "step": 8245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403876909622763e-06,
+      "loss": 0.8789,
+      "step": 8246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401603622901845e-06,
+      "loss": 2.5222,
+      "step": 8247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399330757489898e-06,
+      "loss": 0.896,
+      "step": 8248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3970583135579725e-06,
+      "loss": 0.9175,
+      "step": 8249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394786291277075e-06,
+      "loss": 0.9075,
+      "step": 8250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392514690818194e-06,
+      "loss": 0.8947,
+      "step": 8251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39024351235227e-06,
+      "loss": 0.9373,
+      "step": 8252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387972756050228e-06,
+      "loss": 0.9025,
+      "step": 8253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385702422082953e-06,
+      "loss": 0.904,
+      "step": 8254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3834325106213e-06,
+      "loss": 0.9194,
+      "step": 8255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381163021836097e-06,
+      "loss": 0.8891,
+      "step": 8256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378893955898123e-06,
+      "loss": 0.8859,
+      "step": 8257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376625312978147e-06,
+      "loss": 0.9434,
+      "step": 8258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3743570932468916e-06,
+      "loss": 0.8659,
+      "step": 8259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372089296875054e-06,
+      "loss": 0.9038,
+      "step": 8260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369821924033298e-06,
+      "loss": 0.8935,
+      "step": 8261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367554974892259e-06,
+      "loss": 0.8863,
+      "step": 8262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365288449622531e-06,
+      "loss": 0.8656,
+      "step": 8263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3630223483946874e-06,
+      "loss": 0.8582,
+      "step": 8264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360756671379258e-06,
+      "loss": 0.9356,
+      "step": 8265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35849141874675e-06,
+      "loss": 0.8968,
+      "step": 8266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356226590667635e-06,
+      "loss": 0.915,
+      "step": 8267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353962187312355e-06,
+      "loss": 0.9406,
+      "step": 8268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3516982088513186e-06,
+      "loss": 0.9449,
+      "step": 8269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349434655454903e-06,
+      "loss": 0.9075,
+      "step": 8270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347171527293448e-06,
+      "loss": 0.8991,
+      "step": 8271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344908824537272e-06,
+      "loss": 0.9127,
+      "step": 8272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3426465473566484e-06,
+      "loss": 0.9178,
+      "step": 8273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340384695921827e-06,
+      "loss": 0.9347,
+      "step": 8274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338123270403032e-06,
+      "loss": 0.8955,
+      "step": 8275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335862270970438e-06,
+      "loss": 0.8885,
+      "step": 8276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333601697794202e-06,
+      "loss": 0.86,
+      "step": 8277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331341551044446e-06,
+      "loss": 2.4831,
+      "step": 8278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.329081830891253e-06,
+      "loss": 0.8998,
+      "step": 8279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3268225375046814e-06,
+      "loss": 0.9165,
+      "step": 8280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324563671054753e-06,
+      "loss": 0.9004,
+      "step": 8281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322305231711462e-06,
+      "loss": 0.9411,
+      "step": 8282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320047219644772e-06,
+      "loss": 0.9079,
+      "step": 8283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3177896350246006e-06,
+      "loss": 0.9359,
+      "step": 8284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315532478020847e-06,
+      "loss": 0.8982,
+      "step": 8285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313275748803381e-06,
+      "loss": 2.6378,
+      "step": 8286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311019447542022e-06,
+      "loss": 0.9066,
+      "step": 8287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3087635744065745e-06,
+      "loss": 0.8435,
+      "step": 8288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306508129566805e-06,
+      "loss": 0.8571,
+      "step": 8289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304253113192447e-06,
+      "loss": 0.9042,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301998525453207e-06,
+      "loss": 0.9258,
+      "step": 8291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299744366518746e-06,
+      "loss": 0.9159,
+      "step": 8292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29749063655871e-06,
+      "loss": 0.8821,
+      "step": 8293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295237335742695e-06,
+      "loss": 0.9468,
+      "step": 8294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292984464240275e-06,
+      "loss": 0.9443,
+      "step": 8295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2907320222210025e-06,
+      "loss": 0.9358,
+      "step": 8296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288480009854374e-06,
+      "loss": 0.9201,
+      "step": 8297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286228427309867e-06,
+      "loss": 0.9047,
+      "step": 8298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283977274756932e-06,
+      "loss": 0.8831,
+      "step": 8299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28172655236497e-06,
+      "loss": 0.9472,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279476260303365e-06,
+      "loss": 2.5462,
+      "step": 8301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277226398741463e-06,
+      "loss": 0.8742,
+      "step": 8302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2749769678485785e-06,
+      "loss": 0.8959,
+      "step": 8303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272727967793996e-06,
+      "loss": 0.8808,
+      "step": 8304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270479398746957e-06,
+      "loss": 0.9313,
+      "step": 8305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268231260876683e-06,
+      "loss": 0.9104,
+      "step": 8306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265983554352361e-06,
+      "loss": 2.5872,
+      "step": 8307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263736279343137e-06,
+      "loss": 0.9448,
+      "step": 8308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2614894360181324e-06,
+      "loss": 0.872,
+      "step": 8309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2592430245464345e-06,
+      "loss": 0.8829,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256997045097097e-06,
+      "loss": 0.8747,
+      "step": 8311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254751497839147e-06,
+      "loss": 0.8635,
+      "step": 8312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252506382941566e-06,
+      "loss": 0.8965,
+      "step": 8313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250261700573314e-06,
+      "loss": 0.882,
+      "step": 8314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248017450903314e-06,
+      "loss": 0.8893,
+      "step": 8315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245773634100461e-06,
+      "loss": 0.9394,
+      "step": 8316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243530250333614e-06,
+      "loss": 0.9161,
+      "step": 8317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241287299771595e-06,
+      "loss": 0.89,
+      "step": 8318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2390447825832e-06,
+      "loss": 0.9027,
+      "step": 8319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.236802698937195e-06,
+      "loss": 0.9201,
+      "step": 8320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2345610490023e-06,
+      "loss": 0.9082,
+      "step": 8321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232319832947216e-06,
+      "loss": 0.8791,
+      "step": 8322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230079050940607e-06,
+      "loss": 2.394,
+      "step": 8323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227838703151101e-06,
+      "loss": 0.9103,
+      "step": 8324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225598789747302e-06,
+      "loss": 0.8745,
+      "step": 8325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223359310897767e-06,
+      "loss": 0.9098,
+      "step": 8326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221120266771033e-06,
+      "loss": 0.9001,
+      "step": 8327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218881657535603e-06,
+      "loss": 0.936,
+      "step": 8328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216643483359937e-06,
+      "loss": 0.836,
+      "step": 8329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214405744412473e-06,
+      "loss": 2.5243,
+      "step": 8330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212168440861613e-06,
+      "loss": 0.915,
+      "step": 8331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209931572875726e-06,
+      "loss": 0.8874,
+      "step": 8332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207695140623151e-06,
+      "loss": 0.8468,
+      "step": 8333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205459144272186e-06,
+      "loss": 0.8965,
+      "step": 8334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203223583991103e-06,
+      "loss": 0.9021,
+      "step": 8335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2009884599481406e-06,
+      "loss": 0.8886,
+      "step": 8336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198753772311505e-06,
+      "loss": 0.8561,
+      "step": 8337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19651952124937e-06,
+      "loss": 0.9343,
+      "step": 8338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194285706929868e-06,
+      "loss": 0.904,
+      "step": 8339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19205232952111e-06,
+      "loss": 0.9211,
+      "step": 8340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189819389191172e-06,
+      "loss": 0.8776,
+      "step": 8341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187586886108087e-06,
+      "loss": 0.9452,
+      "step": 8342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.185354820439867e-06,
+      "loss": 0.9255,
+      "step": 8343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183123192354488e-06,
+      "loss": 0.8784,
+      "step": 8344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18089200201989e-06,
+      "loss": 0.9518,
+      "step": 8345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178661249603986e-06,
+      "loss": 0.917,
+      "step": 8346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176430935274645e-06,
+      "loss": 0.8864,
+      "step": 8347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174201059199713e-06,
+      "loss": 0.8673,
+      "step": 8348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171971621547004e-06,
+      "loss": 0.9042,
+      "step": 8349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1697426224842884e-06,
+      "loss": 0.9279,
+      "step": 8350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167514062179308e-06,
+      "loss": 0.8602,
+      "step": 8351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165285940799788e-06,
+      "loss": 0.8778,
+      "step": 8352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163058258513393e-06,
+      "loss": 0.9358,
+      "step": 8353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160831015487775e-06,
+      "loss": 0.9141,
+      "step": 8354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15860421189054e-06,
+      "loss": 0.864,
+      "step": 8355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156377847889271e-06,
+      "loss": 0.9262,
+      "step": 8356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.154151923651511e-06,
+      "loss": 2.4822,
+      "step": 8357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151926439344774e-06,
+      "loss": 0.9231,
+      "step": 8358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149701395136542e-06,
+      "loss": 0.8895,
+      "step": 8359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147476791194256e-06,
+      "loss": 0.882,
+      "step": 8360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145252627685332e-06,
+      "loss": 0.8399,
+      "step": 8361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143028904777154e-06,
+      "loss": 0.8783,
+      "step": 8362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140805622637062e-06,
+      "loss": 0.8734,
+      "step": 8363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138582781432372e-06,
+      "loss": 0.913,
+      "step": 8364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136360381330366e-06,
+      "loss": 0.9222,
+      "step": 8365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134138422498291e-06,
+      "loss": 0.8715,
+      "step": 8366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1319169051033634e-06,
+      "loss": 0.858,
+      "step": 8367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129695829312759e-06,
+      "loss": 0.9194,
+      "step": 8368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127475195293628e-06,
+      "loss": 0.9046,
+      "step": 8369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.125255003213088e-06,
+      "loss": 0.9186,
+      "step": 8370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12303525323821e-06,
+      "loss": 0.9349,
+      "step": 8371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120815945536053e-06,
+      "loss": 0.903,
+      "step": 8372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118597080273632e-06,
+      "loss": 0.8805,
+      "step": 8373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11637865761792e-06,
+      "loss": 0.8342,
+      "step": 8374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114160677735873e-06,
+      "loss": 0.8929,
+      "step": 8375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111943140794399e-06,
+      "loss": 0.8585,
+      "step": 8376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109726046960382e-06,
+      "loss": 0.8871,
+      "step": 8377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107509396400671e-06,
+      "loss": 0.8909,
+      "step": 8378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105293189282079e-06,
+      "loss": 0.9481,
+      "step": 8379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103077425771389e-06,
+      "loss": 0.9374,
+      "step": 8380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100862106035351e-06,
+      "loss": 0.9143,
+      "step": 8381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0986472302406756e-06,
+      "loss": 0.8955,
+      "step": 8382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096432798554048e-06,
+      "loss": 0.9008,
+      "step": 8383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094218811142109e-06,
+      "loss": 0.8974,
+      "step": 8384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092005268171476e-06,
+      "loss": 0.8895,
+      "step": 8385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089792169808733e-06,
+      "loss": 0.9585,
+      "step": 8386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0875795162204245e-06,
+      "loss": 0.8544,
+      "step": 8387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085367307573065e-06,
+      "loss": 0.9029,
+      "step": 8388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083155544033138e-06,
+      "loss": 0.9237,
+      "step": 8389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080944225767085e-06,
+      "loss": 0.9268,
+      "step": 8390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078733352941322e-06,
+      "loss": 2.5723,
+      "step": 8391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0765229257222285e-06,
+      "loss": 0.9049,
+      "step": 8392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07431294427615e-06,
+      "loss": 0.932,
+      "step": 8393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072103408769406e-06,
+      "loss": 0.9484,
+      "step": 8394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0698943193682645e-06,
+      "loss": 0.9102,
+      "step": 8395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067685676238981e-06,
+      "loss": 0.9232,
+      "step": 8396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065477479547761e-06,
+      "loss": 0.9426,
+      "step": 8397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0632697294607846e-06,
+      "loss": 0.9015,
+      "step": 8398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061062426144198e-06,
+      "loss": 0.8576,
+      "step": 8399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058855569764112e-06,
+      "loss": 2.4965,
+      "step": 8400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056649160486604e-06,
+      "loss": 0.8724,
+      "step": 8401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054443198477722e-06,
+      "loss": 0.9272,
+      "step": 8402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0522376839034686e-06,
+      "loss": 0.9519,
+      "step": 8403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050032616929828e-06,
+      "loss": 0.921,
+      "step": 8404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047827997722736e-06,
+      "loss": 0.8612,
+      "step": 8405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045623826448102e-06,
+      "loss": 0.8998,
+      "step": 8406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043420103271813e-06,
+      "loss": 0.8926,
+      "step": 8407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041216828359698e-06,
+      "loss": 0.895,
+      "step": 8408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039014001877569e-06,
+      "loss": 2.4816,
+      "step": 8409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036811623991206e-06,
+      "loss": 0.9071,
+      "step": 8410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034609694866341e-06,
+      "loss": 0.9226,
+      "step": 8411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032408214668686e-06,
+      "loss": 0.905,
+      "step": 8412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030207183563911e-06,
+      "loss": 0.9203,
+      "step": 8413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028006601717658e-06,
+      "loss": 0.8912,
+      "step": 8414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025806469295534e-06,
+      "loss": 0.899,
+      "step": 8415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0236067864631065e-06,
+      "loss": 0.8975,
+      "step": 8416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021407553385913e-06,
+      "loss": 2.4923,
+      "step": 8417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019208770229464e-06,
+      "loss": 0.9419,
+      "step": 8418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.01701043715922e-06,
+      "loss": 0.8906,
+      "step": 8419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014812554340624e-06,
+      "loss": 0.8628,
+      "step": 8420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012615121939074e-06,
+      "loss": 0.8659,
+      "step": 8421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0104181401199415e-06,
+      "loss": 0.9055,
+      "step": 8422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0082216090485646e-06,
+      "loss": 0.8834,
+      "step": 8423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006025528890236e-06,
+      "loss": 0.8969,
+      "step": 8424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003829899810226e-06,
+      "loss": 0.8604,
+      "step": 8425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001634721973769e-06,
+      "loss": 0.9337,
+      "step": 8426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999439995546057e-06,
+      "loss": 0.901,
+      "step": 8427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997245720692267e-06,
+      "loss": 0.9228,
+      "step": 8428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995051897577518e-06,
+      "loss": 0.9192,
+      "step": 8429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992858526366911e-06,
+      "loss": 2.4583,
+      "step": 8430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990665607225513e-06,
+      "loss": 2.5196,
+      "step": 8431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988473140318347e-06,
+      "loss": 0.9121,
+      "step": 8432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986281125810408e-06,
+      "loss": 0.8799,
+      "step": 8433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984089563866659e-06,
+      "loss": 0.9028,
+      "step": 8434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981898454652025e-06,
+      "loss": 0.8424,
+      "step": 8435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979707798331406e-06,
+      "loss": 0.9483,
+      "step": 8436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977517595069649e-06,
+      "loss": 0.9146,
+      "step": 8437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9753278450315824e-06,
+      "loss": 0.8958,
+      "step": 8438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973138548382003e-06,
+      "loss": 0.9318,
+      "step": 8439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970949705285657e-06,
+      "loss": 2.5296,
+      "step": 8440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968761315907271e-06,
+      "loss": 0.8919,
+      "step": 8441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966573380411533e-06,
+      "loss": 0.884,
+      "step": 8442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964385898963097e-06,
+      "loss": 0.9213,
+      "step": 8443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9621988717265855e-06,
+      "loss": 0.887,
+      "step": 8444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960012298866577e-06,
+      "loss": 0.9548,
+      "step": 8445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9578261805476256e-06,
+      "loss": 0.9229,
+      "step": 8446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95564051693425e-06,
+      "loss": 0.8635,
+      "step": 8447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953455308190932e-06,
+      "loss": 0.9,
+      "step": 8448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951270554482123e-06,
+      "loss": 0.9007,
+      "step": 8449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949086255972231e-06,
+      "loss": 0.856,
+      "step": 8450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9469024128256396e-06,
+      "loss": 0.9162,
+      "step": 8451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944719025206697e-06,
+      "loss": 2.5236,
+      "step": 8452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942536093279708e-06,
+      "loss": 0.9065,
+      "step": 8453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940353617208955e-06,
+      "loss": 0.9695,
+      "step": 8454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93817159715868e-06,
+      "loss": 0.9289,
+      "step": 8455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93599003329309e-06,
+      "loss": 0.9148,
+      "step": 8456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9338089257763646e-06,
+      "loss": 0.8974,
+      "step": 8457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931628274772637e-06,
+      "loss": 0.8883,
+      "step": 8458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929448080446014e-06,
+      "loss": 0.896,
+      "step": 8459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927268342960572e-06,
+      "loss": 0.9256,
+      "step": 8460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925089062480339e-06,
+      "loss": 0.9148,
+      "step": 8461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922910239169324e-06,
+      "loss": 0.9227,
+      "step": 8462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9207318731914924e-06,
+      "loss": 0.9272,
+      "step": 8463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918553964710779e-06,
+      "loss": 0.9252,
+      "step": 8464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9163765138910856e-06,
+      "loss": 0.8787,
+      "step": 8465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91419952089627e-06,
+      "loss": 0.8883,
+      "step": 8466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912022985890167e-06,
+      "loss": 0.9603,
+      "step": 8467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909846909036572e-06,
+      "loss": 0.9412,
+      "step": 8468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907671290499245e-06,
+      "loss": 0.9157,
+      "step": 8469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905496130441918e-06,
+      "loss": 0.9546,
+      "step": 8470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903321429028276e-06,
+      "loss": 0.9067,
+      "step": 8471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901147186421979e-06,
+      "loss": 0.897,
+      "step": 8472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898973402786654e-06,
+      "loss": 0.9351,
+      "step": 8473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896800078285885e-06,
+      "loss": 0.9008,
+      "step": 8474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894627213083227e-06,
+      "loss": 0.9498,
+      "step": 8475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892454807342202e-06,
+      "loss": 0.9501,
+      "step": 8476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8902828612262936e-06,
+      "loss": 0.9078,
+      "step": 8477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8881113748989544e-06,
+      "loss": 0.85,
+      "step": 8478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885940348523596e-06,
+      "loss": 0.8865,
+      "step": 8479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883769782263601e-06,
+      "loss": 0.9181,
+      "step": 8480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881599676282321e-06,
+      "loss": 0.8752,
+      "step": 8481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87943003074306e-06,
+      "loss": 2.5001,
+      "step": 8482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8772608458090955e-06,
+      "loss": 0.891,
+      "step": 8483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875092121643681e-06,
+      "loss": 0.9055,
+      "step": 8484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872923858410014e-06,
+      "loss": 0.8857,
+      "step": 8485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870756056271274e-06,
+      "loss": 2.5529,
+      "step": 8486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8685887153905935e-06,
+      "loss": 0.8896,
+      "step": 8487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86642183593108e-06,
+      "loss": 0.9016,
+      "step": 8488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864255418055801e-06,
+      "loss": 0.898,
+      "step": 8489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862089461927793e-06,
+      "loss": 0.8949,
+      "step": 8490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8599239677100545e-06,
+      "loss": 0.8942,
+      "step": 8491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857758935565554e-06,
+      "loss": 0.9456,
+      "step": 8492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855594365657216e-06,
+      "loss": 0.9206,
+      "step": 8493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85343025814794e-06,
+      "loss": 2.536,
+      "step": 8494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851266613200581e-06,
+      "loss": 0.8934,
+      "step": 8495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849103430977971e-06,
+      "loss": 0.9014,
+      "step": 8496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846940711642896e-06,
+      "loss": 0.9262,
+      "step": 8497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844778455358115e-06,
+      "loss": 0.841,
+      "step": 8498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842616662286352e-06,
+      "loss": 0.9715,
+      "step": 8499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840455332590287e-06,
+      "loss": 0.9228,
+      "step": 8500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838294466432575e-06,
+      "loss": 0.9201,
+      "step": 8501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8361340639758345e-06,
+      "loss": 0.9304,
+      "step": 8502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833974125382639e-06,
+      "loss": 0.9044,
+      "step": 8503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831814650815544e-06,
+      "loss": 2.3806,
+      "step": 8504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829655640437062e-06,
+      "loss": 0.8902,
+      "step": 8505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827497094409663e-06,
+      "loss": 0.9012,
+      "step": 8506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825339012895798e-06,
+      "loss": 0.8493,
+      "step": 8507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823181396057863e-06,
+      "loss": 0.9095,
+      "step": 8508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821024244058237e-06,
+      "loss": 0.9016,
+      "step": 8509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818867557059255e-06,
+      "loss": 0.9518,
+      "step": 8510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81671133522322e-06,
+      "loss": 0.9128,
+      "step": 8511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8145555787124e-06,
+      "loss": 0.9191,
+      "step": 8512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8124002876890287e-06,
+      "loss": 0.9155,
+      "step": 8513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8102454623152973e-06,
+      "loss": 0.9212,
+      "step": 8514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808091102753375e-06,
+      "loss": 0.9289,
+      "step": 8515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805937209165381e-06,
+      "loss": 0.8818,
+      "step": 8516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8037837817134116e-06,
+      "loss": 0.9026,
+      "step": 8517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801630820559523e-06,
+      "loss": 0.9003,
+      "step": 8518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7994783258657374e-06,
+      "loss": 0.9012,
+      "step": 8519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7973262977940405e-06,
+      "loss": 0.9538,
+      "step": 8520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7951747365063884e-06,
+      "loss": 0.8889,
+      "step": 8521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7930236421646894e-06,
+      "loss": 0.9315,
+      "step": 8522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7908730149308292e-06,
+      "loss": 0.8944,
+      "step": 8523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7887228549666534e-06,
+      "loss": 0.889,
+      "step": 8524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786573162433973e-06,
+      "loss": 0.8869,
+      "step": 8525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7844239374945678e-06,
+      "loss": 0.9252,
+      "step": 8526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78227518031017e-06,
+      "loss": 0.908,
+      "step": 8527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7801268910424904e-06,
+      "loss": 2.4973,
+      "step": 8528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7779790698532005e-06,
+      "loss": 0.9165,
+      "step": 8529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775831716903929e-06,
+      "loss": 0.8836,
+      "step": 8530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7736848323562803e-06,
+      "loss": 0.9195,
+      "step": 8531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7715384163718172e-06,
+      "loss": 0.9056,
+      "step": 8532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7693924691120696e-06,
+      "loss": 0.883,
+      "step": 8533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7672469907385358e-06,
+      "loss": 0.9262,
+      "step": 8534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.9332,
+      "step": 8535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76295744129589e-06,
+      "loss": 0.887,
+      "step": 8536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760813370549591e-06,
+      "loss": 0.9373,
+      "step": 8537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7586697693351193e-06,
+      "loss": 0.9028,
+      "step": 8538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756526637813804e-06,
+      "loss": 0.8967,
+      "step": 8539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.754383976146918e-06,
+      "loss": 0.8899,
+      "step": 8540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.752241784495708e-06,
+      "loss": 0.9061,
+      "step": 8541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.750100063021391e-06,
+      "loss": 0.8937,
+      "step": 8542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7479588118851373e-06,
+      "loss": 0.8845,
+      "step": 8543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7458180312480876e-06,
+      "loss": 0.8729,
+      "step": 8544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743677721271348e-06,
+      "loss": 0.9349,
+      "step": 8545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.74153788211599e-06,
+      "loss": 0.9236,
+      "step": 8546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739398513943049e-06,
+      "loss": 0.9142,
+      "step": 8547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737259616913519e-06,
+      "loss": 0.896,
+      "step": 8548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7351211911883644e-06,
+      "loss": 0.8935,
+      "step": 8549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7329832369285178e-06,
+      "loss": 0.8584,
+      "step": 8550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730845754294865e-06,
+      "loss": 0.891,
+      "step": 8551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728708743448266e-06,
+      "loss": 0.8726,
+      "step": 8552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7265722045495423e-06,
+      "loss": 2.5258,
+      "step": 8553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7244361377594785e-06,
+      "loss": 0.851,
+      "step": 8554
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7223005432388306e-06,
+      "loss": 0.8725,
+      "step": 8555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720165421148305e-06,
+      "loss": 0.8504,
+      "step": 8556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718030771648585e-06,
+      "loss": 0.8824,
+      "step": 8557
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715896594900317e-06,
+      "loss": 0.9262,
+      "step": 8558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7137628910640997e-06,
+      "loss": 0.9259,
+      "step": 8559
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7116296603005186e-06,
+      "loss": 0.9203,
+      "step": 8560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.709496902770101e-06,
+      "loss": 0.9309,
+      "step": 8561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.707364618633351e-06,
+      "loss": 0.8991,
+      "step": 8562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7052328080507382e-06,
+      "loss": 0.8245,
+      "step": 8563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703101471182685e-06,
+      "loss": 0.8798,
+      "step": 8564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.70097060818959e-06,
+      "loss": 0.8645,
+      "step": 8565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6988402192318117e-06,
+      "loss": 0.9284,
+      "step": 8566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696710304469673e-06,
+      "loss": 0.873,
+      "step": 8567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.694580864063465e-06,
+      "loss": 0.8858,
+      "step": 8568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6924518981734325e-06,
+      "loss": 0.9087,
+      "step": 8569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6903234069597947e-06,
+      "loss": 0.9043,
+      "step": 8570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688195390582735e-06,
+      "loss": 0.9092,
+      "step": 8571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6860678492023916e-06,
+      "loss": 0.91,
+      "step": 8572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6839407829788752e-06,
+      "loss": 0.9369,
+      "step": 8573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6818141920722615e-06,
+      "loss": 0.9092,
+      "step": 8574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.679688076642586e-06,
+      "loss": 0.9239,
+      "step": 8575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677562436849853e-06,
+      "loss": 0.9112,
+      "step": 8576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675437272854023e-06,
+      "loss": 2.5474,
+      "step": 8577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6733125848150286e-06,
+      "loss": 0.8754,
+      "step": 8578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.671188372892762e-06,
+      "loss": 0.9085,
+      "step": 8579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669064637247085e-06,
+      "loss": 0.8828,
+      "step": 8580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6669413780378204e-06,
+      "loss": 0.8627,
+      "step": 8581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6648185954247483e-06,
+      "loss": 0.8421,
+      "step": 8582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662696289567623e-06,
+      "loss": 0.8786,
+      "step": 8583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660574460626164e-06,
+      "loss": 0.8958,
+      "step": 8584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6584531087600426e-06,
+      "loss": 0.8874,
+      "step": 8585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6563322341289043e-06,
+      "loss": 0.8903,
+      "step": 8586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6542118368923562e-06,
+      "loss": 0.9112,
+      "step": 8587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65209191720997e-06,
+      "loss": 0.8868,
+      "step": 8588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6499724752412845e-06,
+      "loss": 0.8672,
+      "step": 8589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6478535111457924e-06,
+      "loss": 0.9286,
+      "step": 8590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6457350250829592e-06,
+      "loss": 0.9119,
+      "step": 8591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6436170172122164e-06,
+      "loss": 0.8954,
+      "step": 8592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6414994876929476e-06,
+      "loss": 0.9419,
+      "step": 8593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63938243668451e-06,
+      "loss": 0.8797,
+      "step": 8594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6372658643462312e-06,
+      "loss": 0.9045,
+      "step": 8595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.635149770837385e-06,
+      "loss": 0.9101,
+      "step": 8596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6330341563172244e-06,
+      "loss": 0.8863,
+      "step": 8597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630919020944956e-06,
+      "loss": 0.892,
+      "step": 8598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6288043648797577e-06,
+      "loss": 2.4509,
+      "step": 8599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626690188280767e-06,
+      "loss": 0.9238,
+      "step": 8600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624576491307088e-06,
+      "loss": 0.8851,
+      "step": 8601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6224632741177914e-06,
+      "loss": 0.9283,
+      "step": 8602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6203505368719004e-06,
+      "loss": 0.9058,
+      "step": 8603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618238279728413e-06,
+      "loss": 0.9011,
+      "step": 8604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616126502846292e-06,
+      "loss": 0.8879,
+      "step": 8605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6140152063844525e-06,
+      "loss": 0.8907,
+      "step": 8606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6119043905017837e-06,
+      "loss": 0.866,
+      "step": 8607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6097940553571365e-06,
+      "loss": 0.9164,
+      "step": 8608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607684201109325e-06,
+      "loss": 0.9371,
+      "step": 8609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.605574827917129e-06,
+      "loss": 0.9179,
+      "step": 8610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603465935939283e-06,
+      "loss": 0.9336,
+      "step": 8611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601357525334498e-06,
+      "loss": 0.9214,
+      "step": 8612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5992495962614436e-06,
+      "loss": 0.8982,
+      "step": 8613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5971421488787482e-06,
+      "loss": 0.8742,
+      "step": 8614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.595035183345007e-06,
+      "loss": 0.8537,
+      "step": 8615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.592928699818792e-06,
+      "loss": 0.8344,
+      "step": 8616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590822698458615e-06,
+      "loss": 0.8748,
+      "step": 8617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588717179422971e-06,
+      "loss": 0.9124,
+      "step": 8618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.586612142870306e-06,
+      "loss": 0.9251,
+      "step": 8619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584507588959039e-06,
+      "loss": 0.8781,
+      "step": 8620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.582403517847546e-06,
+      "loss": 0.9066,
+      "step": 8621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580299929694171e-06,
+      "loss": 0.8886,
+      "step": 8622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5781968246572207e-06,
+      "loss": 0.8771,
+      "step": 8623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5760942028949673e-06,
+      "loss": 0.8874,
+      "step": 8624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573992064565639e-06,
+      "loss": 0.8922,
+      "step": 8625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5718904098274376e-06,
+      "loss": 0.9137,
+      "step": 8626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569789238838518e-06,
+      "loss": 2.6168,
+      "step": 8627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567688551757008e-06,
+      "loss": 0.8912,
+      "step": 8628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565588348740995e-06,
+      "loss": 0.8928,
+      "step": 8629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5634886299485304e-06,
+      "loss": 0.8891,
+      "step": 8630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.56138939553763e-06,
+      "loss": 0.8625,
+      "step": 8631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5592906456662746e-06,
+      "loss": 0.8819,
+      "step": 8632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5571923804924e-06,
+      "loss": 0.8784,
+      "step": 8633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5550946001739183e-06,
+      "loss": 0.9265,
+      "step": 8634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5529973048686883e-06,
+      "loss": 0.902,
+      "step": 8635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550900494734554e-06,
+      "loss": 0.8973,
+      "step": 8636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5488041699293095e-06,
+      "loss": 0.8752,
+      "step": 8637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5467083306107084e-06,
+      "loss": 0.893,
+      "step": 8638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5446129769364813e-06,
+      "loss": 0.8961,
+      "step": 8639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5425181090643066e-06,
+      "loss": 0.8844,
+      "step": 8640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5404237271518393e-06,
+      "loss": 0.8814,
+      "step": 8641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5383298313566906e-06,
+      "loss": 0.9174,
+      "step": 8642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5362364218364387e-06,
+      "loss": 0.9226,
+      "step": 8643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5341434987486236e-06,
+      "loss": 0.9298,
+      "step": 8644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5320510622507507e-06,
+      "loss": 0.906,
+      "step": 8645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529959112500282e-06,
+      "loss": 0.9209,
+      "step": 8646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527867649654655e-06,
+      "loss": 0.9065,
+      "step": 8647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525776673871254e-06,
+      "loss": 0.8869,
+      "step": 8648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523686185307442e-06,
+      "loss": 0.8655,
+      "step": 8649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.521596184120537e-06,
+      "loss": 0.8354,
+      "step": 8650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519506670467825e-06,
+      "loss": 0.9091,
+      "step": 8651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5174176445065524e-06,
+      "loss": 0.8944,
+      "step": 8652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515329106393931e-06,
+      "loss": 0.895,
+      "step": 8653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.51324105628713e-06,
+      "loss": 0.8826,
+      "step": 8654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511153494343288e-06,
+      "loss": 0.9335,
+      "step": 8655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.509066420719506e-06,
+      "loss": 0.9031,
+      "step": 8656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5069798355728457e-06,
+      "loss": 2.4819,
+      "step": 8657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.50489373906034e-06,
+      "loss": 0.8755,
+      "step": 8658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5028081313389696e-06,
+      "loss": 0.8771,
+      "step": 8659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.500723012565691e-06,
+      "loss": 0.9327,
+      "step": 8660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4986383828974246e-06,
+      "loss": 0.905,
+      "step": 8661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4965542424910436e-06,
+      "loss": 0.9393,
+      "step": 8662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494470591503393e-06,
+      "loss": 0.9281,
+      "step": 8663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4923874300912773e-06,
+      "loss": 0.8845,
+      "step": 8664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490304758411469e-06,
+      "loss": 0.9041,
+      "step": 8665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4882225766207e-06,
+      "loss": 0.8911,
+      "step": 8666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.486140884875662e-06,
+      "loss": 0.8839,
+      "step": 8667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4840596833330133e-06,
+      "loss": 0.8605,
+      "step": 8668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4819789721493813e-06,
+      "loss": 0.9104,
+      "step": 8669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479898751481339e-06,
+      "loss": 0.9459,
+      "step": 8670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.477819021485448e-06,
+      "loss": 0.881,
+      "step": 8671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4757397823182084e-06,
+      "loss": 0.9194,
+      "step": 8672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473661034136099e-06,
+      "loss": 0.9204,
+      "step": 8673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4715827770955568e-06,
+      "loss": 0.8619,
+      "step": 8674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469505011352978e-06,
+      "loss": 0.8942,
+      "step": 8675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467427737064728e-06,
+      "loss": 0.9025,
+      "step": 8676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4653509543871322e-06,
+      "loss": 0.8782,
+      "step": 8677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4632746634764802e-06,
+      "loss": 0.932,
+      "step": 8678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461198864489026e-06,
+      "loss": 0.8746,
+      "step": 8679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459123557580978e-06,
+      "loss": 0.883,
+      "step": 8680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4570487429085187e-06,
+      "loss": 0.8986,
+      "step": 8681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4549744206277915e-06,
+      "loss": 0.9033,
+      "step": 8682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452900590894893e-06,
+      "loss": 0.8951,
+      "step": 8683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450827253865894e-06,
+      "loss": 0.8924,
+      "step": 8684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4487544096968227e-06,
+      "loss": 0.9449,
+      "step": 8685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4466820585436735e-06,
+      "loss": 0.9358,
+      "step": 8686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444610200562404e-06,
+      "loss": 0.9048,
+      "step": 8687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442538835908926e-06,
+      "loss": 0.9031,
+      "step": 8688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4404679647391237e-06,
+      "loss": 0.9002,
+      "step": 8689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4383975872088447e-06,
+      "loss": 2.478,
+      "step": 8690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4363277034738864e-06,
+      "loss": 0.9049,
+      "step": 8691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434258313690031e-06,
+      "loss": 0.8791,
+      "step": 8692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4321894180130023e-06,
+      "loss": 0.9128,
+      "step": 8693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4301210165984965e-06,
+      "loss": 0.9482,
+      "step": 8694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4280531096021772e-06,
+      "loss": 0.877,
+      "step": 8695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.425985697179659e-06,
+      "loss": 0.8856,
+      "step": 8696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423918779486528e-06,
+      "loss": 0.9171,
+      "step": 8697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42185235667833e-06,
+      "loss": 0.9067,
+      "step": 8698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4197864289105763e-06,
+      "loss": 0.849,
+      "step": 8699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.41772099633874e-06,
+      "loss": 0.9419,
+      "step": 8700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415656059118251e-06,
+      "loss": 0.8809,
+      "step": 8701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.413591617404508e-06,
+      "loss": 0.8943,
+      "step": 8702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411527671352877e-06,
+      "loss": 0.8996,
+      "step": 8703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4094642211186725e-06,
+      "loss": 0.9848,
+      "step": 8704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407401266857184e-06,
+      "loss": 0.9268,
+      "step": 8705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4053388087236605e-06,
+      "loss": 0.9228,
+      "step": 8706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4032768468733115e-06,
+      "loss": 0.9193,
+      "step": 8707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4012153814613136e-06,
+      "loss": 0.8918,
+      "step": 8708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399154412642798e-06,
+      "loss": 0.8654,
+      "step": 8709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3970939405728654e-06,
+      "loss": 0.89,
+      "step": 8710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395033965406578e-06,
+      "loss": 0.9141,
+      "step": 8711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39297448729896e-06,
+      "loss": 0.9035,
+      "step": 8712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3909155064050005e-06,
+      "loss": 0.9006,
+      "step": 8713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3888570228796437e-06,
+      "loss": 0.8998,
+      "step": 8714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3867990368778025e-06,
+      "loss": 0.9111,
+      "step": 8715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3847415485543554e-06,
+      "loss": 0.8796,
+      "step": 8716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382684558064133e-06,
+      "loss": 2.5699,
+      "step": 8717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3806280655619394e-06,
+      "loss": 0.9158,
+      "step": 8718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3785720712025338e-06,
+      "loss": 0.8225,
+      "step": 8719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3765165751406425e-06,
+      "loss": 0.9075,
+      "step": 8720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374461577530954e-06,
+      "loss": 0.8621,
+      "step": 8721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3724070785281138e-06,
+      "loss": 0.9342,
+      "step": 8722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3703530782867357e-06,
+      "loss": 0.918,
+      "step": 8723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368299576961397e-06,
+      "loss": 0.9105,
+      "step": 8724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3662465747066277e-06,
+      "loss": 0.8978,
+      "step": 8725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3641940716769285e-06,
+      "loss": 0.8918,
+      "step": 8726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.36214206802677e-06,
+      "loss": 0.8891,
+      "step": 8727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3600905639105676e-06,
+      "loss": 0.9029,
+      "step": 8728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.358039559482712e-06,
+      "loss": 0.887,
+      "step": 8729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3559890548975483e-06,
+      "loss": 0.9057,
+      "step": 8730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3539390503093896e-06,
+      "loss": 0.8636,
+      "step": 8731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3518895458725097e-06,
+      "loss": 0.8924,
+      "step": 8732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.349840541741145e-06,
+      "loss": 0.8864,
+      "step": 8733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347792038069494e-06,
+      "loss": 0.9318,
+      "step": 8734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3457440350117197e-06,
+      "loss": 0.8782,
+      "step": 8735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3436965327219406e-06,
+      "loss": 0.922,
+      "step": 8736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3416495313542462e-06,
+      "loss": 0.8849,
+      "step": 8737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3396030310626803e-06,
+      "loss": 0.8739,
+      "step": 8738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337557032001254e-06,
+      "loss": 0.9093,
+      "step": 8739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3355115343239408e-06,
+      "loss": 0.8839,
+      "step": 8740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3334665381846742e-06,
+      "loss": 0.8718,
+      "step": 8741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3314220437373557e-06,
+      "loss": 0.8973,
+      "step": 8742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3293780511358365e-06,
+      "loss": 0.886,
+      "step": 8743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3273345605339415e-06,
+      "loss": 0.972,
+      "step": 8744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3252915720854572e-06,
+      "loss": 0.934,
+      "step": 8745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3232490859441235e-06,
+      "loss": 0.9064,
+      "step": 8746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3212071022636473e-06,
+      "loss": 0.8967,
+      "step": 8747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31916562119771e-06,
+      "loss": 0.8894,
+      "step": 8748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3171246428999325e-06,
+      "loss": 0.9113,
+      "step": 8749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3150841675239166e-06,
+      "loss": 0.9034,
+      "step": 8750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3130441952232116e-06,
+      "loss": 0.8794,
+      "step": 8751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311004726151341e-06,
+      "loss": 0.9313,
+      "step": 8752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308965760461784e-06,
+      "loss": 0.8703,
+      "step": 8753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3069272983079848e-06,
+      "loss": 0.892,
+      "step": 8754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304889339843347e-06,
+      "loss": 0.9178,
+      "step": 8755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.302851885221242e-06,
+      "loss": 0.9252,
+      "step": 8756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3008149345949936e-06,
+      "loss": 0.8857,
+      "step": 8757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2987784881178976e-06,
+      "loss": 0.8942,
+      "step": 8758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2967425459432023e-06,
+      "loss": 0.8605,
+      "step": 8759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2947071082241257e-06,
+      "loss": 0.9122,
+      "step": 8760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2926721751138444e-06,
+      "loss": 0.9585,
+      "step": 8761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2906377467655003e-06,
+      "loss": 0.8904,
+      "step": 8762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2886038233321925e-06,
+      "loss": 0.9288,
+      "step": 8763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.28657040496699e-06,
+      "loss": 0.8888,
+      "step": 8764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284537491822909e-06,
+      "loss": 0.9015,
+      "step": 8765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282505084052946e-06,
+      "loss": 0.9067,
+      "step": 8766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.28047318181004e-06,
+      "loss": 0.922,
+      "step": 8767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2784417852471105e-06,
+      "loss": 0.8906,
+      "step": 8768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2764108945170336e-06,
+      "loss": 0.9127,
+      "step": 8769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2743805097726366e-06,
+      "loss": 0.9195,
+      "step": 8770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2723506311667198e-06,
+      "loss": 0.8959,
+      "step": 8771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270321258852046e-06,
+      "loss": 0.9116,
+      "step": 8772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2682923929813292e-06,
+      "loss": 0.8605,
+      "step": 8773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.266264033707256e-06,
+      "loss": 0.9039,
+      "step": 8774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264236181182472e-06,
+      "loss": 0.8735,
+      "step": 8775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2622088355595816e-06,
+      "loss": 0.8472,
+      "step": 8776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2601819969911586e-06,
+      "loss": 0.927,
+      "step": 8777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2581556656297262e-06,
+      "loss": 0.8987,
+      "step": 8778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2561298416277833e-06,
+      "loss": 0.9367,
+      "step": 8779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2541045251377755e-06,
+      "loss": 0.8797,
+      "step": 8780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2520797163121243e-06,
+      "loss": 0.8743,
+      "step": 8781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250055415303207e-06,
+      "loss": 0.8889,
+      "step": 8782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2480316222633614e-06,
+      "loss": 0.9029,
+      "step": 8783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24600833734489e-06,
+      "loss": 0.8342,
+      "step": 8784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243985560700059e-06,
+      "loss": 0.9024,
+      "step": 8785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241963292481085e-06,
+      "loss": 0.928,
+      "step": 8786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2399415328401597e-06,
+      "loss": 0.9174,
+      "step": 8787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23792028192943e-06,
+      "loss": 0.9,
+      "step": 8788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2358995399010062e-06,
+      "loss": 0.8535,
+      "step": 8789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233879306906963e-06,
+      "loss": 0.9042,
+      "step": 8790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231859583099326e-06,
+      "loss": 0.881,
+      "step": 8791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229840368630095e-06,
+      "loss": 0.8835,
+      "step": 8792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2278216636512296e-06,
+      "loss": 2.6168,
+      "step": 8793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22580346831464e-06,
+      "loss": 0.9422,
+      "step": 8794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223785782772212e-06,
+      "loss": 2.5847,
+      "step": 8795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2217686071757847e-06,
+      "loss": 0.9268,
+      "step": 8796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2197519416771615e-06,
+      "loss": 0.8659,
+      "step": 8797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217735786428112e-06,
+      "loss": 0.8945,
+      "step": 8798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2157201415803553e-06,
+      "loss": 0.9134,
+      "step": 8799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2137050072855825e-06,
+      "loss": 0.9291,
+      "step": 8800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211690383695445e-06,
+      "loss": 0.9207,
+      "step": 8801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209676270961547e-06,
+      "loss": 0.904,
+      "step": 8802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2076626692354716e-06,
+      "loss": 0.9124,
+      "step": 8803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2056495786687447e-06,
+      "loss": 0.9302,
+      "step": 8804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203636999412865e-06,
+      "loss": 0.9023,
+      "step": 8805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2016249316192915e-06,
+      "loss": 0.9051,
+      "step": 8806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1996133754394387e-06,
+      "loss": 0.8562,
+      "step": 8807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1976023310246895e-06,
+      "loss": 0.8488,
+      "step": 8808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1955917985263853e-06,
+      "loss": 0.9303,
+      "step": 8809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1935817780958278e-06,
+      "loss": 0.885,
+      "step": 8810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1915722698842877e-06,
+      "loss": 0.8956,
+      "step": 8811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189563274042984e-06,
+      "loss": 0.9032,
+      "step": 8812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187554790723105e-06,
+      "loss": 0.8974,
+      "step": 8813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1855468200758067e-06,
+      "loss": 0.8999,
+      "step": 8814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1835393622521905e-06,
+      "loss": 0.8886,
+      "step": 8815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1815324174033314e-06,
+      "loss": 0.8653,
+      "step": 8816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179525985680264e-06,
+      "loss": 0.8951,
+      "step": 8817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1775200672339824e-06,
+      "loss": 0.859,
+      "step": 8818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1755146622154466e-06,
+      "loss": 0.9048,
+      "step": 8819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1735097707755657e-06,
+      "loss": 0.9051,
+      "step": 8820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1715053930652216e-06,
+      "loss": 0.8626,
+      "step": 8821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1695015292352603e-06,
+      "loss": 2.4448,
+      "step": 8822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16749817943647e-06,
+      "loss": 0.8508,
+      "step": 8823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165495343819629e-06,
+      "loss": 0.923,
+      "step": 8824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1634930225354508e-06,
+      "loss": 0.8698,
+      "step": 8825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1614912157346233e-06,
+      "loss": 0.9396,
+      "step": 8826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1594899235677966e-06,
+      "loss": 0.9271,
+      "step": 8827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157489146185572e-06,
+      "loss": 0.9302,
+      "step": 8828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1554888837385222e-06,
+      "loss": 0.904,
+      "step": 8829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153489136377177e-06,
+      "loss": 0.9246,
+      "step": 8830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1514899042520288e-06,
+      "loss": 0.8686,
+      "step": 8831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1494911875135323e-06,
+      "loss": 0.895,
+      "step": 8832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.147492986312096e-06,
+      "loss": 0.9034,
+      "step": 8833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145495300798099e-06,
+      "loss": 0.8864,
+      "step": 8834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1434981311218793e-06,
+      "loss": 0.9059,
+      "step": 8835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1415014774337292e-06,
+      "loss": 0.8714,
+      "step": 8836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1395053398839104e-06,
+      "loss": 0.8678,
+      "step": 8837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1375097186226433e-06,
+      "loss": 0.8981,
+      "step": 8838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135514613800108e-06,
+      "loss": 0.874,
+      "step": 8839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.133520025566451e-06,
+      "loss": 2.4866,
+      "step": 8840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1315259540717677e-06,
+      "loss": 2.4784,
+      "step": 8841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1295323994661275e-06,
+      "loss": 0.8913,
+      "step": 8842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.127539361899554e-06,
+      "loss": 0.913,
+      "step": 8843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1255468415220357e-06,
+      "loss": 0.8775,
+      "step": 8844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1235548384835236e-06,
+      "loss": 0.8521,
+      "step": 8845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1215633529339173e-06,
+      "loss": 0.93,
+      "step": 8846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119572385023093e-06,
+      "loss": 0.8829,
+      "step": 8847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.117581934900883e-06,
+      "loss": 0.9024,
+      "step": 8848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115592002717074e-06,
+      "loss": 0.8729,
+      "step": 8849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1136025886214204e-06,
+      "loss": 0.8777,
+      "step": 8850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111613692763639e-06,
+      "loss": 0.9343,
+      "step": 8851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1096253152934007e-06,
+      "loss": 0.8715,
+      "step": 8852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1076374563603486e-06,
+      "loss": 0.8921,
+      "step": 8853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105650116114071e-06,
+      "loss": 0.9091,
+      "step": 8854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103663294704129e-06,
+      "loss": 0.878,
+      "step": 8855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101676992280045e-06,
+      "loss": 0.8702,
+      "step": 8856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0996912089912933e-06,
+      "loss": 0.9133,
+      "step": 8857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0977059449873126e-06,
+      "loss": 0.8804,
+      "step": 8858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095721200417516e-06,
+      "loss": 0.9224,
+      "step": 8859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0937369754312563e-06,
+      "loss": 0.8897,
+      "step": 8860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091753270177863e-06,
+      "loss": 0.8463,
+      "step": 8861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0897700848066125e-06,
+      "loss": 0.94,
+      "step": 8862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0877874194667555e-06,
+      "loss": 0.9419,
+      "step": 8863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085805274307497e-06,
+      "loss": 0.8899,
+      "step": 8864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0838236494780037e-06,
+      "loss": 0.8988,
+      "step": 8865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081842545127404e-06,
+      "loss": 0.9037,
+      "step": 8866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0798619614047885e-06,
+      "loss": 0.8605,
+      "step": 8867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077881898459202e-06,
+      "loss": 0.8784,
+      "step": 8868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0759023564396594e-06,
+      "loss": 0.9289,
+      "step": 8869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073923335495127e-06,
+      "loss": 0.87,
+      "step": 8870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0719448357745387e-06,
+      "loss": 0.928,
+      "step": 8871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.069966857426787e-06,
+      "loss": 0.9372,
+      "step": 8872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.067989400600726e-06,
+      "loss": 0.8981,
+      "step": 8873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0660124654451685e-06,
+      "loss": 0.9124,
+      "step": 8874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0640360521088943e-06,
+      "loss": 0.913,
+      "step": 8875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062060160740631e-06,
+      "loss": 0.8852,
+      "step": 8876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060084791489082e-06,
+      "loss": 0.9294,
+      "step": 8877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0581099445028984e-06,
+      "loss": 0.925,
+      "step": 8878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.056135619930697e-06,
+      "loss": 0.8698,
+      "step": 8879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0541618179210664e-06,
+      "loss": 0.9033,
+      "step": 8880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.052188538622535e-06,
+      "loss": 0.9125,
+      "step": 8881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.05021578218361e-06,
+      "loss": 0.8743,
+      "step": 8882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.048243548752745e-06,
+      "loss": 0.8875,
+      "step": 8883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046271838478364e-06,
+      "loss": 0.956,
+      "step": 8884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04430065150885e-06,
+      "loss": 0.9105,
+      "step": 8885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0423299879925427e-06,
+      "loss": 0.8824,
+      "step": 8886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0403598480777472e-06,
+      "loss": 0.9098,
+      "step": 8887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03839023191273e-06,
+      "loss": 0.8969,
+      "step": 8888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0364211396457067e-06,
+      "loss": 0.89,
+      "step": 8889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.034452571424871e-06,
+      "loss": 0.9263,
+      "step": 8890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0324845273983592e-06,
+      "loss": 0.9522,
+      "step": 8891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.030517007714283e-06,
+      "loss": 0.8577,
+      "step": 8892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0285500125207055e-06,
+      "loss": 0.931,
+      "step": 8893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0265835419656564e-06,
+      "loss": 0.9344,
+      "step": 8894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024617596197121e-06,
+      "loss": 0.9115,
+      "step": 8895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022652175363051e-06,
+      "loss": 0.8817,
+      "step": 8896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0206872796113497e-06,
+      "loss": 0.9238,
+      "step": 8897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0187229090898897e-06,
+      "loss": 0.9123,
+      "step": 8898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0167590639464927e-06,
+      "loss": 0.9002,
+      "step": 8899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0147957443289564e-06,
+      "loss": 0.9596,
+      "step": 8900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012832950385033e-06,
+      "loss": 0.9008,
+      "step": 8901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0108706822624256e-06,
+      "loss": 0.9198,
+      "step": 8902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0089089401088077e-06,
+      "loss": 2.6241,
+      "step": 8903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006947724071816e-06,
+      "loss": 0.8843,
+      "step": 8904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0049870342990338e-06,
+      "loss": 0.8936,
+      "step": 8905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0030268709380184e-06,
+      "loss": 0.9141,
+      "step": 8906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001067234136281e-06,
+      "loss": 0.9153,
+      "step": 8907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9991081240412954e-06,
+      "loss": 0.9001,
+      "step": 8908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997149540800498e-06,
+      "loss": 0.8639,
+      "step": 8909
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9951914845612772e-06,
+      "loss": 0.9121,
+      "step": 8910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9932339554709923e-06,
+      "loss": 0.8994,
+      "step": 8911
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9912769536769515e-06,
+      "loss": 0.9106,
+      "step": 8912
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989320479326433e-06,
+      "loss": 0.8959,
+      "step": 8913
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987364532566671e-06,
+      "loss": 0.8675,
+      "step": 8914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9854091135448626e-06,
+      "loss": 0.8986,
+      "step": 8915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.983454222408162e-06,
+      "loss": 0.9024,
+      "step": 8916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981499859303688e-06,
+      "loss": 0.9038,
+      "step": 8917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9795460243785124e-06,
+      "loss": 2.4584,
+      "step": 8918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9775927177796725e-06,
+      "loss": 0.896,
+      "step": 8919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975639939654167e-06,
+      "loss": 0.8782,
+      "step": 8920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9736876901489508e-06,
+      "loss": 0.8514,
+      "step": 8921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.971735969410946e-06,
+      "loss": 0.9089,
+      "step": 8922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9697847775870227e-06,
+      "loss": 0.8803,
+      "step": 8923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.967834114824021e-06,
+      "loss": 0.87,
+      "step": 8924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9658839812687425e-06,
+      "loss": 0.9423,
+      "step": 8925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9639343770679386e-06,
+      "loss": 2.5852,
+      "step": 8926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961985302368331e-06,
+      "loss": 0.8642,
+      "step": 8927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.960036757316598e-06,
+      "loss": 0.8933,
+      "step": 8928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9580887420593763e-06,
+      "loss": 0.8738,
+      "step": 8929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.956141256743269e-06,
+      "loss": 0.8659,
+      "step": 8930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.954194301514828e-06,
+      "loss": 0.8952,
+      "step": 8931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9522478765205753e-06,
+      "loss": 0.8733,
+      "step": 8932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.950301981906992e-06,
+      "loss": 0.8787,
+      "step": 8933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948356617820509e-06,
+      "loss": 0.8981,
+      "step": 8934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.946411784407537e-06,
+      "loss": 0.9625,
+      "step": 8935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9444674818144247e-06,
+      "loss": 0.883,
+      "step": 8936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942523710187496e-06,
+      "loss": 0.8834,
+      "step": 8937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.940580469673032e-06,
+      "loss": 0.9377,
+      "step": 8938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9386377604172657e-06,
+      "loss": 0.8782,
+      "step": 8939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9366955825664e-06,
+      "loss": 2.5606,
+      "step": 8940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9347539362665935e-06,
+      "loss": 0.9308,
+      "step": 8941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932812821663966e-06,
+      "loss": 0.882,
+      "step": 8942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9308722389045986e-06,
+      "loss": 0.922,
+      "step": 8943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.8967,
+      "step": 8944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9269926694997487e-06,
+      "loss": 2.5026,
+      "step": 8945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9250536831462296e-06,
+      "loss": 0.9118,
+      "step": 8946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9231152292198805e-06,
+      "loss": 0.8962,
+      "step": 8947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.921177307866585e-06,
+      "loss": 0.8517,
+      "step": 8948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9192399192321806e-06,
+      "loss": 0.9022,
+      "step": 8949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917303063462468e-06,
+      "loss": 0.8763,
+      "step": 8950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9153667407032073e-06,
+      "loss": 0.8963,
+      "step": 8951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91343095110011e-06,
+      "loss": 0.8954,
+      "step": 8952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9114956947988595e-06,
+      "loss": 0.9258,
+      "step": 8953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909560971945097e-06,
+      "loss": 0.9276,
+      "step": 8954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9076267826844105e-06,
+      "loss": 0.8632,
+      "step": 8955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.90569312716237e-06,
+      "loss": 0.9038,
+      "step": 8956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903760005524485e-06,
+      "loss": 0.9115,
+      "step": 8957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9018274179162365e-06,
+      "loss": 0.8826,
+      "step": 8958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8998953644830643e-06,
+      "loss": 0.9162,
+      "step": 8959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8979638453703597e-06,
+      "loss": 0.8888,
+      "step": 8960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8960328607234823e-06,
+      "loss": 0.896,
+      "step": 8961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89410241068775e-06,
+      "loss": 0.9239,
+      "step": 8962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8921724954084383e-06,
+      "loss": 0.8671,
+      "step": 8963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8902431150307876e-06,
+      "loss": 0.9049,
+      "step": 8964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8883142696999875e-06,
+      "loss": 0.9168,
+      "step": 8965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8863859595611966e-06,
+      "loss": 0.9063,
+      "step": 8966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884458184759533e-06,
+      "loss": 0.9018,
+      "step": 8967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8825309454400673e-06,
+      "loss": 0.9033,
+      "step": 8968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8806042417478365e-06,
+      "loss": 0.9106,
+      "step": 8969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8786780738278363e-06,
+      "loss": 0.8742,
+      "step": 8970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8767524418250193e-06,
+      "loss": 0.8968,
+      "step": 8971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8748273458843036e-06,
+      "loss": 0.9145,
+      "step": 8972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8729027861505565e-06,
+      "loss": 0.8498,
+      "step": 8973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8709787627686146e-06,
+      "loss": 2.5519,
+      "step": 8974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8690552758832714e-06,
+      "loss": 0.9239,
+      "step": 8975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867132325639278e-06,
+      "loss": 0.8766,
+      "step": 8976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.865209912181347e-06,
+      "loss": 0.8685,
+      "step": 8977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8632880356541547e-06,
+      "loss": 0.8932,
+      "step": 8978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.861366696202326e-06,
+      "loss": 2.4478,
+      "step": 8979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859445893970455e-06,
+      "loss": 0.8925,
+      "step": 8980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8575256291030905e-06,
+      "loss": 0.8654,
+      "step": 8981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8556059017447422e-06,
+      "loss": 0.9505,
+      "step": 8982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8536867120398825e-06,
+      "loss": 0.8564,
+      "step": 8983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851768060132939e-06,
+      "loss": 0.9137,
+      "step": 8984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849849946168303e-06,
+      "loss": 0.877,
+      "step": 8985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8479323702903173e-06,
+      "loss": 0.908,
+      "step": 8986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846015332643293e-06,
+      "loss": 0.8918,
+      "step": 8987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8440988333715e-06,
+      "loss": 0.9243,
+      "step": 8988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8421828726191593e-06,
+      "loss": 0.9032,
+      "step": 8989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8402674505304574e-06,
+      "loss": 0.9478,
+      "step": 8990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.838352567249547e-06,
+      "loss": 0.884,
+      "step": 8991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8364382229205266e-06,
+      "loss": 0.9093,
+      "step": 8992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8345244176874664e-06,
+      "loss": 0.9117,
+      "step": 8993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832611151694382e-06,
+      "loss": 0.8687,
+      "step": 8994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830698425085263e-06,
+      "loss": 0.9009,
+      "step": 8995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828786238004051e-06,
+      "loss": 0.9138,
+      "step": 8996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826874590594647e-06,
+      "loss": 0.9197,
+      "step": 8997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8249634830009134e-06,
+      "loss": 0.9081,
+      "step": 8998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823052915366675e-06,
+      "loss": 0.8753,
+      "step": 8999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.821142887835705e-06,
+      "loss": 0.8771,
+      "step": 9000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8192334005517497e-06,
+      "loss": 0.9599,
+      "step": 9001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8173244536585022e-06,
+      "loss": 0.8471,
+      "step": 9002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815416047299624e-06,
+      "loss": 0.8932,
+      "step": 9003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8135081816187315e-06,
+      "loss": 0.9135,
+      "step": 9004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811600856759403e-06,
+      "loss": 0.8579,
+      "step": 9005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8096940728651743e-06,
+      "loss": 0.9118,
+      "step": 9006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8077878300795446e-06,
+      "loss": 2.5561,
+      "step": 9007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8058821285459615e-06,
+      "loss": 0.8738,
+      "step": 9008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8039769684078464e-06,
+      "loss": 0.8899,
+      "step": 9009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8020723498085635e-06,
+      "loss": 0.9301,
+      "step": 9010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8001682728914535e-06,
+      "loss": 0.9174,
+      "step": 9011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7982647377998086e-06,
+      "loss": 0.9212,
+      "step": 9012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7963617446768744e-06,
+      "loss": 0.8768,
+      "step": 9013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7944592936658676e-06,
+      "loss": 0.9148,
+      "step": 9014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7925573849099507e-06,
+      "loss": 0.9357,
+      "step": 9015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790656018552256e-06,
+      "loss": 0.8906,
+      "step": 9016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7887551947358702e-06,
+      "loss": 0.9355,
+      "step": 9017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7868549136038424e-06,
+      "loss": 0.8735,
+      "step": 9018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7849551752991777e-06,
+      "loss": 0.8782,
+      "step": 9019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783055979964844e-06,
+      "loss": 0.898,
+      "step": 9020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.781157327743761e-06,
+      "loss": 0.8892,
+      "step": 9021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779259218778817e-06,
+      "loss": 0.9143,
+      "step": 9022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77736165321285e-06,
+      "loss": 0.8415,
+      "step": 9023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.775464631188665e-06,
+      "loss": 0.9197,
+      "step": 9024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7735681528490198e-06,
+      "loss": 0.87,
+      "step": 9025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7716722183366383e-06,
+      "loss": 0.8898,
+      "step": 9026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7697768277941985e-06,
+      "loss": 0.916,
+      "step": 9027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767881981364341e-06,
+      "loss": 0.8967,
+      "step": 9028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7659876791896576e-06,
+      "loss": 0.8876,
+      "step": 9029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7640939214127094e-06,
+      "loss": 0.9003,
+      "step": 9030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762200708176005e-06,
+      "loss": 0.8715,
+      "step": 9031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760308039622025e-06,
+      "loss": 0.9013,
+      "step": 9032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7584159158932047e-06,
+      "loss": 0.8816,
+      "step": 9033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7565243371319282e-06,
+      "loss": 0.8877,
+      "step": 9034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7546333034805528e-06,
+      "loss": 0.8969,
+      "step": 9035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7527428150813905e-06,
+      "loss": 0.8901,
+      "step": 9036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7508528720767034e-06,
+      "loss": 0.8843,
+      "step": 9037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7489634746087246e-06,
+      "loss": 0.9117,
+      "step": 9038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74707462281964e-06,
+      "loss": 0.9226,
+      "step": 9039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7451863168515967e-06,
+      "loss": 2.4752,
+      "step": 9040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743298556846702e-06,
+      "loss": 0.8898,
+      "step": 9041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7414113429470133e-06,
+      "loss": 0.927,
+      "step": 9042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739524675294557e-06,
+      "loss": 0.9085,
+      "step": 9043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7376385540313188e-06,
+      "loss": 0.9015,
+      "step": 9044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735752979299231e-06,
+      "loss": 0.8875,
+      "step": 9045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7338679512401987e-06,
+      "loss": 0.893,
+      "step": 9046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731983469996079e-06,
+      "loss": 0.8765,
+      "step": 9047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.730099535708689e-06,
+      "loss": 0.917,
+      "step": 9048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7282161485198077e-06,
+      "loss": 0.9173,
+      "step": 9049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7263333085711642e-06,
+      "loss": 0.9062,
+      "step": 9050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724451016004456e-06,
+      "loss": 0.8981,
+      "step": 9051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7225692709613325e-06,
+      "loss": 0.8928,
+      "step": 9052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7206880735834095e-06,
+      "loss": 0.9378,
+      "step": 9053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7188074240122565e-06,
+      "loss": 0.8926,
+      "step": 9054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7169273223893976e-06,
+      "loss": 0.86,
+      "step": 9055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715047768856325e-06,
+      "loss": 0.918,
+      "step": 9056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7131687635544847e-06,
+      "loss": 0.8524,
+      "step": 9057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7112903066252784e-06,
+      "loss": 0.8882,
+      "step": 9058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7094123982100707e-06,
+      "loss": 0.893,
+      "step": 9059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7075350384501863e-06,
+      "loss": 0.9006,
+      "step": 9060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705658227486906e-06,
+      "loss": 0.8845,
+      "step": 9061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703781965461473e-06,
+      "loss": 0.9034,
+      "step": 9062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7019062525150783e-06,
+      "loss": 0.9275,
+      "step": 9063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.700031088788884e-06,
+      "loss": 0.9317,
+      "step": 9064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6981564744240083e-06,
+      "loss": 0.9299,
+      "step": 9065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696282409561517e-06,
+      "loss": 0.9002,
+      "step": 9066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6944088943424553e-06,
+      "loss": 0.8406,
+      "step": 9067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692535928907806e-06,
+      "loss": 0.8829,
+      "step": 9068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6906635133985235e-06,
+      "loss": 0.8878,
+      "step": 9069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6887916479555187e-06,
+      "loss": 0.9109,
+      "step": 9070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6869203327196546e-06,
+      "loss": 0.8888,
+      "step": 9071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68504956783176e-06,
+      "loss": 0.9069,
+      "step": 9072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68317935343262e-06,
+      "loss": 0.899,
+      "step": 9073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6813096896629777e-06,
+      "loss": 0.8676,
+      "step": 9074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679440576663539e-06,
+      "loss": 0.9032,
+      "step": 9075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6775720145749574e-06,
+      "loss": 0.901,
+      "step": 9076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.675704003537857e-06,
+      "loss": 0.8957,
+      "step": 9077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.673836543692816e-06,
+      "loss": 0.9013,
+      "step": 9078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671969635180366e-06,
+      "loss": 0.9474,
+      "step": 9079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.670103278141005e-06,
+      "loss": 0.9204,
+      "step": 9080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6682374727151873e-06,
+      "loss": 0.8979,
+      "step": 9081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6663722190433216e-06,
+      "loss": 0.8528,
+      "step": 9082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.664507517265783e-06,
+      "loss": 0.893,
+      "step": 9083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6626433675228946e-06,
+      "loss": 0.9428,
+      "step": 9084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.660779769954945e-06,
+      "loss": 0.9146,
+      "step": 9085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6589167247021854e-06,
+      "loss": 0.908,
+      "step": 9086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6570542319048074e-06,
+      "loss": 0.8997,
+      "step": 9087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6551922917029883e-06,
+      "loss": 0.8937,
+      "step": 9088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6533309042368393e-06,
+      "loss": 0.8487,
+      "step": 9089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6514700696464412e-06,
+      "loss": 0.8954,
+      "step": 9090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6496097880718364e-06,
+      "loss": 0.8655,
+      "step": 9091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6477500596530138e-06,
+      "loss": 0.8374,
+      "step": 9092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6458908845299304e-06,
+      "loss": 0.9414,
+      "step": 9093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644032262842501e-06,
+      "loss": 0.8813,
+      "step": 9094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6421741947305944e-06,
+      "loss": 0.9196,
+      "step": 9095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640316680334045e-06,
+      "loss": 2.4612,
+      "step": 9096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6384597197926343e-06,
+      "loss": 2.5483,
+      "step": 9097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63660331324611e-06,
+      "loss": 0.9404,
+      "step": 9098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6347474608341807e-06,
+      "loss": 0.9114,
+      "step": 9099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632892162696502e-06,
+      "loss": 0.9068,
+      "step": 9100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6310374189726995e-06,
+      "loss": 0.9334,
+      "step": 9101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629183229802352e-06,
+      "loss": 0.8816,
+      "step": 9102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6273295953249965e-06,
+      "loss": 0.8977,
+      "step": 9103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625476515680132e-06,
+      "loss": 0.9202,
+      "step": 9104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.623623991007207e-06,
+      "loss": 0.9053,
+      "step": 9105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6217720214456364e-06,
+      "loss": 0.8874,
+      "step": 9106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6199206071347905e-06,
+      "loss": 0.852,
+      "step": 9107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618069748213998e-06,
+      "loss": 0.8534,
+      "step": 9108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6162194448225463e-06,
+      "loss": 0.8792,
+      "step": 9109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6143696970996835e-06,
+      "loss": 0.8954,
+      "step": 9110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.612520505184607e-06,
+      "loss": 0.924,
+      "step": 9111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6106718692164844e-06,
+      "loss": 0.8822,
+      "step": 9112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6088237893344282e-06,
+      "loss": 0.8978,
+      "step": 9113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60697626567752e-06,
+      "loss": 2.4787,
+      "step": 9114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6051292983847965e-06,
+      "loss": 0.915,
+      "step": 9115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60328288759525e-06,
+      "loss": 0.8799,
+      "step": 9116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6014370334478367e-06,
+      "loss": 0.886,
+      "step": 9117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5995917360814614e-06,
+      "loss": 0.9307,
+      "step": 9118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5977469956349956e-06,
+      "loss": 0.8838,
+      "step": 9119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5959028122472663e-06,
+      "loss": 0.8845,
+      "step": 9120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594059186057055e-06,
+      "loss": 0.8897,
+      "step": 9121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5922161172031035e-06,
+      "loss": 0.8962,
+      "step": 9122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590373605824121e-06,
+      "loss": 0.9216,
+      "step": 9123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5885316520587577e-06,
+      "loss": 0.886,
+      "step": 9124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5866902560456364e-06,
+      "loss": 0.9364,
+      "step": 9125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584849417923325e-06,
+      "loss": 0.8942,
+      "step": 9126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58300913783036e-06,
+      "loss": 0.8914,
+      "step": 9127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5811694159052327e-06,
+      "loss": 2.3563,
+      "step": 9128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5793302522863916e-06,
+      "loss": 0.8976,
+      "step": 9129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5774916471122422e-06,
+      "loss": 0.918,
+      "step": 9130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5756536005211552e-06,
+      "loss": 2.5285,
+      "step": 9131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573816112651444e-06,
+      "loss": 2.5355,
+      "step": 9132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5719791836413976e-06,
+      "loss": 0.861,
+      "step": 9133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570142813629247e-06,
+      "loss": 0.8765,
+      "step": 9134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568307002753193e-06,
+      "loss": 0.8877,
+      "step": 9135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5664717511513902e-06,
+      "loss": 0.9111,
+      "step": 9136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5646370589619507e-06,
+      "loss": 0.9003,
+      "step": 9137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5628029263229437e-06,
+      "loss": 0.8639,
+      "step": 9138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560969353372402e-06,
+      "loss": 0.8616,
+      "step": 9139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559136340248305e-06,
+      "loss": 0.9016,
+      "step": 9140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5573038870886025e-06,
+      "loss": 0.8643,
+      "step": 9141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5554719940311867e-06,
+      "loss": 0.8811,
+      "step": 9142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5536406612139288e-06,
+      "loss": 0.9611,
+      "step": 9143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5518098887746423e-06,
+      "loss": 0.8783,
+      "step": 9144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5499796768510997e-06,
+      "loss": 0.9439,
+      "step": 9145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5481500255810365e-06,
+      "loss": 0.8731,
+      "step": 9146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5463209351021457e-06,
+      "loss": 0.9402,
+      "step": 9147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5444924055520713e-06,
+      "loss": 0.8907,
+      "step": 9148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5426644370684215e-06,
+      "loss": 0.9069,
+      "step": 9149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540837029788762e-06,
+      "loss": 0.8963,
+      "step": 9150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.539010183850614e-06,
+      "loss": 0.8902,
+      "step": 9151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.53718389939146e-06,
+      "loss": 0.9302,
+      "step": 9152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5353581765487323e-06,
+      "loss": 0.9034,
+      "step": 9153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5335330154598313e-06,
+      "loss": 0.8913,
+      "step": 9154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5317084162621053e-06,
+      "loss": 0.8743,
+      "step": 9155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5298843790928686e-06,
+      "loss": 0.8665,
+      "step": 9156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5280609040893866e-06,
+      "loss": 0.8858,
+      "step": 9157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526237991388888e-06,
+      "loss": 0.9124,
+      "step": 9158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524415641128557e-06,
+      "loss": 0.8581,
+      "step": 9159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5225938534455353e-06,
+      "loss": 0.9144,
+      "step": 9160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5207726284769197e-06,
+      "loss": 0.8688,
+      "step": 9161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5189519663597696e-06,
+      "loss": 0.9192,
+      "step": 9162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.517131867231093e-06,
+      "loss": 0.8705,
+      "step": 9163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5153123312278703e-06,
+      "loss": 0.9346,
+      "step": 9164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5134933584870304e-06,
+      "loss": 0.9283,
+      "step": 9165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511674949145455e-06,
+      "loss": 0.898,
+      "step": 9166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509857103339992e-06,
+      "loss": 0.8904,
+      "step": 9167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5080398212074484e-06,
+      "loss": 0.8725,
+      "step": 9168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5062231028845764e-06,
+      "loss": 0.8864,
+      "step": 9169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504406948508097e-06,
+      "loss": 0.8814,
+      "step": 9170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5025913582146866e-06,
+      "loss": 0.901,
+      "step": 9171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5007763321409763e-06,
+      "loss": 0.9055,
+      "step": 9172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4989618704235606e-06,
+      "loss": 0.9034,
+      "step": 9173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4971479731989812e-06,
+      "loss": 0.8456,
+      "step": 9174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.495334640603746e-06,
+      "loss": 0.9404,
+      "step": 9175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4935218727743216e-06,
+      "loss": 0.9237,
+      "step": 9176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.491709669847121e-06,
+      "loss": 0.874,
+      "step": 9177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4898980319585277e-06,
+      "loss": 0.9386,
+      "step": 9178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488086959244874e-06,
+      "loss": 0.8734,
+      "step": 9179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4862764518424544e-06,
+      "loss": 0.8961,
+      "step": 9180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.484466509887522e-06,
+      "loss": 0.9253,
+      "step": 9181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4826571335162787e-06,
+      "loss": 0.8842,
+      "step": 9182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4808483228648916e-06,
+      "loss": 0.8835,
+      "step": 9183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4790400780694836e-06,
+      "loss": 0.878,
+      "step": 9184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477232399266136e-06,
+      "loss": 2.4793,
+      "step": 9185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4754252865908877e-06,
+      "loss": 0.8847,
+      "step": 9186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4736187401797295e-06,
+      "loss": 0.89,
+      "step": 9187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4718127601686137e-06,
+      "loss": 0.8999,
+      "step": 9188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4700073466934547e-06,
+      "loss": 0.8926,
+      "step": 9189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4682024998901133e-06,
+      "loss": 0.879,
+      "step": 9190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466398219894417e-06,
+      "loss": 0.8551,
+      "step": 9191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4645945068421463e-06,
+      "loss": 0.8962,
+      "step": 9192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46279136086904e-06,
+      "loss": 0.9016,
+      "step": 9193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.460988782110799e-06,
+      "loss": 0.9021,
+      "step": 9194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45918677070307e-06,
+      "loss": 0.8952,
+      "step": 9195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4573853267814683e-06,
+      "loss": 0.9287,
+      "step": 9196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4555844504815628e-06,
+      "loss": 0.8948,
+      "step": 9197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453784141938871e-06,
+      "loss": 0.8722,
+      "step": 9198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451984401288887e-06,
+      "loss": 0.8682,
+      "step": 9199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4501852286670436e-06,
+      "loss": 0.9159,
+      "step": 9200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4483866242087385e-06,
+      "loss": 0.8714,
+      "step": 9201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4465885880493313e-06,
+      "loss": 0.9037,
+      "step": 9202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444791120324127e-06,
+      "loss": 0.876,
+      "step": 9203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4429942211683976e-06,
+      "loss": 0.9424,
+      "step": 9204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4411978907173706e-06,
+      "loss": 0.8889,
+      "step": 9205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4394021291062263e-06,
+      "loss": 0.9174,
+      "step": 9206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4376069364701104e-06,
+      "loss": 0.8432,
+      "step": 9207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4358123129441146e-06,
+      "loss": 0.9291,
+      "step": 9208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4340182586632965e-06,
+      "loss": 0.893,
+      "step": 9209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4322247737626716e-06,
+      "loss": 0.8943,
+      "step": 9210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4304318583772026e-06,
+      "loss": 0.8367,
+      "step": 9211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428639512641818e-06,
+      "loss": 0.9092,
+      "step": 9212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4268477366914035e-06,
+      "loss": 0.8722,
+      "step": 9213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4250565306607987e-06,
+      "loss": 0.9089,
+      "step": 9214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423265894684803e-06,
+      "loss": 0.9495,
+      "step": 9215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4214758288981675e-06,
+      "loss": 0.9114,
+      "step": 9216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419686333435606e-06,
+      "loss": 0.9393,
+      "step": 9217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4178974084317895e-06,
+      "loss": 0.8725,
+      "step": 9218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4161090540213372e-06,
+      "loss": 0.8602,
+      "step": 9219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414321270338843e-06,
+      "loss": 0.8768,
+      "step": 9220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412534057518838e-06,
+      "loss": 0.9311,
+      "step": 9221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4107474156958234e-06,
+      "loss": 0.9108,
+      "step": 9222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4089613450042546e-06,
+      "loss": 0.8978,
+      "step": 9223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4071758455785387e-06,
+      "loss": 0.886,
+      "step": 9224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4053909175530467e-06,
+      "loss": 0.9092,
+      "step": 9225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403606561062103e-06,
+      "loss": 0.9068,
+      "step": 9226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4018227762399903e-06,
+      "loss": 0.9128,
+      "step": 9227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400039563220952e-06,
+      "loss": 0.8535,
+      "step": 9228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398256922139176e-06,
+      "loss": 0.8439,
+      "step": 9229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3964748531288206e-06,
+      "loss": 0.8952,
+      "step": 9230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394693356323997e-06,
+      "loss": 0.8918,
+      "step": 9231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392912431858768e-06,
+      "loss": 0.8817,
+      "step": 9232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39113207986716e-06,
+      "loss": 0.8793,
+      "step": 9233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389352300483154e-06,
+      "loss": 0.9291,
+      "step": 9234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3875730938406873e-06,
+      "loss": 0.8529,
+      "step": 9235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3857944600736583e-06,
+      "loss": 0.876,
+      "step": 9236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3840163993159126e-06,
+      "loss": 0.8518,
+      "step": 9237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38223891170126e-06,
+      "loss": 0.9355,
+      "step": 9238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.380461997363469e-06,
+      "loss": 0.9281,
+      "step": 9239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378685656436259e-06,
+      "loss": 0.9154,
+      "step": 9240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3769098890533104e-06,
+      "loss": 0.8963,
+      "step": 9241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3751346953482624e-06,
+      "loss": 0.9249,
+      "step": 9242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3733600754547005e-06,
+      "loss": 0.8665,
+      "step": 9243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.371586029506181e-06,
+      "loss": 0.8683,
+      "step": 9244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.369812557636204e-06,
+      "loss": 0.9251,
+      "step": 9245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3680396599782362e-06,
+      "loss": 0.9195,
+      "step": 9246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.366267336665696e-06,
+      "loss": 0.9023,
+      "step": 9247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364495587831963e-06,
+      "loss": 0.9685,
+      "step": 9248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.362724413610368e-06,
+      "loss": 0.8776,
+      "step": 9249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3609538141342046e-06,
+      "loss": 0.9038,
+      "step": 9250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3591837895367154e-06,
+      "loss": 0.9092,
+      "step": 9251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3574143399511084e-06,
+      "loss": 0.8585,
+      "step": 9252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3556454655105385e-06,
+      "loss": 0.8408,
+      "step": 9253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.353877166348124e-06,
+      "loss": 0.8741,
+      "step": 9254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352109442596947e-06,
+      "loss": 0.9045,
+      "step": 9255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350342294390028e-06,
+      "loss": 0.9209,
+      "step": 9256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348575721860361e-06,
+      "loss": 0.874,
+      "step": 9257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346809725140885e-06,
+      "loss": 0.9225,
+      "step": 9258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3450443043645035e-06,
+      "loss": 0.8667,
+      "step": 9259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343279459664072e-06,
+      "loss": 0.8491,
+      "step": 9260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3415151911724043e-06,
+      "loss": 0.8845,
+      "step": 9261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3397514990222735e-06,
+      "loss": 0.9236,
+      "step": 9262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337988383346408e-06,
+      "loss": 0.9552,
+      "step": 9263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3362258442774866e-06,
+      "loss": 0.862,
+      "step": 9264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3344638819481534e-06,
+      "loss": 0.89,
+      "step": 9265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332702496491003e-06,
+      "loss": 0.9063,
+      "step": 9266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3309416880385883e-06,
+      "loss": 0.9028,
+      "step": 9267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3291814567234218e-06,
+      "loss": 0.8445,
+      "step": 9268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3274218026779694e-06,
+      "loss": 0.8765,
+      "step": 9269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325662726034655e-06,
+      "loss": 0.9272,
+      "step": 9270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3239042269258606e-06,
+      "loss": 0.8894,
+      "step": 9271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322146305483918e-06,
+      "loss": 0.919,
+      "step": 9272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3203889618411235e-06,
+      "loss": 0.8734,
+      "step": 9273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318632196129721e-06,
+      "loss": 0.9001,
+      "step": 9274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316876008481922e-06,
+      "loss": 0.9194,
+      "step": 9275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315120399029891e-06,
+      "loss": 0.9365,
+      "step": 9276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31336536790574e-06,
+      "loss": 0.8723,
+      "step": 9277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311610915241548e-06,
+      "loss": 0.8723,
+      "step": 9278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.309857041169349e-06,
+      "loss": 0.9229,
+      "step": 9279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3081037458211263e-06,
+      "loss": 0.9163,
+      "step": 9280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3063510293288273e-06,
+      "loss": 0.8725,
+      "step": 9281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3045988918243535e-06,
+      "loss": 0.8754,
+      "step": 9282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3028473334395617e-06,
+      "loss": 0.8405,
+      "step": 9283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3010963543062693e-06,
+      "loss": 0.9331,
+      "step": 9284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2993459545562412e-06,
+      "loss": 0.9179,
+      "step": 9285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297596134321207e-06,
+      "loss": 0.8936,
+      "step": 9286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2958468937328528e-06,
+      "loss": 0.909,
+      "step": 9287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2940982329228125e-06,
+      "loss": 0.9275,
+      "step": 9288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2923501520226843e-06,
+      "loss": 0.8721,
+      "step": 9289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2906026511640213e-06,
+      "loss": 0.8925,
+      "step": 9290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288855730478332e-06,
+      "loss": 0.899,
+      "step": 9291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287109390097084e-06,
+      "loss": 0.9235,
+      "step": 9292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2853636301516936e-06,
+      "loss": 0.8801,
+      "step": 9293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2836184507735438e-06,
+      "loss": 0.8495,
+      "step": 9294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28187385209396e-06,
+      "loss": 0.9054,
+      "step": 9295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280129834244241e-06,
+      "loss": 0.9233,
+      "step": 9296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.278386397355633e-06,
+      "loss": 0.8794,
+      "step": 9297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2766435415593346e-06,
+      "loss": 0.9031,
+      "step": 9298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274901266986507e-06,
+      "loss": 0.9337,
+      "step": 9299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2731595737682677e-06,
+      "loss": 0.9332,
+      "step": 9300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2714184620356828e-06,
+      "loss": 0.9146,
+      "step": 9301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2696779319197835e-06,
+      "loss": 0.9143,
+      "step": 9302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2679379835515546e-06,
+      "loss": 0.9101,
+      "step": 9303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.266198617061934e-06,
+      "loss": 0.8776,
+      "step": 9304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2644598325818245e-06,
+      "loss": 0.8973,
+      "step": 9305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262721630242071e-06,
+      "loss": 0.8807,
+      "step": 9306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2609840101734857e-06,
+      "loss": 0.8914,
+      "step": 9307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2592469725068358e-06,
+      "loss": 0.8656,
+      "step": 9308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2575105173728384e-06,
+      "loss": 0.8945,
+      "step": 9309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2557746449021722e-06,
+      "loss": 0.904,
+      "step": 9310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2540393552254726e-06,
+      "loss": 2.4601,
+      "step": 9311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252304648473327e-06,
+      "loss": 0.8991,
+      "step": 9312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2505705247762865e-06,
+      "loss": 0.9314,
+      "step": 9313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.248836984264846e-06,
+      "loss": 0.9148,
+      "step": 9314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247104027069467e-06,
+      "loss": 0.8671,
+      "step": 9315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2453716533205627e-06,
+      "loss": 0.8858,
+      "step": 9316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2436398631485046e-06,
+      "loss": 0.904,
+      "step": 9317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2419086566836222e-06,
+      "loss": 0.9055,
+      "step": 9318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240178034056191e-06,
+      "loss": 0.9077,
+      "step": 9319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238447995396453e-06,
+      "loss": 0.8657,
+      "step": 9320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2367185408346057e-06,
+      "loss": 0.926,
+      "step": 9321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2349896705007936e-06,
+      "loss": 0.9075,
+      "step": 9322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2332613845251274e-06,
+      "loss": 0.9072,
+      "step": 9323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2315336830376676e-06,
+      "loss": 0.8877,
+      "step": 9324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2298065661684343e-06,
+      "loss": 0.9172,
+      "step": 9325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.228080034047405e-06,
+      "loss": 0.9006,
+      "step": 9326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226354086804505e-06,
+      "loss": 0.8863,
+      "step": 9327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224628724569623e-06,
+      "loss": 0.9184,
+      "step": 9328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2229039474726054e-06,
+      "loss": 2.4859,
+      "step": 9329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2211797556432414e-06,
+      "loss": 0.9015,
+      "step": 9330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2194561492112963e-06,
+      "loss": 0.9393,
+      "step": 9331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217733128306474e-06,
+      "loss": 0.918,
+      "step": 9332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2160106930584433e-06,
+      "loss": 0.8552,
+      "step": 9333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214288843596828e-06,
+      "loss": 0.8818,
+      "step": 9334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2125675800512026e-06,
+      "loss": 0.8984,
+      "step": 9335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2108469025511027e-06,
+      "loss": 0.8906,
+      "step": 9336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209126811226019e-06,
+      "loss": 0.8801,
+      "step": 9337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207407306205398e-06,
+      "loss": 0.9044,
+      "step": 9338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205688387618643e-06,
+      "loss": 0.8706,
+      "step": 9339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203970055595106e-06,
+      "loss": 0.869,
+      "step": 9340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202252310264106e-06,
+      "loss": 0.9219,
+      "step": 9341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2005351517549123e-06,
+      "loss": 0.8743,
+      "step": 9342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1988185801967464e-06,
+      "loss": 2.4642,
+      "step": 9343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1971025957187908e-06,
+      "loss": 0.8793,
+      "step": 9344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195387198450184e-06,
+      "loss": 0.9152,
+      "step": 9345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1936723885200173e-06,
+      "loss": 0.9139,
+      "step": 9346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.191958166057343e-06,
+      "loss": 0.8191,
+      "step": 9347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19024453119116e-06,
+      "loss": 0.8963,
+      "step": 9348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18853148405043e-06,
+      "loss": 0.9056,
+      "step": 9349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.186819024764072e-06,
+      "loss": 0.907,
+      "step": 9350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1851071534609503e-06,
+      "loss": 0.9067,
+      "step": 9351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1833958702698988e-06,
+      "loss": 0.8573,
+      "step": 9352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.899,
+      "step": 9353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1799750687390943e-06,
+      "loss": 0.9032,
+      "step": 9354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178265550656773e-06,
+      "loss": 0.8957,
+      "step": 9355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1765566212013843e-06,
+      "loss": 0.8741,
+      "step": 9356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.174848280501538e-06,
+      "loss": 0.8598,
+      "step": 9357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1731405286857943e-06,
+      "loss": 0.9374,
+      "step": 9358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1714333658826705e-06,
+      "loss": 0.9219,
+      "step": 9359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169726792220643e-06,
+      "loss": 0.9386,
+      "step": 9360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.168020807828136e-06,
+      "loss": 0.9091,
+      "step": 9361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166315412833534e-06,
+      "loss": 0.8998,
+      "step": 9362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1646106073651818e-06,
+      "loss": 0.9269,
+      "step": 9363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1629063915513694e-06,
+      "loss": 0.9165,
+      "step": 9364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1612027655203505e-06,
+      "loss": 0.904,
+      "step": 9365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1594997294003315e-06,
+      "loss": 0.9227,
+      "step": 9366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157797283319475e-06,
+      "loss": 0.9023,
+      "step": 9367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156095427405902e-06,
+      "loss": 0.8687,
+      "step": 9368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1543941617876817e-06,
+      "loss": 0.8532,
+      "step": 9369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152693486592845e-06,
+      "loss": 0.8468,
+      "step": 9370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150993401949376e-06,
+      "loss": 0.8999,
+      "step": 9371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.149293907985217e-06,
+      "loss": 0.8903,
+      "step": 9372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1475950048282623e-06,
+      "loss": 0.8434,
+      "step": 9373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1458966926063663e-06,
+      "loss": 0.9106,
+      "step": 9374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144198971447332e-06,
+      "loss": 0.9008,
+      "step": 9375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1425018414789256e-06,
+      "loss": 0.9104,
+      "step": 9376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14080530282886e-06,
+      "loss": 0.9105,
+      "step": 9377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1391093556248123e-06,
+      "loss": 2.4203,
+      "step": 9378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137413999994411e-06,
+      "loss": 0.9095,
+      "step": 9379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1357192360652402e-06,
+      "loss": 0.924,
+      "step": 9380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1340250639648406e-06,
+      "loss": 2.4435,
+      "step": 9381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13233148382071e-06,
+      "loss": 0.9021,
+      "step": 9382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1306384957602934e-06,
+      "loss": 0.9055,
+      "step": 9383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1289460999110035e-06,
+      "loss": 0.8642,
+      "step": 9384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1272542964001953e-06,
+      "loss": 0.9242,
+      "step": 9385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1255630853551866e-06,
+      "loss": 0.887,
+      "step": 9386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1238724669032594e-06,
+      "loss": 0.8866,
+      "step": 9387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1221824411716317e-06,
+      "loss": 0.8965,
+      "step": 9388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120493008287491e-06,
+      "loss": 0.9258,
+      "step": 9389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.118804168377977e-06,
+      "loss": 0.8894,
+      "step": 9390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1171159215701807e-06,
+      "loss": 0.873,
+      "step": 9391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115428267991152e-06,
+      "loss": 0.8969,
+      "step": 9392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1137412077678975e-06,
+      "loss": 0.8806,
+      "step": 9393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1120547410273775e-06,
+      "loss": 0.8869,
+      "step": 9394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1103688678965085e-06,
+      "loss": 0.9031,
+      "step": 9395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.108683588502157e-06,
+      "loss": 0.8745,
+      "step": 9396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106998902971156e-06,
+      "loss": 0.8987,
+      "step": 9397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1053148114302792e-06,
+      "loss": 0.8685,
+      "step": 9398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103631314006267e-06,
+      "loss": 0.9014,
+      "step": 9399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1019484108258124e-06,
+      "loss": 0.89,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1002661020155622e-06,
+      "loss": 0.8935,
+      "step": 9401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0985843877021182e-06,
+      "loss": 0.9287,
+      "step": 9402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0969032680120416e-06,
+      "loss": 0.8618,
+      "step": 9403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.095222743071841e-06,
+      "loss": 0.8931,
+      "step": 9404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0935428130079894e-06,
+      "loss": 0.8583,
+      "step": 9405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0918634779469017e-06,
+      "loss": 0.9064,
+      "step": 9406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0901847380149665e-06,
+      "loss": 0.902,
+      "step": 9407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885065933385164e-06,
+      "loss": 2.3922,
+      "step": 9408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0868290440438356e-06,
+      "loss": 0.8839,
+      "step": 9409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.085152090257171e-06,
+      "loss": 0.9134,
+      "step": 9410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083475732104726e-06,
+      "loss": 0.9288,
+      "step": 9411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0817999697126502e-06,
+      "loss": 0.9055,
+      "step": 9412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0801248032070543e-06,
+      "loss": 0.8956,
+      "step": 9413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0784502327140056e-06,
+      "loss": 0.8734,
+      "step": 9414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076776258359523e-06,
+      "loss": 0.8979,
+      "step": 9415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075102880269585e-06,
+      "loss": 0.8548,
+      "step": 9416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.073430098570117e-06,
+      "loss": 0.887,
+      "step": 9417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071757913387007e-06,
+      "loss": 0.908,
+      "step": 9418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070086324846099e-06,
+      "loss": 0.9403,
+      "step": 9419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0684153330731836e-06,
+      "loss": 0.8932,
+      "step": 9420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.066744938194014e-06,
+      "loss": 0.9082,
+      "step": 9421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0650751403342963e-06,
+      "loss": 0.8409,
+      "step": 9422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0634059396196915e-06,
+      "loss": 0.8997,
+      "step": 9423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0617373361758186e-06,
+      "loss": 0.9042,
+      "step": 9424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0600693301282436e-06,
+      "loss": 0.8661,
+      "step": 9425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058401921602495e-06,
+      "loss": 0.8629,
+      "step": 9426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0567351107240563e-06,
+      "loss": 0.8601,
+      "step": 9427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055068897618361e-06,
+      "loss": 0.9042,
+      "step": 9428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.053403282410804e-06,
+      "loss": 0.9455,
+      "step": 9429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.051738265226727e-06,
+      "loss": 0.865,
+      "step": 9430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0500738461914327e-06,
+      "loss": 0.935,
+      "step": 9431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0484100254301818e-06,
+      "loss": 0.9018,
+      "step": 9432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0467468030681792e-06,
+      "loss": 0.9054,
+      "step": 9433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045084179230594e-06,
+      "loss": 0.8683,
+      "step": 9434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043422154042548e-06,
+      "loss": 0.9279,
+      "step": 9435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0417607276291172e-06,
+      "loss": 0.8769,
+      "step": 9436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0400999001153344e-06,
+      "loss": 0.8991,
+      "step": 9437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038439671626182e-06,
+      "loss": 0.9737,
+      "step": 9438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0367800422866024e-06,
+      "loss": 0.8714,
+      "step": 9439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035121012221494e-06,
+      "loss": 0.8784,
+      "step": 9440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033462581555703e-06,
+      "loss": 2.6062,
+      "step": 9441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031804750414038e-06,
+      "loss": 0.9092,
+      "step": 9442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0301475189212593e-06,
+      "loss": 0.8708,
+      "step": 9443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0284908872020815e-06,
+      "loss": 0.8951,
+      "step": 9444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.026834855381179e-06,
+      "loss": 0.8831,
+      "step": 9445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.02517942358317e-06,
+      "loss": 0.8724,
+      "step": 9446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0235245919326385e-06,
+      "loss": 2.4981,
+      "step": 9447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021870360554119e-06,
+      "loss": 0.8754,
+      "step": 9448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020216729572101e-06,
+      "loss": 0.9005,
+      "step": 9449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0185636991110314e-06,
+      "loss": 0.8799,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016911269295305e-06,
+      "loss": 0.8897,
+      "step": 9451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0152594402492774e-06,
+      "loss": 0.9512,
+      "step": 9452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013608212097261e-06,
+      "loss": 0.8732,
+      "step": 9453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0119575849635142e-06,
+      "loss": 0.9158,
+      "step": 9454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0103075589722576e-06,
+      "loss": 0.8596,
+      "step": 9455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008658134247665e-06,
+      "loss": 0.898,
+      "step": 9456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.007009310913863e-06,
+      "loss": 0.8912,
+      "step": 9457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0053610890949394e-06,
+      "loss": 0.9083,
+      "step": 9458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.003713468914924e-06,
+      "loss": 0.8809,
+      "step": 9459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0020664504978126e-06,
+      "loss": 0.8843,
+      "step": 9460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000420033967554e-06,
+      "loss": 0.9084,
+      "step": 9461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998774219448043e-06,
+      "loss": 0.9191,
+      "step": 9462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971290070631457e-06,
+      "loss": 0.87,
+      "step": 9463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995484396936664e-06,
+      "loss": 0.8945,
+      "step": 9464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9938403891923685e-06,
+      "loss": 0.8713,
+      "step": 9465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9921969839539814e-06,
+      "loss": 0.8739,
+      "step": 9466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990554181345171e-06,
+      "loss": 0.8969,
+      "step": 9467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988911981489571e-06,
+      "loss": 0.9274,
+      "step": 9468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9872703845107644e-06,
+      "loss": 0.9108,
+      "step": 9469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985629390532291e-06,
+      "loss": 0.8954,
+      "step": 9470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9839889996776475e-06,
+      "loss": 0.8545,
+      "step": 9471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982349212070275e-06,
+      "loss": 0.8813,
+      "step": 9472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98071002783358e-06,
+      "loss": 0.8496,
+      "step": 9473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9790714470909223e-06,
+      "loss": 0.8671,
+      "step": 9474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9774334699656083e-06,
+      "loss": 0.8832,
+      "step": 9475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9757960965809075e-06,
+      "loss": 0.8775,
+      "step": 9476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974159327060041e-06,
+      "loss": 0.9126,
+      "step": 9477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972523161526183e-06,
+      "loss": 0.9063,
+      "step": 9478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970887600102468e-06,
+      "loss": 0.8498,
+      "step": 9479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969252642911974e-06,
+      "loss": 0.8992,
+      "step": 9480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967618290077744e-06,
+      "loss": 0.8946,
+      "step": 9481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965984541722774e-06,
+      "loss": 0.8903,
+      "step": 9482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9643513979700035e-06,
+      "loss": 0.9285,
+      "step": 9483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962718858942344e-06,
+      "loss": 0.8825,
+      "step": 9484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9610869247626517e-06,
+      "loss": 0.9228,
+      "step": 9485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9594555955537342e-06,
+      "loss": 0.9063,
+      "step": 9486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957824871438363e-06,
+      "loss": 0.9298,
+      "step": 9487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956194752539252e-06,
+      "loss": 0.9126,
+      "step": 9488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9545652389790805e-06,
+      "loss": 0.897,
+      "step": 9489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952936330880477e-06,
+      "loss": 0.8761,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951308028366027e-06,
+      "loss": 0.8769,
+      "step": 9491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949680331558266e-06,
+      "loss": 0.9252,
+      "step": 9492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9480532405796914e-06,
+      "loss": 0.9055,
+      "step": 9493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946426755552746e-06,
+      "loss": 0.8778,
+      "step": 9494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944800876599835e-06,
+      "loss": 0.8954,
+      "step": 9495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943175603843309e-06,
+      "loss": 0.9419,
+      "step": 9496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941550937405483e-06,
+      "loss": 0.8908,
+      "step": 9497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93992687740862e-06,
+      "loss": 0.877,
+      "step": 9498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9383034239749387e-06,
+      "loss": 0.8959,
+      "step": 9499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936680577226617e-06,
+      "loss": 0.8864,
+      "step": 9500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.935058337285777e-06,
+      "loss": 0.8679,
+      "step": 9501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9334367042745015e-06,
+      "loss": 0.904,
+      "step": 9502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93181567831483e-06,
+      "loss": 0.8801,
+      "step": 9503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930195259528751e-06,
+      "loss": 0.8684,
+      "step": 9504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.92857544803821e-06,
+      "loss": 0.8732,
+      "step": 9505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9269562439651103e-06,
+      "loss": 0.8741,
+      "step": 9506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925337647431299e-06,
+      "loss": 0.9098,
+      "step": 9507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9237196585585893e-06,
+      "loss": 0.878,
+      "step": 9508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.922102277468739e-06,
+      "loss": 0.9122,
+      "step": 9509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9204855042834668e-06,
+      "loss": 0.8512,
+      "step": 9510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9188693391244438e-06,
+      "loss": 0.9042,
+      "step": 9511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917253782113294e-06,
+      "loss": 0.8747,
+      "step": 9512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.915638833371598e-06,
+      "loss": 0.9084,
+      "step": 9513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914024493020891e-06,
+      "loss": 0.8908,
+      "step": 9514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9124107611826547e-06,
+      "loss": 0.91,
+      "step": 9515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107976379783377e-06,
+      "loss": 2.5392,
+      "step": 9516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90918512352933e-06,
+      "loss": 0.863,
+      "step": 9517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9075732179569815e-06,
+      "loss": 0.9318,
+      "step": 9518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9059619213826064e-06,
+      "loss": 0.8803,
+      "step": 9519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9043512339274527e-06,
+      "loss": 0.8894,
+      "step": 9520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9027411557127385e-06,
+      "loss": 0.9013,
+      "step": 9521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9011316868596307e-06,
+      "loss": 0.8812,
+      "step": 9522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8995228274892474e-06,
+      "loss": 2.6074,
+      "step": 9523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897914577722665e-06,
+      "loss": 0.8551,
+      "step": 9524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8963069376809129e-06,
+      "loss": 0.9,
+      "step": 9525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8946999074849749e-06,
+      "loss": 0.8896,
+      "step": 9526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8930934872557905e-06,
+      "loss": 0.9242,
+      "step": 9527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8914876771142477e-06,
+      "loss": 0.9054,
+      "step": 9528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.889882477181194e-06,
+      "loss": 0.8658,
+      "step": 9529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888277887577431e-06,
+      "loss": 0.9029,
+      "step": 9530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8866739084237074e-06,
+      "loss": 0.8649,
+      "step": 9531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8850705398407354e-06,
+      "loss": 0.8823,
+      "step": 9532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8834677819491753e-06,
+      "loss": 0.8766,
+      "step": 9533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881865634869644e-06,
+      "loss": 0.8783,
+      "step": 9534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8802640987227127e-06,
+      "loss": 0.9293,
+      "step": 9535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878663173628903e-06,
+      "loss": 0.8445,
+      "step": 9536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.877062859708696e-06,
+      "loss": 0.8937,
+      "step": 9537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8754631570825166e-06,
+      "loss": 0.9263,
+      "step": 9538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8738640658707585e-06,
+      "loss": 0.8623,
+      "step": 9539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8722655861937621e-06,
+      "loss": 0.9391,
+      "step": 9540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870667718171816e-06,
+      "loss": 0.9246,
+      "step": 9541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.869070461925171e-06,
+      "loss": 0.8657,
+      "step": 9542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8674738175740314e-06,
+      "loss": 0.8917,
+      "step": 9543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865877785238549e-06,
+      "loss": 0.8451,
+      "step": 9544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8642823650388354e-06,
+      "loss": 0.8993,
+      "step": 9545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8626875570949533e-06,
+      "loss": 0.9234,
+      "step": 9546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8610933615269233e-06,
+      "loss": 0.9192,
+      "step": 9547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8594997784547165e-06,
+      "loss": 0.909,
+      "step": 9548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8579068079982555e-06,
+      "loss": 2.4106,
+      "step": 9549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.856314450277421e-06,
+      "loss": 0.8273,
+      "step": 9550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.854722705412051e-06,
+      "loss": 0.9459,
+      "step": 9551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8531315735219247e-06,
+      "loss": 0.8709,
+      "step": 9552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8515410547267875e-06,
+      "loss": 2.4079,
+      "step": 9553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8499511491463351e-06,
+      "loss": 0.8819,
+      "step": 9554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8483618569002138e-06,
+      "loss": 0.8749,
+      "step": 9555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8467731781080312e-06,
+      "loss": 0.8855,
+      "step": 9556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451851128893373e-06,
+      "loss": 0.9299,
+      "step": 9557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8435976613636464e-06,
+      "loss": 0.906,
+      "step": 9558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8420108236504208e-06,
+      "loss": 0.8823,
+      "step": 9559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8404245998690795e-06,
+      "loss": 0.8845,
+      "step": 9560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838838990138997e-06,
+      "loss": 0.9072,
+      "step": 9561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837253994579492e-06,
+      "loss": 0.8862,
+      "step": 9562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8356696133098496e-06,
+      "loss": 0.9024,
+      "step": 9563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8340858464493017e-06,
+      "loss": 0.88,
+      "step": 9564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8325026941170331e-06,
+      "loss": 0.8567,
+      "step": 9565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8309201564321854e-06,
+      "loss": 0.881,
+      "step": 9566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8293382335138533e-06,
+      "loss": 0.9167,
+      "step": 9567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8277569254810845e-06,
+      "loss": 0.9105,
+      "step": 9568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8261762324528843e-06,
+      "loss": 0.8883,
+      "step": 9569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8245961545482015e-06,
+      "loss": 0.9155,
+      "step": 9570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8230166918859503e-06,
+      "loss": 0.9085,
+      "step": 9571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8214378445849935e-06,
+      "loss": 0.8951,
+      "step": 9572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198596127641444e-06,
+      "loss": 0.8771,
+      "step": 9573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.818281996542175e-06,
+      "loss": 0.848,
+      "step": 9574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8167049960378103e-06,
+      "loss": 0.9085,
+      "step": 9575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8151286113697275e-06,
+      "loss": 0.8859,
+      "step": 9576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8135528426565584e-06,
+      "loss": 0.9009,
+      "step": 9577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8119776900168862e-06,
+      "loss": 0.897,
+      "step": 9578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8104031535692501e-06,
+      "loss": 0.8628,
+      "step": 9579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8088292334321433e-06,
+      "loss": 0.8885,
+      "step": 9580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80725592972401e-06,
+      "loss": 0.8978,
+      "step": 9581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.805683242563253e-06,
+      "loss": 0.8558,
+      "step": 9582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8041111720682214e-06,
+      "loss": 0.9082,
+      "step": 9583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8025397183572224e-06,
+      "loss": 2.4466,
+      "step": 9584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8009688815485204e-06,
+      "loss": 0.8863,
+      "step": 9585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7993986617603232e-06,
+      "loss": 0.8794,
+      "step": 9586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7978290591108017e-06,
+      "loss": 0.918,
+      "step": 9587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7962600737180747e-06,
+      "loss": 0.8519,
+      "step": 9588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7946917057002189e-06,
+      "loss": 0.8972,
+      "step": 9589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7931239551752645e-06,
+      "loss": 0.8818,
+      "step": 9590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7915568222611856e-06,
+      "loss": 0.8995,
+      "step": 9591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7899903070759229e-06,
+      "loss": 0.8491,
+      "step": 9592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.788424409737366e-06,
+      "loss": 0.8889,
+      "step": 9593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7868591303633498e-06,
+      "loss": 0.8587,
+      "step": 9594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7852944690716766e-06,
+      "loss": 0.8898,
+      "step": 9595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7837304259800969e-06,
+      "loss": 0.8985,
+      "step": 9596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7821670012063064e-06,
+      "loss": 0.9229,
+      "step": 9597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7806041948679675e-06,
+      "loss": 0.89,
+      "step": 9598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7790420070826842e-06,
+      "loss": 0.865,
+      "step": 9599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.777480437968022e-06,
+      "loss": 0.8878,
+      "step": 9600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7759194876414965e-06,
+      "loss": 0.8905,
+      "step": 9601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7743591562205786e-06,
+      "loss": 0.908,
+      "step": 9602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7727994438226937e-06,
+      "loss": 0.8864,
+      "step": 9603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7712403505652131e-06,
+      "loss": 0.8928,
+      "step": 9604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.76968187656547e-06,
+      "loss": 0.9066,
+      "step": 9605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7681240219407492e-06,
+      "loss": 0.8921,
+      "step": 9606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7665667868082837e-06,
+      "loss": 0.8782,
+      "step": 9607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7650101712852642e-06,
+      "loss": 0.8777,
+      "step": 9608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7634541754888367e-06,
+      "loss": 0.8652,
+      "step": 9609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761898799536097e-06,
+      "loss": 0.9602,
+      "step": 9610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7603440435440976e-06,
+      "loss": 0.8828,
+      "step": 9611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7587899076298375e-06,
+      "loss": 0.8403,
+      "step": 9612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7572363919102763e-06,
+      "loss": 0.8871,
+      "step": 9613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7556834965023272e-06,
+      "loss": 0.96,
+      "step": 9614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7541312215228446e-06,
+      "loss": 0.8624,
+      "step": 9615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7525795670886548e-06,
+      "loss": 0.8851,
+      "step": 9616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7510285333165266e-06,
+      "loss": 0.8802,
+      "step": 9617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7494781203231803e-06,
+      "loss": 0.8729,
+      "step": 9618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7479283282252956e-06,
+      "loss": 0.8565,
+      "step": 9619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7463791571394983e-06,
+      "loss": 0.8591,
+      "step": 9620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744830607182375e-06,
+      "loss": 0.8872,
+      "step": 9621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432826784704615e-06,
+      "loss": 0.8748,
+      "step": 9622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7417353711202478e-06,
+      "loss": 0.8538,
+      "step": 9623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7401886852481775e-06,
+      "loss": 0.8784,
+      "step": 9624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7386426209706487e-06,
+      "loss": 0.8743,
+      "step": 9625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737097178404006e-06,
+      "loss": 0.9093,
+      "step": 9626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7355523576645573e-06,
+      "loss": 0.872,
+      "step": 9627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.734008158868553e-06,
+      "loss": 0.8817,
+      "step": 9628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732464582132205e-06,
+      "loss": 0.8882,
+      "step": 9629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7309216275716767e-06,
+      "loss": 0.8964,
+      "step": 9630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.729379295303083e-06,
+      "loss": 0.9235,
+      "step": 9631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7278375854424911e-06,
+      "loss": 0.8845,
+      "step": 9632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7262964981059283e-06,
+      "loss": 0.9253,
+      "step": 9633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7247560334093616e-06,
+      "loss": 0.9129,
+      "step": 9634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7232161914687229e-06,
+      "loss": 0.8898,
+      "step": 9635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7216769723998939e-06,
+      "loss": 0.9062,
+      "step": 9636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7201383763187086e-06,
+      "loss": 0.9057,
+      "step": 9637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7186004033409564e-06,
+      "loss": 0.8871,
+      "step": 9638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.717063053582374e-06,
+      "loss": 0.9287,
+      "step": 9639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7155263271586586e-06,
+      "loss": 0.8835,
+      "step": 9640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7139902241854544e-06,
+      "loss": 0.9533,
+      "step": 9641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7124547447783613e-06,
+      "loss": 0.9105,
+      "step": 9642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7109198890529333e-06,
+      "loss": 0.8921,
+      "step": 9643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709385657124677e-06,
+      "loss": 0.8928,
+      "step": 9644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707852049109051e-06,
+      "loss": 0.8996,
+      "step": 9645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063190651214689e-06,
+      "loss": 0.9203,
+      "step": 9646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047867052772927e-06,
+      "loss": 0.9351,
+      "step": 9647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7032549696918433e-06,
+      "loss": 0.9083,
+      "step": 9648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017238584803897e-06,
+      "loss": 0.8741,
+      "step": 9649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7001933717581542e-06,
+      "loss": 0.8224,
+      "step": 9650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6986635096403213e-06,
+      "loss": 0.8709,
+      "step": 9651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6971342722420158e-06,
+      "loss": 2.4029,
+      "step": 9652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.695605659678321e-06,
+      "loss": 0.8711,
+      "step": 9653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694077672064277e-06,
+      "loss": 0.9274,
+      "step": 9654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6925503095148687e-06,
+      "loss": 0.8681,
+      "step": 9655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6910235721450385e-06,
+      "loss": 0.9113,
+      "step": 9656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6894974600696822e-06,
+      "loss": 0.889,
+      "step": 9657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6879719734036494e-06,
+      "loss": 0.881,
+      "step": 9658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6864471122617422e-06,
+      "loss": 0.8734,
+      "step": 9659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.684922876758709e-06,
+      "loss": 0.8983,
+      "step": 9660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6833992670092592e-06,
+      "loss": 0.8728,
+      "step": 9661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6818762831280565e-06,
+      "loss": 0.9296,
+      "step": 9662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6803539252297074e-06,
+      "loss": 0.8985,
+      "step": 9663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6788321934287799e-06,
+      "loss": 0.9296,
+      "step": 9664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6773110878397937e-06,
+      "loss": 0.9062,
+      "step": 9665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6757906085772169e-06,
+      "loss": 0.8533,
+      "step": 9666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6742707557554794e-06,
+      "loss": 0.8998,
+      "step": 9667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6727515294889518e-06,
+      "loss": 2.4554,
+      "step": 9668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6712329298919684e-06,
+      "loss": 0.8721,
+      "step": 9669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6697149570788052e-06,
+      "loss": 0.9096,
+      "step": 9670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6681976111637045e-06,
+      "loss": 0.8977,
+      "step": 9671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6666808922608557e-06,
+      "loss": 2.414,
+      "step": 9672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6651648004843935e-06,
+      "loss": 0.9095,
+      "step": 9673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6636493359484162e-06,
+      "loss": 0.8797,
+      "step": 9674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6621344987669707e-06,
+      "loss": 0.8776,
+      "step": 9675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6606202890540535e-06,
+      "loss": 0.8406,
+      "step": 9676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6591067069236188e-06,
+      "loss": 0.9411,
+      "step": 9677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6575937524895713e-06,
+      "loss": 0.8578,
+      "step": 9678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6560814258657687e-06,
+      "loss": 0.8965,
+      "step": 9679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6545697271660254e-06,
+      "loss": 0.8554,
+      "step": 9680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6530586565040984e-06,
+      "loss": 0.9055,
+      "step": 9681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6515482139937057e-06,
+      "loss": 0.9079,
+      "step": 9682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6500383997485204e-06,
+      "loss": 0.8951,
+      "step": 9683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6485292138821585e-06,
+      "loss": 0.8944,
+      "step": 9684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6470206565081959e-06,
+      "loss": 0.8815,
+      "step": 9685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6455127277401595e-06,
+      "loss": 0.8886,
+      "step": 9686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6440054276915308e-06,
+      "loss": 0.9236,
+      "step": 9687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6424987564757423e-06,
+      "loss": 0.8914,
+      "step": 9688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6409927142061743e-06,
+      "loss": 0.8891,
+      "step": 9689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6394873009961688e-06,
+      "loss": 0.9164,
+      "step": 9690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6379825169590146e-06,
+      "loss": 0.884,
+      "step": 9691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636478362207955e-06,
+      "loss": 0.885,
+      "step": 9692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6349748368561879e-06,
+      "loss": 0.8793,
+      "step": 9693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6334719410168565e-06,
+      "loss": 0.8851,
+      "step": 9694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6319696748030655e-06,
+      "loss": 0.9394,
+      "step": 9695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6304680383278682e-06,
+      "loss": 0.9189,
+      "step": 9696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6289670317042682e-06,
+      "loss": 0.9068,
+      "step": 9697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6274666550452256e-06,
+      "loss": 0.9237,
+      "step": 9698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.625966908463652e-06,
+      "loss": 0.8559,
+      "step": 9699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6244677920724117e-06,
+      "loss": 0.8621,
+      "step": 9700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.622969305984322e-06,
+      "loss": 0.9045,
+      "step": 9701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6214714503121486e-06,
+      "loss": 0.8798,
+      "step": 9702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6199742251686146e-06,
+      "loss": 0.8823,
+      "step": 9703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6184776306663975e-06,
+      "loss": 0.9083,
+      "step": 9704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6169816669181182e-06,
+      "loss": 0.8775,
+      "step": 9705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6154863340363591e-06,
+      "loss": 0.8588,
+      "step": 9706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6139916321336513e-06,
+      "loss": 0.8852,
+      "step": 9707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6124975613224781e-06,
+      "loss": 0.8986,
+      "step": 9708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6110041217152816e-06,
+      "loss": 0.8808,
+      "step": 9709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6095113134244434e-06,
+      "loss": 0.948,
+      "step": 9710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6080191365623078e-06,
+      "loss": 0.8586,
+      "step": 9711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6065275912411704e-06,
+      "loss": 0.9171,
+      "step": 9712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6050366775732785e-06,
+      "loss": 0.9138,
+      "step": 9713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6035463956708309e-06,
+      "loss": 0.9046,
+      "step": 9714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6020567456459767e-06,
+      "loss": 0.8885,
+      "step": 9715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6005677276108212e-06,
+      "loss": 0.8587,
+      "step": 9716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5990793416774241e-06,
+      "loss": 0.8976,
+      "step": 9717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5975915879577885e-06,
+      "loss": 0.9043,
+      "step": 9718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5961044665638804e-06,
+      "loss": 0.8357,
+      "step": 9719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5946179776076098e-06,
+      "loss": 0.8682,
+      "step": 9720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5931321212008465e-06,
+      "loss": 0.9216,
+      "step": 9721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5916468974554089e-06,
+      "loss": 0.868,
+      "step": 9722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5901623064830651e-06,
+      "loss": 0.8929,
+      "step": 9723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5886783483955393e-06,
+      "loss": 0.9028,
+      "step": 9724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5871950233045107e-06,
+      "loss": 0.9156,
+      "step": 9725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5857123313215994e-06,
+      "loss": 0.8893,
+      "step": 9726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5842302725583935e-06,
+      "loss": 0.9239,
+      "step": 9727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5827488471264251e-06,
+      "loss": 0.864,
+      "step": 9728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5812680551371762e-06,
+      "loss": 0.895,
+      "step": 9729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5797878967020874e-06,
+      "loss": 0.9056,
+      "step": 9730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5783083719325432e-06,
+      "loss": 0.8872,
+      "step": 9731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5768294809398899e-06,
+      "loss": 0.8497,
+      "step": 9732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5753512238354207e-06,
+      "loss": 2.4597,
+      "step": 9733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5738736007303824e-06,
+      "loss": 0.8607,
+      "step": 9734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5723966117359745e-06,
+      "loss": 0.8957,
+      "step": 9735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5709202569633497e-06,
+      "loss": 0.8953,
+      "step": 9736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5694445365236078e-06,
+      "loss": 0.9178,
+      "step": 9737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5679694505278076e-06,
+      "loss": 0.869,
+      "step": 9738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5664949990869548e-06,
+      "loss": 0.929,
+      "step": 9739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565021182312011e-06,
+      "loss": 0.9195,
+      "step": 9740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5635480003138882e-06,
+      "loss": 0.8521,
+      "step": 9741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.562075453203451e-06,
+      "loss": 0.8621,
+      "step": 9742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.56060354109152e-06,
+      "loss": 0.8611,
+      "step": 9743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5591322640888584e-06,
+      "loss": 0.9147,
+      "step": 9744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5576616223061902e-06,
+      "loss": 0.9241,
+      "step": 9745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5561916158541923e-06,
+      "loss": 0.8568,
+      "step": 9746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.554722244843483e-06,
+      "loss": 0.843,
+      "step": 9747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5532535093846457e-06,
+      "loss": 0.9689,
+      "step": 9748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5517854095882123e-06,
+      "loss": 2.4784,
+      "step": 9749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5503179455646599e-06,
+      "loss": 0.8368,
+      "step": 9750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.548851117424428e-06,
+      "loss": 0.8628,
+      "step": 9751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5473849252778982e-06,
+      "loss": 0.9041,
+      "step": 9752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5459193692354114e-06,
+      "loss": 0.8774,
+      "step": 9753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5444544494072577e-06,
+      "loss": 0.9056,
+      "step": 9754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5429901659036818e-06,
+      "loss": 0.857,
+      "step": 9755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5415265188348783e-06,
+      "loss": 0.8937,
+      "step": 9756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.540063508310995e-06,
+      "loss": 0.8859,
+      "step": 9757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5386011344421292e-06,
+      "loss": 0.9202,
+      "step": 9758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5371393973383341e-06,
+      "loss": 0.8735,
+      "step": 9759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535678297109612e-06,
+      "loss": 0.8824,
+      "step": 9760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5342178338659163e-06,
+      "loss": 0.8749,
+      "step": 9761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.917,
+      "step": 9762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5312988187731969e-06,
+      "loss": 0.8958,
+      "step": 9763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5298402671438427e-06,
+      "loss": 0.9103,
+      "step": 9764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528382352938863e-06,
+      "loss": 0.883,
+      "step": 9765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5269250762679676e-06,
+      "loss": 0.8919,
+      "step": 9766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5254684372408267e-06,
+      "loss": 0.9015,
+      "step": 9767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5240124359670616e-06,
+      "loss": 0.8946,
+      "step": 9768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5225570725562433e-06,
+      "loss": 0.8318,
+      "step": 9769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5211023471178975e-06,
+      "loss": 0.8952,
+      "step": 9770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5196482597614958e-06,
+      "loss": 0.9191,
+      "step": 9771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5181948105964706e-06,
+      "loss": 0.9024,
+      "step": 9772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5167419997321965e-06,
+      "loss": 0.9133,
+      "step": 9773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.515289827278007e-06,
+      "loss": 0.8926,
+      "step": 9774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5138382933431882e-06,
+      "loss": 0.9258,
+      "step": 9775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.512387398036973e-06,
+      "loss": 0.8911,
+      "step": 9776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5109371414685515e-06,
+      "loss": 0.8608,
+      "step": 9777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.509487523747063e-06,
+      "loss": 0.8704,
+      "step": 9778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508038544981596e-06,
+      "loss": 0.8602,
+      "step": 9779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5065902052811976e-06,
+      "loss": 0.8488,
+      "step": 9780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5051425047548584e-06,
+      "loss": 0.8838,
+      "step": 9781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.503695443511526e-06,
+      "loss": 0.8994,
+      "step": 9782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502249021660107e-06,
+      "loss": 0.8862,
+      "step": 9783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.500803239309443e-06,
+      "loss": 0.8994,
+      "step": 9784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4993580965683418e-06,
+      "loss": 0.9181,
+      "step": 9785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497913593545559e-06,
+      "loss": 0.8604,
+      "step": 9786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4964697303497965e-06,
+      "loss": 0.881,
+      "step": 9787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4950265070897152e-06,
+      "loss": 0.8509,
+      "step": 9788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4935839238739248e-06,
+      "loss": 0.8738,
+      "step": 9789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4921419808109882e-06,
+      "loss": 0.8999,
+      "step": 9790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4907006780094212e-06,
+      "loss": 0.8911,
+      "step": 9791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4892600155776837e-06,
+      "loss": 0.9071,
+      "step": 9792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4878199936241978e-06,
+      "loss": 0.9463,
+      "step": 9793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.486380612257333e-06,
+      "loss": 0.9124,
+      "step": 9794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4849418715854068e-06,
+      "loss": 0.8921,
+      "step": 9795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4835037717166944e-06,
+      "loss": 0.8912,
+      "step": 9796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4820663127594205e-06,
+      "loss": 0.8603,
+      "step": 9797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4806294948217604e-06,
+      "loss": 0.9,
+      "step": 9798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4791933180118456e-06,
+      "loss": 0.8739,
+      "step": 9799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4777577824377509e-06,
+      "loss": 0.9271,
+      "step": 9800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476322888207511e-06,
+      "loss": 0.8564,
+      "step": 9801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4748886354291092e-06,
+      "loss": 0.8832,
+      "step": 9802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4734550242104795e-06,
+      "loss": 0.8561,
+      "step": 9803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472022054659511e-06,
+      "loss": 0.9203,
+      "step": 9804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4705897268840398e-06,
+      "loss": 0.8977,
+      "step": 9805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469158040991856e-06,
+      "loss": 0.8621,
+      "step": 9806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4677269970907049e-06,
+      "loss": 0.8987,
+      "step": 9807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4662965952882757e-06,
+      "loss": 0.9259,
+      "step": 9808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4648668356922157e-06,
+      "loss": 0.8593,
+      "step": 9809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4634377184101222e-06,
+      "loss": 0.8812,
+      "step": 9810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4620092435495437e-06,
+      "loss": 0.8902,
+      "step": 9811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4605814112179827e-06,
+      "loss": 0.9021,
+      "step": 9812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.459154221522886e-06,
+      "loss": 0.857,
+      "step": 9813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577276745716607e-06,
+      "loss": 0.9052,
+      "step": 9814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4563017704716642e-06,
+      "loss": 0.8509,
+      "step": 9815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4548765093301974e-06,
+      "loss": 0.888,
+      "step": 9816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4534518912545216e-06,
+      "loss": 0.9529,
+      "step": 9817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452027916351848e-06,
+      "loss": 0.8328,
+      "step": 9818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.450604584729336e-06,
+      "loss": 0.8832,
+      "step": 9819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449181896494104e-06,
+      "loss": 0.8796,
+      "step": 9820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.44775985175321e-06,
+      "loss": 0.9211,
+      "step": 9821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4463384506136734e-06,
+      "loss": 0.8936,
+      "step": 9822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4449176931824615e-06,
+      "loss": 0.8758,
+      "step": 9823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4434975795664952e-06,
+      "loss": 0.9073,
+      "step": 9824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4420781098726477e-06,
+      "loss": 0.8923,
+      "step": 9825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.440659284207736e-06,
+      "loss": 0.8819,
+      "step": 9826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4392411026785369e-06,
+      "loss": 0.876,
+      "step": 9827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4378235653917782e-06,
+      "loss": 0.9421,
+      "step": 9828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4364066724541337e-06,
+      "loss": 0.9186,
+      "step": 9829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4349904239722324e-06,
+      "loss": 0.9089,
+      "step": 9830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4335748200526567e-06,
+      "loss": 0.8671,
+      "step": 9831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.432159860801936e-06,
+      "loss": 0.8742,
+      "step": 9832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4307455463265574e-06,
+      "loss": 0.892,
+      "step": 9833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4293318767329512e-06,
+      "loss": 0.9122,
+      "step": 9834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4279188521275044e-06,
+      "loss": 0.8927,
+      "step": 9835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4265064726165568e-06,
+      "loss": 0.8765,
+      "step": 9836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4250947383063952e-06,
+      "loss": 0.8647,
+      "step": 9837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.423683649303258e-06,
+      "loss": 0.8722,
+      "step": 9838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4222732057133436e-06,
+      "loss": 0.8887,
+      "step": 9839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4208634076427908e-06,
+      "loss": 2.667,
+      "step": 9840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4194542551976963e-06,
+      "loss": 0.8795,
+      "step": 9841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4180457484841036e-06,
+      "loss": 0.8656,
+      "step": 9842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4166378876080123e-06,
+      "loss": 0.8617,
+      "step": 9843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4152306726753706e-06,
+      "loss": 0.8362,
+      "step": 9844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4138241037920798e-06,
+      "loss": 0.8786,
+      "step": 9845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4124181810639925e-06,
+      "loss": 0.8815,
+      "step": 9846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.411012904596909e-06,
+      "loss": 0.9043,
+      "step": 9847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4096082744965845e-06,
+      "loss": 0.8686,
+      "step": 9848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4082042908687276e-06,
+      "loss": 0.8476,
+      "step": 9849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.406800953818993e-06,
+      "loss": 0.8995,
+      "step": 9850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4053982634529884e-06,
+      "loss": 0.8431,
+      "step": 9851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4039962198762757e-06,
+      "loss": 0.8623,
+      "step": 9852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4025948231943643e-06,
+      "loss": 0.9236,
+      "step": 9853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4011940735127216e-06,
+      "loss": 0.8573,
+      "step": 9854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3997939709367548e-06,
+      "loss": 0.8995,
+      "step": 9855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3983945155718314e-06,
+      "loss": 0.9092,
+      "step": 9856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396995707523271e-06,
+      "loss": 0.862,
+      "step": 9857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3955975468963345e-06,
+      "loss": 0.8893,
+      "step": 9858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.394200033796247e-06,
+      "loss": 0.8903,
+      "step": 9859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3928031683281796e-06,
+      "loss": 0.9118,
+      "step": 9860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3914069505972484e-06,
+      "loss": 0.8701,
+      "step": 9861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3900113807085314e-06,
+      "loss": 0.8707,
+      "step": 9862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.388616458767047e-06,
+      "loss": 0.9284,
+      "step": 9863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3872221848777746e-06,
+      "loss": 0.8703,
+      "step": 9864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.385828559145639e-06,
+      "loss": 0.9795,
+      "step": 9865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3844355816755183e-06,
+      "loss": 0.8618,
+      "step": 9866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3830432525722414e-06,
+      "loss": 0.8964,
+      "step": 9867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.381651571940592e-06,
+      "loss": 0.8441,
+      "step": 9868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3802605398852954e-06,
+      "loss": 0.9344,
+      "step": 9869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3788701565110385e-06,
+      "loss": 0.8582,
+      "step": 9870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3774804219224525e-06,
+      "loss": 0.8031,
+      "step": 9871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3760913362241223e-06,
+      "loss": 0.8813,
+      "step": 9872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3747028995205847e-06,
+      "loss": 0.9112,
+      "step": 9873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373315111916327e-06,
+      "loss": 2.4513,
+      "step": 9874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3719279735157875e-06,
+      "loss": 0.8959,
+      "step": 9875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370541484423359e-06,
+      "loss": 0.9228,
+      "step": 9876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3691556447433752e-06,
+      "loss": 0.9138,
+      "step": 9877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3677704545801352e-06,
+      "loss": 0.9223,
+      "step": 9878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3663859140378733e-06,
+      "loss": 2.4689,
+      "step": 9879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3650020232207907e-06,
+      "loss": 0.9125,
+      "step": 9880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3636187822330326e-06,
+      "loss": 0.8635,
+      "step": 9881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3622361911786909e-06,
+      "loss": 0.883,
+      "step": 9882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3608542501618172e-06,
+      "loss": 0.859,
+      "step": 9883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3594729592864042e-06,
+      "loss": 0.9008,
+      "step": 9884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3580923186564065e-06,
+      "loss": 0.8707,
+      "step": 9885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3567123283757221e-06,
+      "loss": 0.9179,
+      "step": 9886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3553329885482035e-06,
+      "loss": 0.9329,
+      "step": 9887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3539542992776534e-06,
+      "loss": 0.8913,
+      "step": 9888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3525762606678273e-06,
+      "loss": 0.8929,
+      "step": 9889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3511988728224245e-06,
+      "loss": 0.8642,
+      "step": 9890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.349822135845108e-06,
+      "loss": 0.8844,
+      "step": 9891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.348446049839478e-06,
+      "loss": 0.8714,
+      "step": 9892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.347070614909094e-06,
+      "loss": 0.8631,
+      "step": 9893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3456958311574663e-06,
+      "loss": 0.9006,
+      "step": 9894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3443216986880547e-06,
+      "loss": 0.8784,
+      "step": 9895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3429482176042686e-06,
+      "loss": 0.9018,
+      "step": 9896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3415753880094729e-06,
+      "loss": 0.8591,
+      "step": 9897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.340203210006975e-06,
+      "loss": 0.8779,
+      "step": 9898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3388316837000414e-06,
+      "loss": 0.8806,
+      "step": 9899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3374608091918872e-06,
+      "loss": 2.4163,
+      "step": 9900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3360905865856766e-06,
+      "loss": 0.8973,
+      "step": 9901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3347210159845303e-06,
+      "loss": 0.8402,
+      "step": 9902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3333520974915093e-06,
+      "loss": 0.8794,
+      "step": 9903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3319838312096356e-06,
+      "loss": 0.8663,
+      "step": 9904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330616217241879e-06,
+      "loss": 0.9169,
+      "step": 9905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3292492556911575e-06,
+      "loss": 0.877,
+      "step": 9906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3278829466603437e-06,
+      "loss": 0.8845,
+      "step": 9907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3265172902522582e-06,
+      "loss": 0.8752,
+      "step": 9908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3251522865696754e-06,
+      "loss": 0.8625,
+      "step": 9909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3237879357153205e-06,
+      "loss": 0.8775,
+      "step": 9910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322424237791864e-06,
+      "loss": 0.938,
+      "step": 9911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3210611929019357e-06,
+      "loss": 0.864,
+      "step": 9912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3196988011481071e-06,
+      "loss": 0.8757,
+      "step": 9913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3183370626329074e-06,
+      "loss": 0.8741,
+      "step": 9914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3169759774588187e-06,
+      "loss": 0.8843,
+      "step": 9915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3156155457282638e-06,
+      "loss": 0.92,
+      "step": 9916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3142557675436262e-06,
+      "loss": 0.8578,
+      "step": 9917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3128966430072366e-06,
+      "loss": 0.9326,
+      "step": 9918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3115381722213727e-06,
+      "loss": 0.9561,
+      "step": 9919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101803552882696e-06,
+      "loss": 0.8996,
+      "step": 9920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.308823192310109e-06,
+      "loss": 0.9308,
+      "step": 9921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3074666833890259e-06,
+      "loss": 0.8878,
+      "step": 9922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306110828627105e-06,
+      "loss": 0.9147,
+      "step": 9923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3047556281263796e-06,
+      "loss": 0.9226,
+      "step": 9924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3034010819888365e-06,
+      "loss": 0.8452,
+      "step": 9925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3020471903164145e-06,
+      "loss": 0.9334,
+      "step": 9926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3006939532109975e-06,
+      "loss": 0.8526,
+      "step": 9927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2993413707744252e-06,
+      "loss": 0.9218,
+      "step": 9928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2979894431084861e-06,
+      "loss": 0.8715,
+      "step": 9929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2966381703149222e-06,
+      "loss": 0.8746,
+      "step": 9930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2952875524954233e-06,
+      "loss": 0.8897,
+      "step": 9931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2939375897516281e-06,
+      "loss": 0.9096,
+      "step": 9932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292588282185131e-06,
+      "loss": 0.8148,
+      "step": 9933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2912396298974728e-06,
+      "loss": 0.8828,
+      "step": 9934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2898916329901468e-06,
+      "loss": 0.8787,
+      "step": 9935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2885442915646007e-06,
+      "loss": 0.8562,
+      "step": 9936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2871976057222235e-06,
+      "loss": 0.8876,
+      "step": 9937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.285851575564363e-06,
+      "loss": 0.9044,
+      "step": 9938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.284506201192318e-06,
+      "loss": 2.4554,
+      "step": 9939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2831614827073291e-06,
+      "loss": 0.8901,
+      "step": 9940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281817420210596e-06,
+      "loss": 0.8884,
+      "step": 9941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.280474013803268e-06,
+      "loss": 0.9166,
+      "step": 9942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2791312635864417e-06,
+      "loss": 0.8811,
+      "step": 9943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2777891696611698e-06,
+      "loss": 0.8961,
+      "step": 9944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2764477321284474e-06,
+      "loss": 0.8451,
+      "step": 9945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2751069510892256e-06,
+      "loss": 0.9279,
+      "step": 9946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2737668266444093e-06,
+      "loss": 0.9055,
+      "step": 9947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2724273588948433e-06,
+      "loss": 0.8953,
+      "step": 9948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710885479413349e-06,
+      "loss": 0.9079,
+      "step": 9949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2697503938846334e-06,
+      "loss": 0.852,
+      "step": 9950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.268412896825445e-06,
+      "loss": 0.903,
+      "step": 9951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2670760568644237e-06,
+      "loss": 0.9068,
+      "step": 9952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2657398741021698e-06,
+      "loss": 0.9202,
+      "step": 9953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.264404348639241e-06,
+      "loss": 0.8863,
+      "step": 9954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2630694805761413e-06,
+      "loss": 0.8724,
+      "step": 9955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2617352700133278e-06,
+      "loss": 0.8888,
+      "step": 9956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2604017170512095e-06,
+      "loss": 0.8467,
+      "step": 9957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.259068821790137e-06,
+      "loss": 0.8621,
+      "step": 9958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2577365843304212e-06,
+      "loss": 0.879,
+      "step": 9959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.256405004772323e-06,
+      "loss": 0.8938,
+      "step": 9960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2550740832160446e-06,
+      "loss": 0.8884,
+      "step": 9961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.253743819761749e-06,
+      "loss": 0.8985,
+      "step": 9962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252414214509544e-06,
+      "loss": 0.8623,
+      "step": 9963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2510852675594898e-06,
+      "loss": 0.9019,
+      "step": 9964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2497569790115993e-06,
+      "loss": 0.8944,
+      "step": 9965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2484293489658295e-06,
+      "loss": 0.9241,
+      "step": 9966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471023775220913e-06,
+      "loss": 0.8711,
+      "step": 9967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2457760647802507e-06,
+      "loss": 0.9474,
+      "step": 9968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2444504108401157e-06,
+      "loss": 0.8936,
+      "step": 9969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.243125415801447e-06,
+      "loss": 0.9537,
+      "step": 9970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2418010797639656e-06,
+      "loss": 0.9003,
+      "step": 9971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.240477402827328e-06,
+      "loss": 0.8379,
+      "step": 9972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2391543850911513e-06,
+      "loss": 0.8343,
+      "step": 9973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.237832026654996e-06,
+      "loss": 0.8824,
+      "step": 9974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2365103276183777e-06,
+      "loss": 0.8775,
+      "step": 9975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2351892880807637e-06,
+      "loss": 0.8802,
+      "step": 9976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2338689081415668e-06,
+      "loss": 0.8835,
+      "step": 9977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2325491879001528e-06,
+      "loss": 0.8986,
+      "step": 9978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2312301274558415e-06,
+      "loss": 0.919,
+      "step": 9979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2299117269078932e-06,
+      "loss": 0.8693,
+      "step": 9980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2285939863555296e-06,
+      "loss": 0.9433,
+      "step": 9981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2272769058979128e-06,
+      "loss": 0.8655,
+      "step": 9982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2259604856341633e-06,
+      "loss": 0.8863,
+      "step": 9983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2246447256633475e-06,
+      "loss": 0.9125,
+      "step": 9984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.223329626084484e-06,
+      "loss": 0.9192,
+      "step": 9985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2220151869965425e-06,
+      "loss": 0.8854,
+      "step": 9986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.220701408498438e-06,
+      "loss": 0.8947,
+      "step": 9987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.219388290689042e-06,
+      "loss": 0.8858,
+      "step": 9988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.218075833667174e-06,
+      "loss": 0.9286,
+      "step": 9989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216764037531597e-06,
+      "loss": 0.8926,
+      "step": 9990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2154529023810391e-06,
+      "loss": 0.8737,
+      "step": 9991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2141424283141678e-06,
+      "loss": 0.8862,
+      "step": 9992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2128326154296e-06,
+      "loss": 0.9148,
+      "step": 9993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.211523463825911e-06,
+      "loss": 0.8553,
+      "step": 9994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2102149736016156e-06,
+      "loss": 0.8596,
+      "step": 9995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.208907144855188e-06,
+      "loss": 0.8873,
+      "step": 9996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2075999776850488e-06,
+      "loss": 0.8699,
+      "step": 9997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2062934721895692e-06,
+      "loss": 0.9213,
+      "step": 9998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2049876284670715e-06,
+      "loss": 0.8967,
+      "step": 9999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.203682446615827e-06,
+      "loss": 0.9362,
+      "step": 10000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2023779267340562e-06,
+      "loss": 0.8961,
+      "step": 10001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201074068919934e-06,
+      "loss": 0.8791,
+      "step": 10002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1997708732715785e-06,
+      "loss": 0.902,
+      "step": 10003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1984683398870634e-06,
+      "loss": 0.861,
+      "step": 10004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1971664688644124e-06,
+      "loss": 0.8978,
+      "step": 10005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1958652603015986e-06,
+      "loss": 0.8543,
+      "step": 10006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1945647142965433e-06,
+      "loss": 0.9011,
+      "step": 10007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1932648309471228e-06,
+      "loss": 0.8932,
+      "step": 10008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1919656103511557e-06,
+      "loss": 0.858,
+      "step": 10009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190667052606419e-06,
+      "loss": 0.8751,
+      "step": 10010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1893691578106292e-06,
+      "loss": 0.8924,
+      "step": 10011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1880719260614683e-06,
+      "loss": 0.9013,
+      "step": 10012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1867753574565576e-06,
+      "loss": 0.8915,
+      "step": 10013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1854794520934677e-06,
+      "loss": 0.8653,
+      "step": 10014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1841842100697253e-06,
+      "loss": 0.8826,
+      "step": 10015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.182889631482802e-06,
+      "loss": 0.906,
+      "step": 10016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1815957164301229e-06,
+      "loss": 0.8495,
+      "step": 10017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180302465009061e-06,
+      "loss": 0.9257,
+      "step": 10018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1790098773169424e-06,
+      "loss": 0.8811,
+      "step": 10019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1777179534510387e-06,
+      "loss": 0.9112,
+      "step": 10020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1764266935085777e-06,
+      "loss": 0.8892,
+      "step": 10021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1751360975867287e-06,
+      "loss": 0.853,
+      "step": 10022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1738461657826205e-06,
+      "loss": 0.8993,
+      "step": 10023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1725568981933223e-06,
+      "loss": 0.9149,
+      "step": 10024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1712682949158617e-06,
+      "loss": 0.8986,
+      "step": 10025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1699803560472123e-06,
+      "loss": 0.8333,
+      "step": 10026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168693081684298e-06,
+      "loss": 0.9164,
+      "step": 10027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.167406471923994e-06,
+      "loss": 0.904,
+      "step": 10028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1661205268631247e-06,
+      "loss": 0.8657,
+      "step": 10029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1648352465984625e-06,
+      "loss": 0.9057,
+      "step": 10030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.163550631226732e-06,
+      "loss": 0.8676,
+      "step": 10031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.162266680844607e-06,
+      "loss": 0.8843,
+      "step": 10032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1609833955487126e-06,
+      "loss": 0.9112,
+      "step": 10033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1597007754356249e-06,
+      "loss": 0.9012,
+      "step": 10034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1584188206018632e-06,
+      "loss": 0.9138,
+      "step": 10035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1571375311439037e-06,
+      "loss": 0.8624,
+      "step": 10036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1558569071581727e-06,
+      "loss": 0.8885,
+      "step": 10037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1545769487410386e-06,
+      "loss": 0.8424,
+      "step": 10038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1532976559888287e-06,
+      "loss": 0.8609,
+      "step": 10039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1520190289978151e-06,
+      "loss": 0.8814,
+      "step": 10040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1507410678642216e-06,
+      "loss": 0.9123,
+      "step": 10041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.149463772684224e-06,
+      "loss": 0.8652,
+      "step": 10042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481871435539415e-06,
+      "loss": 0.9009,
+      "step": 10043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146911180569449e-06,
+      "loss": 0.853,
+      "step": 10044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1456358838267712e-06,
+      "loss": 0.8523,
+      "step": 10045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.144361253421875e-06,
+      "loss": 0.9043,
+      "step": 10046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1430872894506927e-06,
+      "loss": 0.8828,
+      "step": 10047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1418139920090876e-06,
+      "loss": 2.4319,
+      "step": 10048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405413611928873e-06,
+      "loss": 0.8422,
+      "step": 10049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1392693970978642e-06,
+      "loss": 0.9134,
+      "step": 10050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137998099819737e-06,
+      "loss": 0.8746,
+      "step": 10051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1367274694541785e-06,
+      "loss": 0.8934,
+      "step": 10052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.135457506096812e-06,
+      "loss": 0.8828,
+      "step": 10053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1341882098432077e-06,
+      "loss": 0.9186,
+      "step": 10054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1329195807888893e-06,
+      "loss": 0.9213,
+      "step": 10055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1316516190293237e-06,
+      "loss": 0.8598,
+      "step": 10056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1303843246599345e-06,
+      "loss": 0.8843,
+      "step": 10057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291176977760931e-06,
+      "loss": 0.8513,
+      "step": 10058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.127851738473117e-06,
+      "loss": 0.9328,
+      "step": 10059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265864468462773e-06,
+      "loss": 0.8409,
+      "step": 10060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1253218229907948e-06,
+      "loss": 0.9265,
+      "step": 10061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1240578670018388e-06,
+      "loss": 0.8735,
+      "step": 10062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.122794578974531e-06,
+      "loss": 0.8459,
+      "step": 10063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1215319590039353e-06,
+      "loss": 0.9021,
+      "step": 10064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1202700071850736e-06,
+      "loss": 0.8646,
+      "step": 10065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1190087236129154e-06,
+      "loss": 0.8818,
+      "step": 10066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117748108382376e-06,
+      "loss": 0.9169,
+      "step": 10067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164881615883283e-06,
+      "loss": 0.8303,
+      "step": 10068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1152288833255853e-06,
+      "loss": 0.8891,
+      "step": 10069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1139702736889147e-06,
+      "loss": 0.874,
+      "step": 10070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1127123327730383e-06,
+      "loss": 0.8484,
+      "step": 10071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.111455060672616e-06,
+      "loss": 0.8738,
+      "step": 10072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1101984574822677e-06,
+      "loss": 0.8668,
+      "step": 10073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.108942523296559e-06,
+      "loss": 0.9134,
+      "step": 10074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1076872582100063e-06,
+      "loss": 0.8523,
+      "step": 10075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1064326623170762e-06,
+      "loss": 0.9111,
+      "step": 10076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10517873571218e-06,
+      "loss": 0.913,
+      "step": 10077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1039254784896846e-06,
+      "loss": 0.8726,
+      "step": 10078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1026728907439056e-06,
+      "loss": 0.8769,
+      "step": 10079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1014209725691028e-06,
+      "loss": 0.8496,
+      "step": 10080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10016972405949e-06,
+      "loss": 0.8837,
+      "step": 10081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0989191453092374e-06,
+      "loss": 0.8848,
+      "step": 10082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0976692364124497e-06,
+      "loss": 0.8864,
+      "step": 10083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.096419997463194e-06,
+      "loss": 0.8761,
+      "step": 10084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0951714285554783e-06,
+      "loss": 0.9178,
+      "step": 10085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0939235297832661e-06,
+      "loss": 2.4164,
+      "step": 10086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0926763012404674e-06,
+      "loss": 0.8559,
+      "step": 10087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0914297430209442e-06,
+      "loss": 0.9035,
+      "step": 10088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0901838552185084e-06,
+      "loss": 0.8789,
+      "step": 10089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088938637926914e-06,
+      "loss": 0.915,
+      "step": 10090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.087694091239875e-06,
+      "loss": 0.899,
+      "step": 10091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0864502152510492e-06,
+      "loss": 0.8758,
+      "step": 10092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0852070100540424e-06,
+      "loss": 0.9136,
+      "step": 10093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083964475742415e-06,
+      "loss": 0.8648,
+      "step": 10094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0827226124096725e-06,
+      "loss": 0.9117,
+      "step": 10095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.081481420149273e-06,
+      "loss": 0.8159,
+      "step": 10096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0802408990546242e-06,
+      "loss": 0.8599,
+      "step": 10097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0790010492190784e-06,
+      "loss": 0.9132,
+      "step": 10098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0777618707359427e-06,
+      "loss": 0.8961,
+      "step": 10099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0765233636984728e-06,
+      "loss": 0.8672,
+      "step": 10100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0752855281998709e-06,
+      "loss": 0.8979,
+      "step": 10101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07404836433329e-06,
+      "loss": 0.8822,
+      "step": 10102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072811872191838e-06,
+      "loss": 0.8959,
+      "step": 10103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0715760518685625e-06,
+      "loss": 0.9158,
+      "step": 10104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0703409034564682e-06,
+      "loss": 0.8773,
+      "step": 10105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0691064270485051e-06,
+      "loss": 0.8991,
+      "step": 10106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0678726227375747e-06,
+      "loss": 0.8667,
+      "step": 10107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0666394906165267e-06,
+      "loss": 0.8902,
+      "step": 10108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0654070307781616e-06,
+      "loss": 0.9219,
+      "step": 10109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0641752433152286e-06,
+      "loss": 0.9359,
+      "step": 10110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.062944128320428e-06,
+      "loss": 0.8978,
+      "step": 10111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061713685886403e-06,
+      "loss": 0.8656,
+      "step": 10112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0604839161057567e-06,
+      "loss": 0.8736,
+      "step": 10113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0592548190710295e-06,
+      "loss": 0.9193,
+      "step": 10114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0580263948747215e-06,
+      "loss": 0.8886,
+      "step": 10115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.056798643609277e-06,
+      "loss": 0.8905,
+      "step": 10116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0555715653670905e-06,
+      "loss": 0.8643,
+      "step": 10117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0543451602405096e-06,
+      "loss": 0.8393,
+      "step": 10118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0531194283218226e-06,
+      "loss": 0.9088,
+      "step": 10119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0518943697032746e-06,
+      "loss": 0.9028,
+      "step": 10120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0506699844770597e-06,
+      "loss": 0.9051,
+      "step": 10121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.049446272735314e-06,
+      "loss": 0.8939,
+      "step": 10122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0482232345701326e-06,
+      "loss": 0.875,
+      "step": 10123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0470008700735568e-06,
+      "loss": 0.9052,
+      "step": 10124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0457791793375727e-06,
+      "loss": 0.9218,
+      "step": 10125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0445581624541223e-06,
+      "loss": 0.8639,
+      "step": 10126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0433378195150889e-06,
+      "loss": 0.8616,
+      "step": 10127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0421181506123124e-06,
+      "loss": 0.8981,
+      "step": 10128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040899155837579e-06,
+      "loss": 0.9136,
+      "step": 10129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.039680835282626e-06,
+      "loss": 0.8802,
+      "step": 10130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0384631890391362e-06,
+      "loss": 0.8944,
+      "step": 10131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0372462171987474e-06,
+      "loss": 0.9093,
+      "step": 10132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0360299198530387e-06,
+      "loss": 0.8751,
+      "step": 10133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.034814297093547e-06,
+      "loss": 0.8875,
+      "step": 10134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0335993490117503e-06,
+      "loss": 0.885,
+      "step": 10135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0323850756990817e-06,
+      "loss": 0.9049,
+      "step": 10136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0311714772469217e-06,
+      "loss": 0.8873,
+      "step": 10137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0299585537466007e-06,
+      "loss": 2.6141,
+      "step": 10138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0287463052893976e-06,
+      "loss": 0.8408,
+      "step": 10139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0275347319665419e-06,
+      "loss": 0.9235,
+      "step": 10140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0263238338692061e-06,
+      "loss": 0.9146,
+      "step": 10141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0251136110885229e-06,
+      "loss": 0.8725,
+      "step": 10142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0239040637155606e-06,
+      "loss": 0.8979,
+      "step": 10143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0226951918413498e-06,
+      "loss": 0.8881,
+      "step": 10144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.021486995556864e-06,
+      "loss": 0.9204,
+      "step": 10145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0202794749530242e-06,
+      "loss": 0.8684,
+      "step": 10146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0190726301207043e-06,
+      "loss": 0.9067,
+      "step": 10147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0178664611507261e-06,
+      "loss": 0.8757,
+      "step": 10148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.016660968133858e-06,
+      "loss": 2.5044,
+      "step": 10149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0154561511608218e-06,
+      "loss": 0.9097,
+      "step": 10150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142520103222841e-06,
+      "loss": 0.9137,
+      "step": 10151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0130485457088657e-06,
+      "loss": 0.9079,
+      "step": 10152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011845757411134e-06,
+      "loss": 0.9435,
+      "step": 10153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.010643645519601e-06,
+      "loss": 0.8754,
+      "step": 10154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.009442210124737e-06,
+      "loss": 0.8427,
+      "step": 10155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0082414513169526e-06,
+      "loss": 0.9443,
+      "step": 10156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.007041369186612e-06,
+      "loss": 0.8686,
+      "step": 10157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.005841963824028e-06,
+      "loss": 0.9232,
+      "step": 10158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0046432353194635e-06,
+      "loss": 0.8582,
+      "step": 10159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0034451837631277e-06,
+      "loss": 0.8205,
+      "step": 10160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0022478092451826e-06,
+      "loss": 0.8641,
+      "step": 10161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.001051111855733e-06,
+      "loss": 0.878,
+      "step": 10162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.998550916848393e-07,
+      "loss": 0.8531,
+      "step": 10163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.986597488225081e-07,
+      "loss": 0.9048,
+      "step": 10164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.974650833586952e-07,
+      "loss": 0.9159,
+      "step": 10165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.962710953833065e-07,
+      "loss": 0.9347,
+      "step": 10166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.950777849861948e-07,
+      "loss": 0.8906,
+      "step": 10167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.938851522571614e-07,
+      "loss": 0.9161,
+      "step": 10168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.926931972859622e-07,
+      "loss": 0.8768,
+      "step": 10169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.915019201622944e-07,
+      "loss": 0.8936,
+      "step": 10170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.9072,
+      "step": 10171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.891213998161052e-07,
+      "loss": 0.8903,
+      "step": 10172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.879321567727317e-07,
+      "loss": 0.8281,
+      "step": 10173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.867435919351854e-07,
+      "loss": 0.9108,
+      "step": 10174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.855557053929099e-07,
+      "loss": 0.8727,
+      "step": 10175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.843684972353017e-07,
+      "loss": 0.8693,
+      "step": 10176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.83181967551705e-07,
+      "loss": 0.9103,
+      "step": 10177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.819961164314085e-07,
+      "loss": 0.883,
+      "step": 10178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.808109439636605e-07,
+      "loss": 0.9021,
+      "step": 10179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.796264502376462e-07,
+      "loss": 0.908,
+      "step": 10180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.784426353425069e-07,
+      "loss": 0.9367,
+      "step": 10181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.772594993673312e-07,
+      "loss": 0.8806,
+      "step": 10182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.760770424011557e-07,
+      "loss": 0.8838,
+      "step": 10183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.748952645329658e-07,
+      "loss": 0.8822,
+      "step": 10184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.737141658516968e-07,
+      "loss": 0.8424,
+      "step": 10185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.725337464462348e-07,
+      "loss": 0.9361,
+      "step": 10186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.713540064054117e-07,
+      "loss": 0.8687,
+      "step": 10187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.701749458180065e-07,
+      "loss": 0.8292,
+      "step": 10188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.689965647727528e-07,
+      "loss": 0.8754,
+      "step": 10189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678188633583307e-07,
+      "loss": 0.8835,
+      "step": 10190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66641841663365e-07,
+      "loss": 0.8927,
+      "step": 10191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.654654997764345e-07,
+      "loss": 0.8952,
+      "step": 10192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642898377860643e-07,
+      "loss": 0.9205,
+      "step": 10193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.631148557807314e-07,
+      "loss": 0.905,
+      "step": 10194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61940553848859e-07,
+      "loss": 0.8615,
+      "step": 10195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.607669320788182e-07,
+      "loss": 0.8926,
+      "step": 10196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.595939905589303e-07,
+      "loss": 0.9002,
+      "step": 10197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.584217293774656e-07,
+      "loss": 0.9102,
+      "step": 10198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.572501486226426e-07,
+      "loss": 0.8721,
+      "step": 10199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.56079248382632e-07,
+      "loss": 0.9148,
+      "step": 10200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.549090287455464e-07,
+      "loss": 0.8645,
+      "step": 10201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.537394897994512e-07,
+      "loss": 0.8925,
+      "step": 10202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.525706316323646e-07,
+      "loss": 0.9228,
+      "step": 10203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.514024543322442e-07,
+      "loss": 0.9105,
+      "step": 10204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.502349579870029e-07,
+      "loss": 0.8843,
+      "step": 10205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.490681426845028e-07,
+      "loss": 0.8913,
+      "step": 10206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.479020085125523e-07,
+      "loss": 0.9152,
+      "step": 10207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467365555589092e-07,
+      "loss": 0.8588,
+      "step": 10208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455717839112788e-07,
+      "loss": 0.8445,
+      "step": 10209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444076936573166e-07,
+      "loss": 0.9586,
+      "step": 10210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.432442848846291e-07,
+      "loss": 0.912,
+      "step": 10211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420815576807651e-07,
+      "loss": 0.909,
+      "step": 10212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.409195121332271e-07,
+      "loss": 0.8926,
+      "step": 10213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.397581483294682e-07,
+      "loss": 0.9278,
+      "step": 10214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.385974663568831e-07,
+      "loss": 0.8924,
+      "step": 10215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374374663028241e-07,
+      "loss": 0.901,
+      "step": 10216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.362781482545813e-07,
+      "loss": 0.8765,
+      "step": 10217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.351195122994027e-07,
+      "loss": 0.8744,
+      "step": 10218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.339615585244821e-07,
+      "loss": 0.8758,
+      "step": 10219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.328042870169606e-07,
+      "loss": 0.8739,
+      "step": 10220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.316476978639311e-07,
+      "loss": 0.8845,
+      "step": 10221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.304917911524303e-07,
+      "loss": 0.8955,
+      "step": 10222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.293365669694477e-07,
+      "loss": 0.9382,
+      "step": 10223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.281820254019213e-07,
+      "loss": 0.88,
+      "step": 10224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.270281665367331e-07,
+      "loss": 0.9008,
+      "step": 10225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258749904607201e-07,
+      "loss": 0.8572,
+      "step": 10226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.247224972606627e-07,
+      "loss": 0.8893,
+      "step": 10227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.235706870232942e-07,
+      "loss": 0.8479,
+      "step": 10228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.224195598352947e-07,
+      "loss": 0.9076,
+      "step": 10229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.212691157832887e-07,
+      "loss": 0.8756,
+      "step": 10230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.201193549538567e-07,
+      "loss": 0.8887,
+      "step": 10231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189702774335263e-07,
+      "loss": 0.882,
+      "step": 10232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.17821883308766e-07,
+      "loss": 2.3597,
+      "step": 10233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.16674172666e-07,
+      "loss": 0.9082,
+      "step": 10234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.155271455916038e-07,
+      "loss": 0.8642,
+      "step": 10235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.143808021718936e-07,
+      "loss": 0.8277,
+      "step": 10236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.132351424931396e-07,
+      "loss": 0.8389,
+      "step": 10237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.12090166641556e-07,
+      "loss": 0.8732,
+      "step": 10238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.109458747033106e-07,
+      "loss": 0.9058,
+      "step": 10239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.098022667645179e-07,
+      "loss": 0.8937,
+      "step": 10240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.086593429112378e-07,
+      "loss": 0.8849,
+      "step": 10241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.075171032294839e-07,
+      "loss": 0.8981,
+      "step": 10242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.063755478052172e-07,
+      "loss": 0.9088,
+      "step": 10243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.052346767243414e-07,
+      "loss": 0.9082,
+      "step": 10244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.040944900727167e-07,
+      "loss": 0.9231,
+      "step": 10245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029549879361454e-07,
+      "loss": 0.8778,
+      "step": 10246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.018161704003813e-07,
+      "loss": 0.9336,
+      "step": 10247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.006780375511293e-07,
+      "loss": 0.8694,
+      "step": 10248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.995405894740372e-07,
+      "loss": 0.8676,
+      "step": 10249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.984038262547046e-07,
+      "loss": 0.8747,
+      "step": 10250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.972677479786817e-07,
+      "loss": 0.9031,
+      "step": 10251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.961323547314594e-07,
+      "loss": 0.8802,
+      "step": 10252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.949976465984878e-07,
+      "loss": 0.8342,
+      "step": 10253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.938636236651521e-07,
+      "loss": 0.9002,
+      "step": 10254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.927302860167996e-07,
+      "loss": 2.4348,
+      "step": 10255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.915976337387211e-07,
+      "loss": 0.8672,
+      "step": 10256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.904656669161493e-07,
+      "loss": 0.8503,
+      "step": 10257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.893343856342751e-07,
+      "loss": 0.9249,
+      "step": 10258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.882037899782303e-07,
+      "loss": 0.9006,
+      "step": 10259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.870738800330991e-07,
+      "loss": 0.8653,
+      "step": 10260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.859446558839146e-07,
+      "loss": 0.8969,
+      "step": 10261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.848161176156545e-07,
+      "loss": 0.8976,
+      "step": 10262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.836882653132495e-07,
+      "loss": 0.8653,
+      "step": 10263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.825610990615773e-07,
+      "loss": 0.9092,
+      "step": 10264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814346189454603e-07,
+      "loss": 0.908,
+      "step": 10265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.80308825049675e-07,
+      "loss": 0.9339,
+      "step": 10266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.791837174589401e-07,
+      "loss": 0.8775,
+      "step": 10267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.780592962579271e-07,
+      "loss": 2.5591,
+      "step": 10268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.769355615312547e-07,
+      "loss": 0.906,
+      "step": 10269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.758125133634898e-07,
+      "loss": 0.8582,
+      "step": 10270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.746901518391493e-07,
+      "loss": 0.8466,
+      "step": 10271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.735684770426967e-07,
+      "loss": 0.8738,
+      "step": 10272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.72447489058541e-07,
+      "loss": 0.8853,
+      "step": 10273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.713271879710472e-07,
+      "loss": 0.8824,
+      "step": 10274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.702075738645165e-07,
+      "loss": 0.8889,
+      "step": 10275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69088646823213e-07,
+      "loss": 2.4564,
+      "step": 10276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.67970406931341e-07,
+      "loss": 0.9342,
+      "step": 10277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.668528542730504e-07,
+      "loss": 0.8889,
+      "step": 10278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.657359889324446e-07,
+      "loss": 0.8889,
+      "step": 10279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.646198109935755e-07,
+      "loss": 2.4557,
+      "step": 10280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.635043205404381e-07,
+      "loss": 0.8874,
+      "step": 10281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623895176569808e-07,
+      "loss": 0.9186,
+      "step": 10282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.612754024270975e-07,
+      "loss": 0.8332,
+      "step": 10283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601619749346335e-07,
+      "loss": 0.8688,
+      "step": 10284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590492352633794e-07,
+      "loss": 0.8697,
+      "step": 10285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.579371834970718e-07,
+      "loss": 0.8735,
+      "step": 10286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568258197194012e-07,
+      "loss": 0.8752,
+      "step": 10287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.557151440140055e-07,
+      "loss": 0.9125,
+      "step": 10288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.546051564644631e-07,
+      "loss": 0.8955,
+      "step": 10289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.534958571543117e-07,
+      "loss": 0.9074,
+      "step": 10290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523872461670291e-07,
+      "loss": 0.8543,
+      "step": 10291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.51279323586045e-07,
+      "loss": 0.892,
+      "step": 10292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.501720894947385e-07,
+      "loss": 0.9061,
+      "step": 10293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.490655439764306e-07,
+      "loss": 0.8828,
+      "step": 10294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47959687114398e-07,
+      "loss": 0.875,
+      "step": 10295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.468545189918609e-07,
+      "loss": 0.8664,
+      "step": 10296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.457500396919893e-07,
+      "loss": 0.8582,
+      "step": 10297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.446462492979035e-07,
+      "loss": 0.879,
+      "step": 10298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.435431478926648e-07,
+      "loss": 0.8633,
+      "step": 10299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.424407355592911e-07,
+      "loss": 0.8629,
+      "step": 10300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.413390123807441e-07,
+      "loss": 0.8681,
+      "step": 10301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.402379784399329e-07,
+      "loss": 0.8699,
+      "step": 10302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.391376338197165e-07,
+      "loss": 0.8743,
+      "step": 10303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.380379786029025e-07,
+      "loss": 0.8886,
+      "step": 10304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.369390128722455e-07,
+      "loss": 0.866,
+      "step": 10305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.358407367104504e-07,
+      "loss": 0.9137,
+      "step": 10306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.347431502001646e-07,
+      "loss": 0.8506,
+      "step": 10307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.336462534239886e-07,
+      "loss": 0.8771,
+      "step": 10308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.325500464644731e-07,
+      "loss": 0.9072,
+      "step": 10309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.314545294041055e-07,
+      "loss": 0.93,
+      "step": 10310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.303597023253396e-07,
+      "loss": 0.9225,
+      "step": 10311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.292655653105586e-07,
+      "loss": 0.9039,
+      "step": 10312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.281721184421054e-07,
+      "loss": 0.9069,
+      "step": 10313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.270793618022688e-07,
+      "loss": 0.897,
+      "step": 10314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259872954732818e-07,
+      "loss": 0.8929,
+      "step": 10315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.248959195373296e-07,
+      "loss": 0.9068,
+      "step": 10316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.238052340765435e-07,
+      "loss": 0.9289,
+      "step": 10317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.227152391730042e-07,
+      "loss": 0.9223,
+      "step": 10318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.216259349087407e-07,
+      "loss": 0.9222,
+      "step": 10319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.205373213657264e-07,
+      "loss": 0.8384,
+      "step": 10320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194493986258855e-07,
+      "loss": 0.8874,
+      "step": 10321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.183621667710928e-07,
+      "loss": 0.8593,
+      "step": 10322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172756258831638e-07,
+      "loss": 0.901,
+      "step": 10323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.161897760438697e-07,
+      "loss": 0.8838,
+      "step": 10324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151046173349242e-07,
+      "loss": 0.9028,
+      "step": 10325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.14020149837994e-07,
+      "loss": 0.8712,
+      "step": 10326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.129363736346907e-07,
+      "loss": 0.8987,
+      "step": 10327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.118532888065711e-07,
+      "loss": 0.8839,
+      "step": 10328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.107708954351456e-07,
+      "loss": 0.8698,
+      "step": 10329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.096891936018691e-07,
+      "loss": 0.8593,
+      "step": 10330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.086081833881453e-07,
+      "loss": 0.8168,
+      "step": 10331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.075278648753293e-07,
+      "loss": 0.8619,
+      "step": 10332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.064482381447147e-07,
+      "loss": 0.8981,
+      "step": 10333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.053693032775534e-07,
+      "loss": 0.888,
+      "step": 10334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.042910603550413e-07,
+      "loss": 0.8637,
+      "step": 10335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.032135094583193e-07,
+      "loss": 2.4382,
+      "step": 10336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.021366506684802e-07,
+      "loss": 0.8893,
+      "step": 10337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.010604840665626e-07,
+      "loss": 0.8857,
+      "step": 10338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.999850097335537e-07,
+      "loss": 0.8604,
+      "step": 10339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.989102277503913e-07,
+      "loss": 0.9243,
+      "step": 10340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.978361381979549e-07,
+      "loss": 0.8944,
+      "step": 10341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.967627411570766e-07,
+      "loss": 0.9036,
+      "step": 10342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.956900367085374e-07,
+      "loss": 0.8812,
+      "step": 10343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.946180249330604e-07,
+      "loss": 0.874,
+      "step": 10344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.9354670591132e-07,
+      "loss": 0.8578,
+      "step": 10345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.924760797239439e-07,
+      "loss": 0.9199,
+      "step": 10346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.914061464514977e-07,
+      "loss": 0.9076,
+      "step": 10347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.903369061745014e-07,
+      "loss": 0.889,
+      "step": 10348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.892683589734196e-07,
+      "loss": 0.8844,
+      "step": 10349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.882005049286668e-07,
+      "loss": 0.9067,
+      "step": 10350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.871333441206053e-07,
+      "loss": 0.8613,
+      "step": 10351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.860668766295432e-07,
+      "loss": 0.9082,
+      "step": 10352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.850011025357385e-07,
+      "loss": 0.901,
+      "step": 10353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.839360219193992e-07,
+      "loss": 0.8726,
+      "step": 10354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.828716348606735e-07,
+      "loss": 0.8709,
+      "step": 10355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.818079414396662e-07,
+      "loss": 0.9006,
+      "step": 10356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.807449417364222e-07,
+      "loss": 0.8813,
+      "step": 10357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.796826358309384e-07,
+      "loss": 0.8775,
+      "step": 10358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.786210238031622e-07,
+      "loss": 0.8367,
+      "step": 10359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.775601057329818e-07,
+      "loss": 0.8728,
+      "step": 10360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.764998817002412e-07,
+      "loss": 0.8698,
+      "step": 10361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.75440351784723e-07,
+      "loss": 0.8986,
+      "step": 10362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.743815160661649e-07,
+      "loss": 0.8428,
+      "step": 10363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.733233746242519e-07,
+      "loss": 0.8906,
+      "step": 10364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.722659275386102e-07,
+      "loss": 0.8654,
+      "step": 10365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.712091748888184e-07,
+      "loss": 0.8618,
+      "step": 10366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.701531167544096e-07,
+      "loss": 0.9139,
+      "step": 10367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.690977532148514e-07,
+      "loss": 0.8926,
+      "step": 10368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.680430843495679e-07,
+      "loss": 0.9191,
+      "step": 10369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.669891102379268e-07,
+      "loss": 0.8938,
+      "step": 10370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.659358309592458e-07,
+      "loss": 0.9146,
+      "step": 10371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.648832465927903e-07,
+      "loss": 0.9208,
+      "step": 10372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.638313572177736e-07,
+      "loss": 0.9222,
+      "step": 10373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.627801629133536e-07,
+      "loss": 0.8981,
+      "step": 10374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.617296637586424e-07,
+      "loss": 0.8902,
+      "step": 10375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606798598326915e-07,
+      "loss": 0.9114,
+      "step": 10376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.596307512145073e-07,
+      "loss": 0.8659,
+      "step": 10377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.58582337983037e-07,
+      "loss": 0.9083,
+      "step": 10378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.575346202171819e-07,
+      "loss": 0.9385,
+      "step": 10379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564875979957864e-07,
+      "loss": 0.9068,
+      "step": 10380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.554412713976478e-07,
+      "loss": 0.9282,
+      "step": 10381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543956405015041e-07,
+      "loss": 0.8633,
+      "step": 10382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.533507053860489e-07,
+      "loss": 0.9381,
+      "step": 10383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.523064661299151e-07,
+      "loss": 0.8993,
+      "step": 10384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.512629228116896e-07,
+      "loss": 0.9002,
+      "step": 10385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.502200755098999e-07,
+      "loss": 0.8522,
+      "step": 10386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.491779243030317e-07,
+      "loss": 2.5339,
+      "step": 10387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.481364692695114e-07,
+      "loss": 0.8895,
+      "step": 10388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.47095710487712e-07,
+      "loss": 0.9454,
+      "step": 10389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46055648035956e-07,
+      "loss": 0.8851,
+      "step": 10390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.450162819925166e-07,
+      "loss": 0.8914,
+      "step": 10391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.439776124356068e-07,
+      "loss": 0.8825,
+      "step": 10392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429396394433952e-07,
+      "loss": 0.9065,
+      "step": 10393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.419023630939948e-07,
+      "loss": 0.8804,
+      "step": 10394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.408657834654653e-07,
+      "loss": 0.8737,
+      "step": 10395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398299006358156e-07,
+      "loss": 0.9282,
+      "step": 10396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.38794714683e-07,
+      "loss": 0.8283,
+      "step": 10397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.377602256849248e-07,
+      "loss": 0.9046,
+      "step": 10398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.367264337194357e-07,
+      "loss": 0.8688,
+      "step": 10399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.356933388643339e-07,
+      "loss": 0.8824,
+      "step": 10400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.34660941197366e-07,
+      "loss": 0.8872,
+      "step": 10401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.336292407962242e-07,
+      "loss": 0.8777,
+      "step": 10402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.325982377385499e-07,
+      "loss": 0.9101,
+      "step": 10403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.315679321019332e-07,
+      "loss": 0.8613,
+      "step": 10404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305383239639052e-07,
+      "loss": 0.8615,
+      "step": 10405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.295094134019554e-07,
+      "loss": 0.9177,
+      "step": 10406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284812004935083e-07,
+      "loss": 0.8696,
+      "step": 10407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.274536853159475e-07,
+      "loss": 0.8923,
+      "step": 10408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.264268679465991e-07,
+      "loss": 0.8849,
+      "step": 10409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.254007484627335e-07,
+      "loss": 0.929,
+      "step": 10410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.243753269415721e-07,
+      "loss": 0.8887,
+      "step": 10411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.233506034602866e-07,
+      "loss": 0.8986,
+      "step": 10412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.223265780959887e-07,
+      "loss": 0.8679,
+      "step": 10413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.213032509257422e-07,
+      "loss": 0.8682,
+      "step": 10414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.202806220265601e-07,
+      "loss": 0.9001,
+      "step": 10415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.192586914753996e-07,
+      "loss": 0.8532,
+      "step": 10416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.182374593491682e-07,
+      "loss": 0.8911,
+      "step": 10417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.172169257247152e-07,
+      "loss": 0.8856,
+      "step": 10418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16197090678843e-07,
+      "loss": 0.8872,
+      "step": 10419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.15177954288302e-07,
+      "loss": 0.8646,
+      "step": 10420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.141595166297833e-07,
+      "loss": 0.9054,
+      "step": 10421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.131417777799321e-07,
+      "loss": 0.9168,
+      "step": 10422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.121247378153384e-07,
+      "loss": 0.8297,
+      "step": 10423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.111083968125399e-07,
+      "loss": 2.591,
+      "step": 10424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.10092754848023e-07,
+      "loss": 0.8622,
+      "step": 10425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090778119982167e-07,
+      "loss": 0.8724,
+      "step": 10426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080635683395032e-07,
+      "loss": 0.8994,
+      "step": 10427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.070500239482092e-07,
+      "loss": 0.8271,
+      "step": 10428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060371789006104e-07,
+      "loss": 0.838,
+      "step": 10429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.050250332729281e-07,
+      "loss": 0.887,
+      "step": 10430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.040135871413301e-07,
+      "loss": 0.9039,
+      "step": 10431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.030028405819344e-07,
+      "loss": 0.8557,
+      "step": 10432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.01992793670806e-07,
+      "loss": 2.2968,
+      "step": 10433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009834464839538e-07,
+      "loss": 0.8692,
+      "step": 10434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.999747990973382e-07,
+      "loss": 0.8766,
+      "step": 10435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.989668515868642e-07,
+      "loss": 0.8431,
+      "step": 10436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.979596040283864e-07,
+      "loss": 0.8931,
+      "step": 10437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969530564977067e-07,
+      "loss": 0.9116,
+      "step": 10438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.959472090705699e-07,
+      "loss": 0.8691,
+      "step": 10439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.949420618226721e-07,
+      "loss": 0.8953,
+      "step": 10440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.939376148296595e-07,
+      "loss": 0.9128,
+      "step": 10441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.92933868167116e-07,
+      "loss": 0.8791,
+      "step": 10442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.919308219105847e-07,
+      "loss": 0.8773,
+      "step": 10443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.909284761355461e-07,
+      "loss": 0.8532,
+      "step": 10444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.899268309174345e-07,
+      "loss": 0.8844,
+      "step": 10445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.889258863316284e-07,
+      "loss": 0.8664,
+      "step": 10446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.87925642453453e-07,
+      "loss": 2.4921,
+      "step": 10447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.869260993581828e-07,
+      "loss": 0.9063,
+      "step": 10448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.859272571210384e-07,
+      "loss": 0.9109,
+      "step": 10449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.849291158171878e-07,
+      "loss": 0.8961,
+      "step": 10450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839316755217496e-07,
+      "loss": 0.8704,
+      "step": 10451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.829349363097803e-07,
+      "loss": 0.8993,
+      "step": 10452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.819388982562947e-07,
+      "loss": 0.8886,
+      "step": 10453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.809435614362503e-07,
+      "loss": 0.88,
+      "step": 10454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.799489259245473e-07,
+      "loss": 0.8965,
+      "step": 10455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.789549917960403e-07,
+      "loss": 0.9041,
+      "step": 10456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.779617591255272e-07,
+      "loss": 0.9222,
+      "step": 10457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.769692279877538e-07,
+      "loss": 0.8752,
+      "step": 10458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.759773984574158e-07,
+      "loss": 0.8779,
+      "step": 10459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.749862706091492e-07,
+      "loss": 0.8585,
+      "step": 10460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.739958445175454e-07,
+      "loss": 2.5321,
+      "step": 10461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.73006120257137e-07,
+      "loss": 0.9087,
+      "step": 10462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.720170979024065e-07,
+      "loss": 2.3883,
+      "step": 10463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.710287775277857e-07,
+      "loss": 0.8765,
+      "step": 10464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.700411592076472e-07,
+      "loss": 0.8948,
+      "step": 10465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.690542430163149e-07,
+      "loss": 0.8478,
+      "step": 10466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.680680290280628e-07,
+      "loss": 0.8502,
+      "step": 10467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.670825173171047e-07,
+      "loss": 0.9059,
+      "step": 10468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.660977079576081e-07,
+      "loss": 0.8985,
+      "step": 10469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.651136010236836e-07,
+      "loss": 0.874,
+      "step": 10470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.641301965893909e-07,
+      "loss": 0.8819,
+      "step": 10471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.631474947287386e-07,
+      "loss": 0.8384,
+      "step": 10472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.621654955156765e-07,
+      "loss": 0.8805,
+      "step": 10473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611841990241075e-07,
+      "loss": 0.8749,
+      "step": 10474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.602036053278804e-07,
+      "loss": 0.894,
+      "step": 10475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.592237145007862e-07,
+      "loss": 0.8769,
+      "step": 10476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.582445266165693e-07,
+      "loss": 0.8917,
+      "step": 10477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.572660417489207e-07,
+      "loss": 0.8376,
+      "step": 10478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.562882599714726e-07,
+      "loss": 0.9125,
+      "step": 10479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.553111813578128e-07,
+      "loss": 0.8552,
+      "step": 10480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.543348059814669e-07,
+      "loss": 0.8699,
+      "step": 10481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.533591339159151e-07,
+      "loss": 0.9086,
+      "step": 10482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.523841652345808e-07,
+      "loss": 0.8857,
+      "step": 10483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.514099000108354e-07,
+      "loss": 0.8304,
+      "step": 10484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.50436338317999e-07,
+      "loss": 0.8744,
+      "step": 10485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.494634802293376e-07,
+      "loss": 2.5473,
+      "step": 10486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.484913258180615e-07,
+      "loss": 0.893,
+      "step": 10487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.47519875157333e-07,
+      "loss": 0.8765,
+      "step": 10488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.465491283202563e-07,
+      "loss": 0.8686,
+      "step": 10489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.455790853798871e-07,
+      "loss": 0.8428,
+      "step": 10490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.446097464092249e-07,
+      "loss": 0.8991,
+      "step": 10491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.436411114812191e-07,
+      "loss": 0.865,
+      "step": 10492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.426731806687637e-07,
+      "loss": 0.8812,
+      "step": 10493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.417059540447012e-07,
+      "loss": 0.8736,
+      "step": 10494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.407394316818205e-07,
+      "loss": 0.9004,
+      "step": 10495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.397736136528576e-07,
+      "loss": 0.9271,
+      "step": 10496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.388085000304933e-07,
+      "loss": 0.9278,
+      "step": 10497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.378440908873584e-07,
+      "loss": 0.8516,
+      "step": 10498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.368803862960338e-07,
+      "loss": 0.9674,
+      "step": 10499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.35917386329038e-07,
+      "loss": 0.918,
+      "step": 10500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.349550910588464e-07,
+      "loss": 0.8656,
+      "step": 10501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.339935005578723e-07,
+      "loss": 0.9438,
+      "step": 10502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.330326148984822e-07,
+      "loss": 0.891,
+      "step": 10503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.320724341529894e-07,
+      "loss": 0.9147,
+      "step": 10504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.311129583936504e-07,
+      "loss": 0.8941,
+      "step": 10505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.301541876926731e-07,
+      "loss": 0.9078,
+      "step": 10506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.291961221222087e-07,
+      "loss": 0.9123,
+      "step": 10507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.282387617543572e-07,
+      "loss": 0.8555,
+      "step": 10508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272821066611645e-07,
+      "loss": 0.8239,
+      "step": 10509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.263261569146239e-07,
+      "loss": 0.9053,
+      "step": 10510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.253709125866758e-07,
+      "loss": 0.8616,
+      "step": 10511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.244163737492071e-07,
+      "loss": 0.8995,
+      "step": 10512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.234625404740524e-07,
+      "loss": 0.903,
+      "step": 10513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225094128329934e-07,
+      "loss": 0.8859,
+      "step": 10514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.215569908977581e-07,
+      "loss": 0.8917,
+      "step": 10515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.206052747400182e-07,
+      "loss": 0.8352,
+      "step": 10516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.196542644314007e-07,
+      "loss": 0.8805,
+      "step": 10517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.187039600434663e-07,
+      "loss": 0.8836,
+      "step": 10518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.177543616477377e-07,
+      "loss": 0.9068,
+      "step": 10519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.168054693156767e-07,
+      "loss": 0.9073,
+      "step": 10520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.158572831186882e-07,
+      "loss": 0.884,
+      "step": 10521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149098031281309e-07,
+      "loss": 0.8226,
+      "step": 10522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.139630294153098e-07,
+      "loss": 0.9103,
+      "step": 10523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.130169620514692e-07,
+      "loss": 0.8977,
+      "step": 10524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120716011078087e-07,
+      "loss": 0.8656,
+      "step": 10525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111269466554726e-07,
+      "loss": 0.8902,
+      "step": 10526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101829987655494e-07,
+      "loss": 0.8591,
+      "step": 10527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.092397575090791e-07,
+      "loss": 0.8628,
+      "step": 10528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.082972229570416e-07,
+      "loss": 0.8497,
+      "step": 10529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07355395180369e-07,
+      "loss": 0.85,
+      "step": 10530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.064142742499412e-07,
+      "loss": 0.8885,
+      "step": 10531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054738602365795e-07,
+      "loss": 0.9167,
+      "step": 10532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.04534153211056e-07,
+      "loss": 0.9363,
+      "step": 10533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035951532440887e-07,
+      "loss": 0.8765,
+      "step": 10534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.026568604063432e-07,
+      "loss": 0.9063,
+      "step": 10535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.017192747684309e-07,
+      "loss": 2.4609,
+      "step": 10536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.007823964009085e-07,
+      "loss": 0.907,
+      "step": 10537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.998462253742831e-07,
+      "loss": 2.4643,
+      "step": 10538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.989107617590017e-07,
+      "loss": 0.8707,
+      "step": 10539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.979760056254691e-07,
+      "loss": 0.8682,
+      "step": 10540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9704195704403e-07,
+      "loss": 0.9174,
+      "step": 10541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.961086160849716e-07,
+      "loss": 0.8601,
+      "step": 10542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.951759828185377e-07,
+      "loss": 0.8546,
+      "step": 10543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942440573149122e-07,
+      "loss": 0.866,
+      "step": 10544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.933128396442256e-07,
+      "loss": 0.8852,
+      "step": 10545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.923823298765597e-07,
+      "loss": 0.8781,
+      "step": 10546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.914525280819383e-07,
+      "loss": 0.8798,
+      "step": 10547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.905234343303356e-07,
+      "loss": 0.9017,
+      "step": 10548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.895950486916701e-07,
+      "loss": 0.8997,
+      "step": 10549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.886673712358071e-07,
+      "loss": 0.9235,
+      "step": 10550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877404020325594e-07,
+      "loss": 0.8863,
+      "step": 10551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.868141411516892e-07,
+      "loss": 0.9064,
+      "step": 10552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.858885886628973e-07,
+      "loss": 0.8982,
+      "step": 10553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849637446358403e-07,
+      "loss": 0.8519,
+      "step": 10554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.840396091401157e-07,
+      "loss": 0.8406,
+      "step": 10555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.831161822452703e-07,
+      "loss": 0.9015,
+      "step": 10556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821934640207993e-07,
+      "loss": 0.8896,
+      "step": 10557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.812714545361375e-07,
+      "loss": 0.9016,
+      "step": 10558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.803501538606737e-07,
+      "loss": 0.8363,
+      "step": 10559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794295620637403e-07,
+      "loss": 0.8665,
+      "step": 10560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.785096792146161e-07,
+      "loss": 0.8543,
+      "step": 10561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775905053825292e-07,
+      "loss": 0.8994,
+      "step": 10562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766720406366499e-07,
+      "loss": 0.8935,
+      "step": 10563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.757542850460984e-07,
+      "loss": 0.862,
+      "step": 10564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748372386799417e-07,
+      "loss": 0.9018,
+      "step": 10565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.739209016071901e-07,
+      "loss": 0.9142,
+      "step": 10566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.730052738968039e-07,
+      "loss": 0.8992,
+      "step": 10567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.720903556176893e-07,
+      "loss": 2.4965,
+      "step": 10568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.711761468386989e-07,
+      "loss": 0.8935,
+      "step": 10569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.702626476286322e-07,
+      "loss": 0.8985,
+      "step": 10570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.69349858056234e-07,
+      "loss": 0.8488,
+      "step": 10571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.684377781901951e-07,
+      "loss": 0.9319,
+      "step": 10572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.675264080991583e-07,
+      "loss": 0.855,
+      "step": 10573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.666157478517031e-07,
+      "loss": 0.8838,
+      "step": 10574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.657057975163682e-07,
+      "loss": 0.916,
+      "step": 10575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.647965571616276e-07,
+      "loss": 0.8971,
+      "step": 10576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.638880268559077e-07,
+      "loss": 0.9187,
+      "step": 10577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.629802066675815e-07,
+      "loss": 0.8958,
+      "step": 10578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.620730966649657e-07,
+      "loss": 0.8954,
+      "step": 10579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.8669,
+      "step": 10580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.602610074898707e-07,
+      "loss": 0.8147,
+      "step": 10581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.593560284537625e-07,
+      "loss": 0.9027,
+      "step": 10582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.584517598761041e-07,
+      "loss": 0.8866,
+      "step": 10583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.575482018249456e-07,
+      "loss": 0.8712,
+      "step": 10584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.566453543682848e-07,
+      "loss": 0.8688,
+      "step": 10585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.557432175740674e-07,
+      "loss": 0.8434,
+      "step": 10586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.548417915101811e-07,
+      "loss": 0.8702,
+      "step": 10587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.539410762444641e-07,
+      "loss": 0.8333,
+      "step": 10588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.53041071844701e-07,
+      "loss": 0.8644,
+      "step": 10589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521417783786209e-07,
+      "loss": 0.8747,
+      "step": 10590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.51243195913902e-07,
+      "loss": 0.8648,
+      "step": 10591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.503453245181656e-07,
+      "loss": 0.8582,
+      "step": 10592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.494481642589811e-07,
+      "loss": 0.8344,
+      "step": 10593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.485517152038655e-07,
+      "loss": 0.9335,
+      "step": 10594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.476559774202805e-07,
+      "loss": 0.8698,
+      "step": 10595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467609509756366e-07,
+      "loss": 0.8461,
+      "step": 10596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.458666359372899e-07,
+      "loss": 0.8955,
+      "step": 10597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.449730323725388e-07,
+      "loss": 0.8953,
+      "step": 10598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.440801403486351e-07,
+      "loss": 0.9253,
+      "step": 10599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.431879599327716e-07,
+      "loss": 0.9206,
+      "step": 10600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.422964911920903e-07,
+      "loss": 0.888,
+      "step": 10601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.414057341936785e-07,
+      "loss": 0.8658,
+      "step": 10602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.405156890045704e-07,
+      "loss": 0.8697,
+      "step": 10603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.396263556917491e-07,
+      "loss": 0.8498,
+      "step": 10604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.387377343221379e-07,
+      "loss": 0.9035,
+      "step": 10605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.37849824962613e-07,
+      "loss": 0.9151,
+      "step": 10606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369626276799934e-07,
+      "loss": 0.9026,
+      "step": 10607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.360761425410443e-07,
+      "loss": 0.8549,
+      "step": 10608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351903696124771e-07,
+      "loss": 0.9099,
+      "step": 10609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.343053089609573e-07,
+      "loss": 0.936,
+      "step": 10610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.334209606530837e-07,
+      "loss": 0.86,
+      "step": 10611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.325373247554122e-07,
+      "loss": 0.8627,
+      "step": 10612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.316544013344393e-07,
+      "loss": 0.8796,
+      "step": 10613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.307721904566088e-07,
+      "loss": 0.8731,
+      "step": 10614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.29890692188314e-07,
+      "loss": 0.8859,
+      "step": 10615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.290099065958898e-07,
+      "loss": 0.9057,
+      "step": 10616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28129833745622e-07,
+      "loss": 0.8746,
+      "step": 10617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.27250473703742e-07,
+      "loss": 0.8631,
+      "step": 10618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263718265364226e-07,
+      "loss": 0.9187,
+      "step": 10619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254938923097907e-07,
+      "loss": 0.8912,
+      "step": 10620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246166710899114e-07,
+      "loss": 0.8754,
+      "step": 10621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.237401629428018e-07,
+      "loss": 0.9027,
+      "step": 10622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.228643679344236e-07,
+      "loss": 0.9185,
+      "step": 10623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219892861306863e-07,
+      "loss": 0.9209,
+      "step": 10624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.21114917597444e-07,
+      "loss": 0.8831,
+      "step": 10625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.202412624004982e-07,
+      "loss": 0.8681,
+      "step": 10626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193683206055933e-07,
+      "loss": 0.8631,
+      "step": 10627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184960922784266e-07,
+      "loss": 0.9174,
+      "step": 10628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.176245774846345e-07,
+      "loss": 0.9512,
+      "step": 10629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.167537762898023e-07,
+      "loss": 0.9282,
+      "step": 10630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158836887594687e-07,
+      "loss": 0.8721,
+      "step": 10631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150143149591058e-07,
+      "loss": 0.8447,
+      "step": 10632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.141456549541412e-07,
+      "loss": 0.9032,
+      "step": 10633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132777088099483e-07,
+      "loss": 0.8877,
+      "step": 10634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.124104765918403e-07,
+      "loss": 0.898,
+      "step": 10635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.115439583650827e-07,
+      "loss": 0.9213,
+      "step": 10636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.106781541948868e-07,
+      "loss": 0.8899,
+      "step": 10637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098130641464083e-07,
+      "loss": 0.8762,
+      "step": 10638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.089486882847505e-07,
+      "loss": 0.8844,
+      "step": 10639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.080850266749604e-07,
+      "loss": 0.8857,
+      "step": 10640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.072220793820349e-07,
+      "loss": 0.9159,
+      "step": 10641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06359846470913e-07,
+      "loss": 0.8764,
+      "step": 10642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.054983280064851e-07,
+      "loss": 0.873,
+      "step": 10643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.046375240535828e-07,
+      "loss": 0.928,
+      "step": 10644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.037774346769875e-07,
+      "loss": 0.9035,
+      "step": 10645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.029180599414252e-07,
+      "loss": 0.8975,
+      "step": 10646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.020593999115686e-07,
+      "loss": 0.9004,
+      "step": 10647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.012014546520361e-07,
+      "loss": 0.8928,
+      "step": 10648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.003442242273937e-07,
+      "loss": 0.8615,
+      "step": 10649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.994877087021488e-07,
+      "loss": 0.8633,
+      "step": 10650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.98631908140762e-07,
+      "loss": 0.9098,
+      "step": 10651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977768226076385e-07,
+      "loss": 0.8719,
+      "step": 10652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.969224521671246e-07,
+      "loss": 0.9117,
+      "step": 10653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.960687968835165e-07,
+      "loss": 0.9101,
+      "step": 10654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.952158568210585e-07,
+      "loss": 0.9117,
+      "step": 10655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943636320439371e-07,
+      "loss": 0.8926,
+      "step": 10656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935121226162864e-07,
+      "loss": 0.8824,
+      "step": 10657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.926613286021876e-07,
+      "loss": 0.9398,
+      "step": 10658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.91811250065668e-07,
+      "loss": 0.8775,
+      "step": 10659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909618870707012e-07,
+      "loss": 0.8954,
+      "step": 10660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.901132396812036e-07,
+      "loss": 0.9071,
+      "step": 10661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.892653079610432e-07,
+      "loss": 0.917,
+      "step": 10662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8841809197403e-07,
+      "loss": 0.8884,
+      "step": 10663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875715917839208e-07,
+      "loss": 0.8677,
+      "step": 10664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867258074544201e-07,
+      "loss": 0.8878,
+      "step": 10665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.858807390491771e-07,
+      "loss": 0.8806,
+      "step": 10666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850363866317886e-07,
+      "loss": 0.8456,
+      "step": 10667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.841927502657973e-07,
+      "loss": 0.8962,
+      "step": 10668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833498300146888e-07,
+      "loss": 0.9009,
+      "step": 10669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.825076259418992e-07,
+      "loss": 0.8942,
+      "step": 10670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.816661381108068e-07,
+      "loss": 0.8781,
+      "step": 10671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.808253665847406e-07,
+      "loss": 0.8943,
+      "step": 10672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.799853114269726e-07,
+      "loss": 2.3441,
+      "step": 10673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.791459727007198e-07,
+      "loss": 2.4213,
+      "step": 10674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783073504691482e-07,
+      "loss": 0.8828,
+      "step": 10675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.774694447953699e-07,
+      "loss": 0.8403,
+      "step": 10676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7663225574243743e-07,
+      "loss": 0.8829,
+      "step": 10677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.757957833733573e-07,
+      "loss": 0.8279,
+      "step": 10678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.749600277510769e-07,
+      "loss": 0.8929,
+      "step": 10679
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7412498893849244e-07,
+      "loss": 0.8594,
+      "step": 10680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.732906669984461e-07,
+      "loss": 0.9258,
+      "step": 10681
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7245706199372186e-07,
+      "loss": 0.8595,
+      "step": 10682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.716241739870542e-07,
+      "loss": 0.9045,
+      "step": 10683
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7079200304112393e-07,
+      "loss": 0.8856,
+      "step": 10684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699605492185533e-07,
+      "loss": 0.9029,
+      "step": 10685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6912981258191547e-07,
+      "loss": 0.8955,
+      "step": 10686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.682997931937283e-07,
+      "loss": 0.8931,
+      "step": 10687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.674704911164529e-07,
+      "loss": 2.4956,
+      "step": 10688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6664190641250275e-07,
+      "loss": 0.8705,
+      "step": 10689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6581403914422896e-07,
+      "loss": 0.9252,
+      "step": 10690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.649868893739351e-07,
+      "loss": 0.9144,
+      "step": 10691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6416045716386806e-07,
+      "loss": 0.9139,
+      "step": 10692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6333474257622026e-07,
+      "loss": 0.8751,
+      "step": 10693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6250974567313423e-07,
+      "loss": 0.8637,
+      "step": 10694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6168546651669255e-07,
+      "loss": 0.9349,
+      "step": 10695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.608619051689267e-07,
+      "loss": 0.8902,
+      "step": 10696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.60039061691816e-07,
+      "loss": 0.9016,
+      "step": 10697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.592169361472798e-07,
+      "loss": 0.9094,
+      "step": 10698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5839552859719196e-07,
+      "loss": 0.9085,
+      "step": 10699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.575748391033641e-07,
+      "loss": 0.9135,
+      "step": 10700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5675486772756015e-07,
+      "loss": 0.9052,
+      "step": 10701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.559356145314875e-07,
+      "loss": 0.8818,
+      "step": 10702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5511707957679673e-07,
+      "loss": 0.9223,
+      "step": 10703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5429926292508863e-07,
+      "loss": 0.8859,
+      "step": 10704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5348216463790837e-07,
+      "loss": 2.4656,
+      "step": 10705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.526657847767446e-07,
+      "loss": 0.9024,
+      "step": 10706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.518501234030381e-07,
+      "loss": 0.9072,
+      "step": 10707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5103518057816873e-07,
+      "loss": 0.8942,
+      "step": 10708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5022095636346516e-07,
+      "loss": 0.8906,
+      "step": 10709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.494074508202062e-07,
+      "loss": 0.8809,
+      "step": 10710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4859466400960615e-07,
+      "loss": 0.8694,
+      "step": 10711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.477825959928361e-07,
+      "loss": 0.8826,
+      "step": 10712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4697124683100725e-07,
+      "loss": 0.9,
+      "step": 10713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.461606165851762e-07,
+      "loss": 0.8798,
+      "step": 10714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4535070531635195e-07,
+      "loss": 0.8922,
+      "step": 10715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.445415130854791e-07,
+      "loss": 0.8614,
+      "step": 10716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.437330399534556e-07,
+      "loss": 0.8638,
+      "step": 10717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4292528598112504e-07,
+      "loss": 0.8843,
+      "step": 10718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.421182512292732e-07,
+      "loss": 0.8871,
+      "step": 10719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4131193575863373e-07,
+      "loss": 0.8905,
+      "step": 10720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4050633962988697e-07,
+      "loss": 0.8316,
+      "step": 10721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.397014629036578e-07,
+      "loss": 0.9045,
+      "step": 10722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.388973056405199e-07,
+      "loss": 0.8588,
+      "step": 10723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3809386790098606e-07,
+      "loss": 0.8757,
+      "step": 10724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3729114974552124e-07,
+      "loss": 0.9153,
+      "step": 10725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3648915123453485e-07,
+      "loss": 0.9008,
+      "step": 10726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3568787242838084e-07,
+      "loss": 0.9214,
+      "step": 10727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.348873133873599e-07,
+      "loss": 0.874,
+      "step": 10728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3408747417171935e-07,
+      "loss": 0.8589,
+      "step": 10729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3328835484165e-07,
+      "loss": 0.8749,
+      "step": 10730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3248995545729035e-07,
+      "loss": 0.8853,
+      "step": 10731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.316922760787223e-07,
+      "loss": 2.4424,
+      "step": 10732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3089531676597796e-07,
+      "loss": 0.8851,
+      "step": 10733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3009907757903146e-07,
+      "loss": 0.9061,
+      "step": 10734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293035585778049e-07,
+      "loss": 0.8638,
+      "step": 10735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2850875982216376e-07,
+      "loss": 0.8663,
+      "step": 10736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2771468137192465e-07,
+      "loss": 0.8807,
+      "step": 10737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2692132328684077e-07,
+      "loss": 0.9018,
+      "step": 10738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.261286856266211e-07,
+      "loss": 2.5228,
+      "step": 10739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2533676845091243e-07,
+      "loss": 0.8794,
+      "step": 10740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2454557181931147e-07,
+      "loss": 0.8936,
+      "step": 10741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2375509579136274e-07,
+      "loss": 0.8728,
+      "step": 10742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2296534042654993e-07,
+      "loss": 0.884,
+      "step": 10743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.221763057843098e-07,
+      "loss": 2.3583,
+      "step": 10744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2138799192401824e-07,
+      "loss": 0.877,
+      "step": 10745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2060039890500114e-07,
+      "loss": 2.4092,
+      "step": 10746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1981352678653e-07,
+      "loss": 0.903,
+      "step": 10747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.190273756278207e-07,
+      "loss": 0.8972,
+      "step": 10748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.182419454880338e-07,
+      "loss": 0.8945,
+      "step": 10749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.174572364262808e-07,
+      "loss": 0.9322,
+      "step": 10750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.166732485016112e-07,
+      "loss": 0.9014,
+      "step": 10751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1588998177302666e-07,
+      "loss": 0.8965,
+      "step": 10752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1510743629947e-07,
+      "loss": 0.8791,
+      "step": 10753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.143256121398331e-07,
+      "loss": 0.8605,
+      "step": 10754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.135445093529522e-07,
+      "loss": 0.8593,
+      "step": 10755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1276412799761025e-07,
+      "loss": 0.902,
+      "step": 10756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.119844681325347e-07,
+      "loss": 0.8903,
+      "step": 10757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.112055298163997e-07,
+      "loss": 0.8406,
+      "step": 10758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.104273131078229e-07,
+      "loss": 0.8567,
+      "step": 10759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0964981806537077e-07,
+      "loss": 0.8363,
+      "step": 10760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0887304474755196e-07,
+      "loss": 0.8979,
+      "step": 10761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.08096993212822e-07,
+      "loss": 0.8928,
+      "step": 10762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0732166351958867e-07,
+      "loss": 0.9347,
+      "step": 10763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.065470557261941e-07,
+      "loss": 0.8638,
+      "step": 10764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0577316989093283e-07,
+      "loss": 0.8275,
+      "step": 10765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0500000607204604e-07,
+      "loss": 0.9103,
+      "step": 10766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0422756432771495e-07,
+      "loss": 0.8395,
+      "step": 10767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.034558447160719e-07,
+      "loss": 0.8489,
+      "step": 10768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0268484729519274e-07,
+      "loss": 0.8601,
+      "step": 10769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019145721230999e-07,
+      "loss": 0.8498,
+      "step": 10770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0114501925775927e-07,
+      "loss": 0.9338,
+      "step": 10771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.003761887570845e-07,
+      "loss": 0.9317,
+      "step": 10772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9960808067893485e-07,
+      "loss": 0.8587,
+      "step": 10773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.988406950811119e-07,
+      "loss": 0.8754,
+      "step": 10774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.980740320213672e-07,
+      "loss": 0.9038,
+      "step": 10775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.973080915573968e-07,
+      "loss": 0.8995,
+      "step": 10776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.965428737468413e-07,
+      "loss": 0.8983,
+      "step": 10777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9577837864728686e-07,
+      "loss": 0.8917,
+      "step": 10778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.950146063162674e-07,
+      "loss": 0.8572,
+      "step": 10779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9425155681125815e-07,
+      "loss": 0.9398,
+      "step": 10780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.934892301896853e-07,
+      "loss": 0.9028,
+      "step": 10781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9272762650891414e-07,
+      "loss": 0.8943,
+      "step": 10782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9196674582626437e-07,
+      "loss": 0.8627,
+      "step": 10783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.912065881989946e-07,
+      "loss": 0.8968,
+      "step": 10784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9044715368430797e-07,
+      "loss": 0.8706,
+      "step": 10785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.896884423393588e-07,
+      "loss": 0.9137,
+      "step": 10786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8893045422124467e-07,
+      "loss": 0.9045,
+      "step": 10787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8817318938700553e-07,
+      "loss": 0.8942,
+      "step": 10788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8741664789363143e-07,
+      "loss": 0.8618,
+      "step": 10789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.866608297980545e-07,
+      "loss": 0.9452,
+      "step": 10790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8590573515715604e-07,
+      "loss": 0.8447,
+      "step": 10791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8515136402776045e-07,
+      "loss": 0.8829,
+      "step": 10792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.843977164666379e-07,
+      "loss": 0.8882,
+      "step": 10793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836447925305031e-07,
+      "loss": 0.908,
+      "step": 10794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.828925922760196e-07,
+      "loss": 0.8848,
+      "step": 10795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.821411157597932e-07,
+      "loss": 0.8944,
+      "step": 10796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.813903630383764e-07,
+      "loss": 0.906,
+      "step": 10797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.806403341682674e-07,
+      "loss": 0.9056,
+      "step": 10798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7989102920591103e-07,
+      "loss": 2.4805,
+      "step": 10799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7914244820769664e-07,
+      "loss": 0.8661,
+      "step": 10800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.783945912299558e-07,
+      "loss": 0.886,
+      "step": 10801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7764745832897244e-07,
+      "loss": 0.8893,
+      "step": 10802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.769010495609693e-07,
+      "loss": 0.8768,
+      "step": 10803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7615536498211926e-07,
+      "loss": 0.9092,
+      "step": 10804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.754104046485396e-07,
+      "loss": 0.9221,
+      "step": 10805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7466616861629003e-07,
+      "loss": 0.954,
+      "step": 10806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.739226569413812e-07,
+      "loss": 0.9208,
+      "step": 10807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.731798696797651e-07,
+      "loss": 0.9077,
+      "step": 10808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.724378068873391e-07,
+      "loss": 0.9132,
+      "step": 10809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.716964686199487e-07,
+      "loss": 0.8856,
+      "step": 10810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.709558549333836e-07,
+      "loss": 0.8756,
+      "step": 10811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7021596588337814e-07,
+      "loss": 0.8702,
+      "step": 10812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.694768015256145e-07,
+      "loss": 0.895,
+      "step": 10813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.68738361915717e-07,
+      "loss": 0.905,
+      "step": 10814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6800064710925675e-07,
+      "loss": 0.9029,
+      "step": 10815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.672636571617538e-07,
+      "loss": 0.9268,
+      "step": 10816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6652739212866717e-07,
+      "loss": 0.8684,
+      "step": 10817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.657918520654058e-07,
+      "loss": 0.9291,
+      "step": 10818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6505703702732207e-07,
+      "loss": 0.8551,
+      "step": 10819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.643229470697174e-07,
+      "loss": 0.885,
+      "step": 10820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.635895822478341e-07,
+      "loss": 0.8701,
+      "step": 10821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6285694261686157e-07,
+      "loss": 0.9378,
+      "step": 10822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.621250282319355e-07,
+      "loss": 0.8984,
+      "step": 10823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.613938391481353e-07,
+      "loss": 0.8821,
+      "step": 10824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6066337542048803e-07,
+      "loss": 0.8509,
+      "step": 10825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5993363710396523e-07,
+      "loss": 0.9101,
+      "step": 10826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.592046242534819e-07,
+      "loss": 2.4957,
+      "step": 10827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5847633692390083e-07,
+      "loss": 0.8625,
+      "step": 10828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5774877517003036e-07,
+      "loss": 0.8766,
+      "step": 10829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.570219390466223e-07,
+      "loss": 0.9365,
+      "step": 10830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.562958286083751e-07,
+      "loss": 0.9488,
+      "step": 10831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.555704439099328e-07,
+      "loss": 0.9148,
+      "step": 10832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5484578500588395e-07,
+      "loss": 0.8637,
+      "step": 10833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.54121851950765e-07,
+      "loss": 0.9074,
+      "step": 10834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5339864479905227e-07,
+      "loss": 0.882,
+      "step": 10835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5267616360517345e-07,
+      "loss": 0.8756,
+      "step": 10836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5195440842349824e-07,
+      "loss": 0.9052,
+      "step": 10837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.512333793083411e-07,
+      "loss": 0.8886,
+      "step": 10838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5051307631396634e-07,
+      "loss": 0.8216,
+      "step": 10839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.497934994945795e-07,
+      "loss": 0.9275,
+      "step": 10840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.490746489043317e-07,
+      "loss": 0.8829,
+      "step": 10841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483565245973208e-07,
+      "loss": 2.4336,
+      "step": 10842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476391266275891e-07,
+      "loss": 0.8637,
+      "step": 10843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4692245504912346e-07,
+      "loss": 0.8191,
+      "step": 10844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462065099158596e-07,
+      "loss": 0.8929,
+      "step": 10845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.454912912816755e-07,
+      "loss": 0.9101,
+      "step": 10846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4477679920039474e-07,
+      "loss": 0.8559,
+      "step": 10847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4406303372578666e-07,
+      "loss": 0.8677,
+      "step": 10848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4334999491156483e-07,
+      "loss": 0.8772,
+      "step": 10849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.42637682811392e-07,
+      "loss": 0.8962,
+      "step": 10850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4192609747887073e-07,
+      "loss": 0.8676,
+      "step": 10851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4121523896755273e-07,
+      "loss": 0.8902,
+      "step": 10852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4050510733093287e-07,
+      "loss": 0.8485,
+      "step": 10853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.397957026224541e-07,
+      "loss": 0.9057,
+      "step": 10854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.390870248955025e-07,
+      "loss": 0.8849,
+      "step": 10855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.383790742034077e-07,
+      "loss": 0.9111,
+      "step": 10856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.376718505994492e-07,
+      "loss": 0.8508,
+      "step": 10857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.369653541368478e-07,
+      "loss": 0.8744,
+      "step": 10858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.362595848687711e-07,
+      "loss": 0.8741,
+      "step": 10859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.355545428483331e-07,
+      "loss": 0.8276,
+      "step": 10860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.348502281285937e-07,
+      "loss": 0.9248,
+      "step": 10861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.341466407625515e-07,
+      "loss": 0.8889,
+      "step": 10862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3344378080315876e-07,
+      "loss": 0.8551,
+      "step": 10863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3274164830330857e-07,
+      "loss": 0.8949,
+      "step": 10864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.320402433158387e-07,
+      "loss": 0.8973,
+      "step": 10865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3133956589353475e-07,
+      "loss": 0.9138,
+      "step": 10866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3063961608912675e-07,
+      "loss": 0.882,
+      "step": 10867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.299403939552903e-07,
+      "loss": 0.9144,
+      "step": 10868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2924189954464447e-07,
+      "loss": 0.8692,
+      "step": 10869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2854413290975384e-07,
+      "loss": 0.9378,
+      "step": 10870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.278470941031309e-07,
+      "loss": 0.914,
+      "step": 10871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2715078317723026e-07,
+      "loss": 0.8757,
+      "step": 10872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.264552001844523e-07,
+      "loss": 2.4991,
+      "step": 10873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2576034517714625e-07,
+      "loss": 2.5012,
+      "step": 10874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.250662182076003e-07,
+      "loss": 0.8413,
+      "step": 10875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2437281932805374e-07,
+      "loss": 0.8257,
+      "step": 10876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2368014859068706e-07,
+      "loss": 0.8579,
+      "step": 10877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2298820604762746e-07,
+      "loss": 0.8741,
+      "step": 10878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2229699175094665e-07,
+      "loss": 0.8502,
+      "step": 10879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2160650575266406e-07,
+      "loss": 0.8537,
+      "step": 10880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.209167481047415e-07,
+      "loss": 0.8665,
+      "step": 10881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2022771885908743e-07,
+      "loss": 0.9129,
+      "step": 10882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1953941806755265e-07,
+      "loss": 0.8892,
+      "step": 10883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1885184578193894e-07,
+      "loss": 0.8685,
+      "step": 10884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1816500205398726e-07,
+      "loss": 0.8919,
+      "step": 10885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.174788869353862e-07,
+      "loss": 0.9108,
+      "step": 10886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.167935004777711e-07,
+      "loss": 0.8934,
+      "step": 10887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.161088427327197e-07,
+      "loss": 0.8944,
+      "step": 10888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1542491375175623e-07,
+      "loss": 0.8677,
+      "step": 10889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1474171358635177e-07,
+      "loss": 0.8767,
+      "step": 10890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.140592422879174e-07,
+      "loss": 0.897,
+      "step": 10891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.133774999078165e-07,
+      "loss": 2.4851,
+      "step": 10892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1269648649735027e-07,
+      "loss": 0.8793,
+      "step": 10893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1201620210776885e-07,
+      "loss": 0.9295,
+      "step": 10894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.113366467902712e-07,
+      "loss": 0.8679,
+      "step": 10895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1065782059599423e-07,
+      "loss": 0.9015,
+      "step": 10896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0997972357602267e-07,
+      "loss": 0.8982,
+      "step": 10897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.09302355781389e-07,
+      "loss": 0.864,
+      "step": 10898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08625717263068e-07,
+      "loss": 0.8717,
+      "step": 10899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0794980807197895e-07,
+      "loss": 0.8956,
+      "step": 10900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0727462825898893e-07,
+      "loss": 0.9107,
+      "step": 10901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.066001778749084e-07,
+      "loss": 0.824,
+      "step": 10902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.059264569704945e-07,
+      "loss": 0.892,
+      "step": 10903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0525346559644675e-07,
+      "loss": 0.9248,
+      "step": 10904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.045812038034113e-07,
+      "loss": 0.8382,
+      "step": 10905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0390967164198093e-07,
+      "loss": 0.9536,
+      "step": 10906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0323886916268976e-07,
+      "loss": 0.8867,
+      "step": 10907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0256879641602067e-07,
+      "loss": 0.8584,
+      "step": 10908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0189945345240003e-07,
+      "loss": 0.8807,
+      "step": 10909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.012308403221997e-07,
+      "loss": 0.8988,
+      "step": 10910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0056295707573736e-07,
+      "loss": 0.8976,
+      "step": 10911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9989580376327155e-07,
+      "loss": 0.8938,
+      "step": 10912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9922938043501324e-07,
+      "loss": 0.8979,
+      "step": 10913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9856368714110905e-07,
+      "loss": 0.8652,
+      "step": 10914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9789872393165997e-07,
+      "loss": 0.9022,
+      "step": 10915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.972344908567082e-07,
+      "loss": 0.8975,
+      "step": 10916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9657098796623817e-07,
+      "loss": 0.8623,
+      "step": 10917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.959082153101833e-07,
+      "loss": 0.8556,
+      "step": 10918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.952461729384226e-07,
+      "loss": 0.9054,
+      "step": 10919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.945848609007751e-07,
+      "loss": 0.8687,
+      "step": 10920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9392427924700983e-07,
+      "loss": 0.878,
+      "step": 10921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.932644280268382e-07,
+      "loss": 0.8418,
+      "step": 10922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.926053072899182e-07,
+      "loss": 0.8813,
+      "step": 10923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9194691708585246e-07,
+      "loss": 0.8689,
+      "step": 10924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.912892574641879e-07,
+      "loss": 0.8937,
+      "step": 10925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9063232847441503e-07,
+      "loss": 0.8508,
+      "step": 10926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8997613016597536e-07,
+      "loss": 0.8686,
+      "step": 10927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.893206625882472e-07,
+      "loss": 0.9266,
+      "step": 10928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8866592579056e-07,
+      "loss": 0.8861,
+      "step": 10929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880119198221865e-07,
+      "loss": 0.866,
+      "step": 10930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.873586447323429e-07,
+      "loss": 0.9335,
+      "step": 10931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.867061005701921e-07,
+      "loss": 0.8755,
+      "step": 10932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.860542873848415e-07,
+      "loss": 0.8777,
+      "step": 10933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8540320522534415e-07,
+      "loss": 0.8773,
+      "step": 10934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8475285414069634e-07,
+      "loss": 0.8563,
+      "step": 10935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8410323417983996e-07,
+      "loss": 0.9031,
+      "step": 10936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.834543453916661e-07,
+      "loss": 0.8979,
+      "step": 10937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.828061878250021e-07,
+      "loss": 0.8675,
+      "step": 10938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.82158761528627e-07,
+      "loss": 0.9002,
+      "step": 10939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8151206655126605e-07,
+      "loss": 0.883,
+      "step": 10940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8086610294158157e-07,
+      "loss": 0.879,
+      "step": 10941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8022087074818906e-07,
+      "loss": 0.8923,
+      "step": 10942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.795763700196441e-07,
+      "loss": 0.9033,
+      "step": 10943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.789326008044491e-07,
+      "loss": 0.8821,
+      "step": 10944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.782895631510529e-07,
+      "loss": 0.8796,
+      "step": 10945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7764725710784477e-07,
+      "loss": 0.8695,
+      "step": 10946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7700568272316377e-07,
+      "loss": 0.8867,
+      "step": 10947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7636484004529006e-07,
+      "loss": 0.8261,
+      "step": 10948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7572472912245183e-07,
+      "loss": 0.8651,
+      "step": 10949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7508535000281943e-07,
+      "loss": 0.8252,
+      "step": 10950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.74446702734511e-07,
+      "loss": 0.9049,
+      "step": 10951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.738087873655859e-07,
+      "loss": 0.897,
+      "step": 10952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7317160394405464e-07,
+      "loss": 0.8956,
+      "step": 10953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725351525178643e-07,
+      "loss": 0.8588,
+      "step": 10954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.718994331349134e-07,
+      "loss": 0.8688,
+      "step": 10955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7126444584304357e-07,
+      "loss": 0.8855,
+      "step": 10956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706301906900399e-07,
+      "loss": 0.9097,
+      "step": 10957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.699966677236343e-07,
+      "loss": 0.8632,
+      "step": 10958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.693638769915019e-07,
+      "loss": 0.9199,
+      "step": 10959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.687318185412635e-07,
+      "loss": 0.8794,
+      "step": 10960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6810049242048665e-07,
+      "loss": 0.8583,
+      "step": 10961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6746989867667885e-07,
+      "loss": 0.8875,
+      "step": 10962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6684003735729766e-07,
+      "loss": 0.8927,
+      "step": 10963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.662109085097442e-07,
+      "loss": 0.8702,
+      "step": 10964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6558251218136154e-07,
+      "loss": 0.8633,
+      "step": 10965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64954848419442e-07,
+      "loss": 0.8996,
+      "step": 10966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6432791727121984e-07,
+      "loss": 0.9038,
+      "step": 10967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.637017187838731e-07,
+      "loss": 0.881,
+      "step": 10968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.630762530045305e-07,
+      "loss": 0.8698,
+      "step": 10969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6245151998025686e-07,
+      "loss": 0.8474,
+      "step": 10970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6182751975807e-07,
+      "loss": 0.8587,
+      "step": 10971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.61204252384929e-07,
+      "loss": 0.8943,
+      "step": 10972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6058171790773747e-07,
+      "loss": 0.9033,
+      "step": 10973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.599599163733446e-07,
+      "loss": 0.8964,
+      "step": 10974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.593388478285419e-07,
+      "loss": 0.9393,
+      "step": 10975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5871851232007197e-07,
+      "loss": 0.8825,
+      "step": 10976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5809890989461516e-07,
+      "loss": 0.8344,
+      "step": 10977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5748004059880204e-07,
+      "loss": 0.8672,
+      "step": 10978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5686190447920535e-07,
+      "loss": 0.8815,
+      "step": 10979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.562445015823423e-07,
+      "loss": 0.8793,
+      "step": 10980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.556278319546768e-07,
+      "loss": 0.8623,
+      "step": 10981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550118956426151e-07,
+      "loss": 0.8791,
+      "step": 10982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5439669269251123e-07,
+      "loss": 0.9131,
+      "step": 10983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.537822231506604e-07,
+      "loss": 0.8818,
+      "step": 10984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5316848706330666e-07,
+      "loss": 0.8908,
+      "step": 10985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5255548447663535e-07,
+      "loss": 0.8903,
+      "step": 10986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5194321543678067e-07,
+      "loss": 0.8783,
+      "step": 10987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5133167998981577e-07,
+      "loss": 0.8938,
+      "step": 10988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.853,
+      "step": 10989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5011081005859137e-07,
+      "loss": 0.8934,
+      "step": 10990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.495014756662084e-07,
+      "loss": 0.8719,
+      "step": 10991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4889287505047046e-07,
+      "loss": 0.8929,
+      "step": 10992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.482850082571786e-07,
+      "loss": 0.8749,
+      "step": 10993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4767787533207744e-07,
+      "loss": 0.8635,
+      "step": 10994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4707147632085815e-07,
+      "loss": 0.9365,
+      "step": 10995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4646581126915314e-07,
+      "loss": 0.8851,
+      "step": 10996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4586088022254374e-07,
+      "loss": 0.8559,
+      "step": 10997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.452566832265524e-07,
+      "loss": 0.8594,
+      "step": 10998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.446532203266505e-07,
+      "loss": 0.8728,
+      "step": 10999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.440504915682507e-07,
+      "loss": 0.8809,
+      "step": 11000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434484969967132e-07,
+      "loss": 0.8565,
+      "step": 11001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4284723665733755e-07,
+      "loss": 0.881,
+      "step": 11002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.422467105953752e-07,
+      "loss": 0.8815,
+      "step": 11003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.416469188560178e-07,
+      "loss": 0.9368,
+      "step": 11004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.410478614844003e-07,
+      "loss": 0.8825,
+      "step": 11005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4044953852561e-07,
+      "loss": 0.8876,
+      "step": 11006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.398519500246699e-07,
+      "loss": 0.8649,
+      "step": 11007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.392550960265527e-07,
+      "loss": 0.8371,
+      "step": 11008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.386589765761771e-07,
+      "loss": 0.8758,
+      "step": 11009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3806359171840044e-07,
+      "loss": 0.9003,
+      "step": 11010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3746894149803024e-07,
+      "loss": 0.8637,
+      "step": 11011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3687502595981847e-07,
+      "loss": 0.8789,
+      "step": 11012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3628184514845832e-07,
+      "loss": 0.8651,
+      "step": 11013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.356893991085918e-07,
+      "loss": 0.9101,
+      "step": 11014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3509768788480102e-07,
+      "loss": 0.8997,
+      "step": 11015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.345067115216182e-07,
+      "loss": 0.8917,
+      "step": 11016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3391647006351548e-07,
+      "loss": 0.832,
+      "step": 11017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.333269635549118e-07,
+      "loss": 0.8486,
+      "step": 11018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3273819204017057e-07,
+      "loss": 0.9171,
+      "step": 11019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3215015556360188e-07,
+      "loss": 0.9005,
+      "step": 11020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.315628541694559e-07,
+      "loss": 0.8344,
+      "step": 11021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3097628790193283e-07,
+      "loss": 0.8976,
+      "step": 11022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3039045680517292e-07,
+      "loss": 0.9264,
+      "step": 11023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.298053609232642e-07,
+      "loss": 0.8608,
+      "step": 11024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2922100030023709e-07,
+      "loss": 0.8558,
+      "step": 11025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2863737498006743e-07,
+      "loss": 0.8901,
+      "step": 11026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2805448500667904e-07,
+      "loss": 0.9152,
+      "step": 11027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2747233042393458e-07,
+      "loss": 0.8512,
+      "step": 11028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2689091127564454e-07,
+      "loss": 0.8777,
+      "step": 11029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.263102276055662e-07,
+      "loss": 0.9037,
+      "step": 11030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2573027945739568e-07,
+      "loss": 0.9278,
+      "step": 11031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2515106687478027e-07,
+      "loss": 0.9274,
+      "step": 11032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2457258990130627e-07,
+      "loss": 0.8691,
+      "step": 11033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2399484858050878e-07,
+      "loss": 2.5719,
+      "step": 11034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2341784295586533e-07,
+      "loss": 0.8851,
+      "step": 11035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228415730707989e-07,
+      "loss": 0.9016,
+      "step": 11036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2226603896867592e-07,
+      "loss": 0.8888,
+      "step": 11037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2169124069280955e-07,
+      "loss": 0.8831,
+      "step": 11038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.211171782864563e-07,
+      "loss": 0.8615,
+      "step": 11039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2054385179281602e-07,
+      "loss": 0.8615,
+      "step": 11040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1997126125503643e-07,
+      "loss": 0.7924,
+      "step": 11041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1939940671620641e-07,
+      "loss": 0.8443,
+      "step": 11042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1882828821936265e-07,
+      "loss": 0.9303,
+      "step": 11043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1825790580748296e-07,
+      "loss": 0.885,
+      "step": 11044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1768825952349304e-07,
+      "loss": 0.895,
+      "step": 11045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171193494102608e-07,
+      "loss": 0.8613,
+      "step": 11046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.165511755106009e-07,
+      "loss": 0.9295,
+      "step": 11047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1598373786727244e-07,
+      "loss": 0.8643,
+      "step": 11048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1541703652297463e-07,
+      "loss": 0.8986,
+      "step": 11049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1485107152035668e-07,
+      "loss": 0.8522,
+      "step": 11050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1428584290201116e-07,
+      "loss": 0.846,
+      "step": 11051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1372135071047296e-07,
+      "loss": 0.9574,
+      "step": 11052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1315759498822476e-07,
+      "loss": 0.8872,
+      "step": 11053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1259457577769038e-07,
+      "loss": 0.8522,
+      "step": 11054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1203229312124152e-07,
+      "loss": 0.8798,
+      "step": 11055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.114707470611932e-07,
+      "loss": 0.901,
+      "step": 11056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1090993763980272e-07,
+      "loss": 0.8816,
+      "step": 11057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1034986489927522e-07,
+      "loss": 0.9004,
+      "step": 11058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0979052888175922e-07,
+      "loss": 0.8277,
+      "step": 11059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.092319296293477e-07,
+      "loss": 0.8363,
+      "step": 11060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.086740671840781e-07,
+      "loss": 0.8836,
+      "step": 11061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.081169415879314e-07,
+      "loss": 0.8485,
+      "step": 11062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0756055288283616e-07,
+      "loss": 0.8541,
+      "step": 11063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.070049011106634e-07,
+      "loss": 0.8965,
+      "step": 11064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0644998631322744e-07,
+      "loss": 0.866,
+      "step": 11065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.058958085322893e-07,
+      "loss": 0.8844,
+      "step": 11066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0534236780955451e-07,
+      "loss": 0.8571,
+      "step": 11067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.047896641866709e-07,
+      "loss": 0.8878,
+      "step": 11068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0423769770523404e-07,
+      "loss": 0.8808,
+      "step": 11069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0368646840678074e-07,
+      "loss": 0.8239,
+      "step": 11070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0313597633279447e-07,
+      "loss": 0.906,
+      "step": 11071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.025862215247032e-07,
+      "loss": 0.8917,
+      "step": 11072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0203720402387828e-07,
+      "loss": 0.8824,
+      "step": 11073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0148892387163666e-07,
+      "loss": 0.8762,
+      "step": 11074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0094138110923867e-07,
+      "loss": 0.8794,
+      "step": 11075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0039457577789024e-07,
+      "loss": 0.8726,
+      "step": 11076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984850791874177e-07,
+      "loss": 0.8867,
+      "step": 11077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9930317757288708e-07,
+      "loss": 0.8758,
+      "step": 11078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9875858478136557e-07,
+      "loss": 0.8823,
+      "step": 11079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.982147295851611e-07,
+      "loss": 0.882,
+      "step": 11080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9767161202520092e-07,
+      "loss": 0.9042,
+      "step": 11081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.971292321423568e-07,
+      "loss": 0.8646,
+      "step": 11082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.965875899774472e-07,
+      "loss": 0.9187,
+      "step": 11083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9604668557123286e-07,
+      "loss": 0.9066,
+      "step": 11084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9550651896442118e-07,
+      "loss": 0.8758,
+      "step": 11085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.949670901976597e-07,
+      "loss": 0.8827,
+      "step": 11086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9442839931154477e-07,
+      "loss": 0.8922,
+      "step": 11087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9389044634661623e-07,
+      "loss": 0.8238,
+      "step": 11088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9335323134335726e-07,
+      "loss": 0.9041,
+      "step": 11089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.928167543421966e-07,
+      "loss": 0.8492,
+      "step": 11090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.922810153835064e-07,
+      "loss": 0.9201,
+      "step": 11091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9174601450760445e-07,
+      "loss": 0.8837,
+      "step": 11092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.912117517547518e-07,
+      "loss": 0.8955,
+      "step": 11093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9067822716515526e-07,
+      "loss": 0.876,
+      "step": 11094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9014544077896379e-07,
+      "loss": 0.8898,
+      "step": 11095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8961339263627422e-07,
+      "loss": 0.8793,
+      "step": 11096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8908208277712446e-07,
+      "loss": 0.853,
+      "step": 11097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8855151124150151e-07,
+      "loss": 0.8132,
+      "step": 11098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8802167806933003e-07,
+      "loss": 0.9117,
+      "step": 11099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.874925833004837e-07,
+      "loss": 0.8674,
+      "step": 11100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8696422697478066e-07,
+      "loss": 0.89,
+      "step": 11101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8643660913198137e-07,
+      "loss": 2.4615,
+      "step": 11102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8590972981179288e-07,
+      "loss": 0.8913,
+      "step": 11103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8538358905386577e-07,
+      "loss": 0.8438,
+      "step": 11104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.84858186897795e-07,
+      "loss": 0.8908,
+      "step": 11105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8433352338311894e-07,
+      "loss": 0.8999,
+      "step": 11106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8380959854932045e-07,
+      "loss": 0.8875,
+      "step": 11107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8328641243583022e-07,
+      "loss": 0.8636,
+      "step": 11108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8276396508201787e-07,
+      "loss": 0.901,
+      "step": 11109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8224225652720307e-07,
+      "loss": 0.8863,
+      "step": 11110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.817212868106455e-07,
+      "loss": 0.9025,
+      "step": 11111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.812010559715527e-07,
+      "loss": 0.8787,
+      "step": 11112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8068156404907223e-07,
+      "loss": 0.8339,
+      "step": 11113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.801628110823006e-07,
+      "loss": 0.9092,
+      "step": 11114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.796447971102744e-07,
+      "loss": 0.8639,
+      "step": 11115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7912752217197903e-07,
+      "loss": 0.8999,
+      "step": 11116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7861098630634232e-07,
+      "loss": 0.8518,
+      "step": 11117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7809518955223536e-07,
+      "loss": 0.9157,
+      "step": 11118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.775801319484749e-07,
+      "loss": 0.8858,
+      "step": 11119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.77065813533821e-07,
+      "loss": 0.9182,
+      "step": 11120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7655223434698054e-07,
+      "loss": 0.8774,
+      "step": 11121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7603939442660145e-07,
+      "loss": 0.8965,
+      "step": 11122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7552729381127732e-07,
+      "loss": 0.8872,
+      "step": 11123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.750159325395484e-07,
+      "loss": 0.9028,
+      "step": 11124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7450531064989728e-07,
+      "loss": 0.8622,
+      "step": 11125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7399542818074877e-07,
+      "loss": 0.8795,
+      "step": 11126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.734862851704766e-07,
+      "loss": 0.8701,
+      "step": 11127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.729778816573935e-07,
+      "loss": 0.8824,
+      "step": 11128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7247021767976213e-07,
+      "loss": 0.8642,
+      "step": 11129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7196329327578643e-07,
+      "loss": 0.8482,
+      "step": 11130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7145710848361475e-07,
+      "loss": 0.8246,
+      "step": 11131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7095166334133993e-07,
+      "loss": 0.8913,
+      "step": 11132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.704469578870005e-07,
+      "loss": 0.9423,
+      "step": 11133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6994299215857711e-07,
+      "loss": 0.8681,
+      "step": 11134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6943976619399615e-07,
+      "loss": 0.899,
+      "step": 11135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6893728003112842e-07,
+      "loss": 0.904,
+      "step": 11136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.68435533707787e-07,
+      "loss": 0.8803,
+      "step": 11137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6793452726173276e-07,
+      "loss": 0.8937,
+      "step": 11138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6743426073066892e-07,
+      "loss": 0.9145,
+      "step": 11139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6693473415224315e-07,
+      "loss": 0.8984,
+      "step": 11140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6643594756404757e-07,
+      "loss": 0.888,
+      "step": 11141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6593790100361662e-07,
+      "loss": 0.927,
+      "step": 11142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6544059450843363e-07,
+      "loss": 0.8921,
+      "step": 11143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6494402811592204e-07,
+      "loss": 0.8797,
+      "step": 11144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6444820186345078e-07,
+      "loss": 0.8432,
+      "step": 11145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.639531157883356e-07,
+      "loss": 0.9147,
+      "step": 11146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6345876992783117e-07,
+      "loss": 0.8718,
+      "step": 11147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.629651643191421e-07,
+      "loss": 0.8789,
+      "step": 11148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.624722989994143e-07,
+      "loss": 0.9161,
+      "step": 11149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6198017400573806e-07,
+      "loss": 0.8819,
+      "step": 11150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6148878937514822e-07,
+      "loss": 0.9166,
+      "step": 11151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609981451446241e-07,
+      "loss": 0.836,
+      "step": 11152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6050824135108944e-07,
+      "loss": 0.9007,
+      "step": 11153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6001907803141258e-07,
+      "loss": 0.8521,
+      "step": 11154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595306552224052e-07,
+      "loss": 0.8575,
+      "step": 11155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.590429729608245e-07,
+      "loss": 0.9484,
+      "step": 11156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5855603128336893e-07,
+      "loss": 0.8937,
+      "step": 11157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5806983022668476e-07,
+      "loss": 0.8943,
+      "step": 11158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5758436982736382e-07,
+      "loss": 2.4512,
+      "step": 11159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5709965012193463e-07,
+      "loss": 0.9078,
+      "step": 11160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5661567114687803e-07,
+      "loss": 0.9099,
+      "step": 11161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5613243293861713e-07,
+      "loss": 0.891,
+      "step": 11162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5564993553351394e-07,
+      "loss": 0.8493,
+      "step": 11163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5516817896788273e-07,
+      "loss": 0.8526,
+      "step": 11164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5468716327797673e-07,
+      "loss": 0.8892,
+      "step": 11165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5420688849999478e-07,
+      "loss": 0.8976,
+      "step": 11166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5372735467008125e-07,
+      "loss": 0.9046,
+      "step": 11167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.532485618243218e-07,
+      "loss": 0.8595,
+      "step": 11168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5277050999874865e-07,
+      "loss": 0.862,
+      "step": 11169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.522931992293397e-07,
+      "loss": 0.8849,
+      "step": 11170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5181662955201293e-07,
+      "loss": 0.8808,
+      "step": 11171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5134080100263292e-07,
+      "loss": 0.8907,
+      "step": 11172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5086571361700886e-07,
+      "loss": 0.9162,
+      "step": 11173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5039136743089434e-07,
+      "loss": 0.9009,
+      "step": 11174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4991776247998524e-07,
+      "loss": 0.9303,
+      "step": 11175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4944489879992307e-07,
+      "loss": 0.8591,
+      "step": 11176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.489727764262927e-07,
+      "loss": 0.8789,
+      "step": 11177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.485013953946246e-07,
+      "loss": 0.9066,
+      "step": 11178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4803075574039372e-07,
+      "loss": 0.8285,
+      "step": 11179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4756085749901728e-07,
+      "loss": 0.8787,
+      "step": 11180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4709170070585588e-07,
+      "loss": 0.9175,
+      "step": 11181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4662328539621907e-07,
+      "loss": 0.8915,
+      "step": 11182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.461556116053553e-07,
+      "loss": 0.837,
+      "step": 11183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4568867936846087e-07,
+      "loss": 0.9083,
+      "step": 11184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4522248872067434e-07,
+      "loss": 0.8695,
+      "step": 11185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4475703969707766e-07,
+      "loss": 0.8748,
+      "step": 11186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4429233233270167e-07,
+      "loss": 0.8746,
+      "step": 11187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.438283666625151e-07,
+      "loss": 0.9384,
+      "step": 11188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4336514272143554e-07,
+      "loss": 0.8931,
+      "step": 11189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4290266054432177e-07,
+      "loss": 0.9029,
+      "step": 11190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4244092016597933e-07,
+      "loss": 0.8553,
+      "step": 11191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.419799216211548e-07,
+      "loss": 0.86,
+      "step": 11192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4151966494454273e-07,
+      "loss": 0.8582,
+      "step": 11193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4106015017077757e-07,
+      "loss": 0.876,
+      "step": 11194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.406013773344428e-07,
+      "loss": 0.8925,
+      "step": 11195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40143346470063e-07,
+      "loss": 0.8888,
+      "step": 11196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3968605761210618e-07,
+      "loss": 0.8827,
+      "step": 11197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3922951079498593e-07,
+      "loss": 0.8926,
+      "step": 11198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3877370605306028e-07,
+      "loss": 0.8852,
+      "step": 11199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3831864342063072e-07,
+      "loss": 0.8791,
+      "step": 11200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.378643229319454e-07,
+      "loss": 0.8827,
+      "step": 11201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.374107446211903e-07,
+      "loss": 0.8848,
+      "step": 11202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3695790852250147e-07,
+      "loss": 0.8834,
+      "step": 11203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3650581466995828e-07,
+      "loss": 0.901,
+      "step": 11204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.360544630975813e-07,
+      "loss": 0.9064,
+      "step": 11205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3560385383933895e-07,
+      "loss": 0.8931,
+      "step": 11206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.351539869291396e-07,
+      "loss": 0.9132,
+      "step": 11207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3470486240084068e-07,
+      "loss": 0.9187,
+      "step": 11208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3425648028824068e-07,
+      "loss": 0.8945,
+      "step": 11209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3380884062508147e-07,
+      "loss": 0.8882,
+      "step": 11210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.333619434450506e-07,
+      "loss": 0.868,
+      "step": 11211
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3291578878178e-07,
+      "loss": 0.8408,
+      "step": 11212
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.324703766688451e-07,
+      "loss": 0.8765,
+      "step": 11213
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3202570713976458e-07,
+      "loss": 0.8978,
+      "step": 11214
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3158178022800506e-07,
+      "loss": 0.9021,
+      "step": 11215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3113859596697086e-07,
+      "loss": 0.925,
+      "step": 11216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3069615439001536e-07,
+      "loss": 0.8919,
+      "step": 11217
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3025445553043525e-07,
+      "loss": 0.8726,
+      "step": 11218
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2981349942146947e-07,
+      "loss": 0.8888,
+      "step": 11219
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2937328609630374e-07,
+      "loss": 0.8643,
+      "step": 11220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2893381558806483e-07,
+      "loss": 0.9097,
+      "step": 11221
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2849508792982746e-07,
+      "loss": 0.8162,
+      "step": 11222
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.280571031546052e-07,
+      "loss": 0.9063,
+      "step": 11223
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2761986129536052e-07,
+      "loss": 0.8999,
+      "step": 11224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.271833623849994e-07,
+      "loss": 0.9169,
+      "step": 11225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2674760645636774e-07,
+      "loss": 0.8818,
+      "step": 11226
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2631259354226155e-07,
+      "loss": 0.8963,
+      "step": 11227
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2587832367541464e-07,
+      "loss": 0.8919,
+      "step": 11228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2544479688851085e-07,
+      "loss": 0.8703,
+      "step": 11229
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.250120132141741e-07,
+      "loss": 0.8551,
+      "step": 11230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2457997268497392e-07,
+      "loss": 0.8732,
+      "step": 11231
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2414867533342422e-07,
+      "loss": 0.8943,
+      "step": 11232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2371812119198135e-07,
+      "loss": 0.9091,
+      "step": 11233
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.23288310293046e-07,
+      "loss": 0.9423,
+      "step": 11234
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2285924266896675e-07,
+      "loss": 0.9256,
+      "step": 11235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2243091835203114e-07,
+      "loss": 0.8856,
+      "step": 11236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2200333737447222e-07,
+      "loss": 0.8704,
+      "step": 11237
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2157649976846986e-07,
+      "loss": 0.8929,
+      "step": 11238
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.211504055661439e-07,
+      "loss": 0.9297,
+      "step": 11239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2072505479956088e-07,
+      "loss": 0.8527,
+      "step": 11240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2030044750073077e-07,
+      "loss": 0.8842,
+      "step": 11241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1987658370160804e-07,
+      "loss": 0.8892,
+      "step": 11242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1945346343408937e-07,
+      "loss": 0.9083,
+      "step": 11243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1903108673001817e-07,
+      "loss": 2.471,
+      "step": 11244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1860945362117904e-07,
+      "loss": 0.879,
+      "step": 11245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1818856413930436e-07,
+      "loss": 0.9318,
+      "step": 11246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1776841831606544e-07,
+      "loss": 0.8516,
+      "step": 11247
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1734901618308259e-07,
+      "loss": 0.8935,
+      "step": 11248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1693035777191719e-07,
+      "loss": 0.8569,
+      "step": 11249
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1651244311407628e-07,
+      "loss": 0.8462,
+      "step": 11250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1609527224100803e-07,
+      "loss": 0.8772,
+      "step": 11251
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1567884518411065e-07,
+      "loss": 0.8606,
+      "step": 11252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1526316197471798e-07,
+      "loss": 0.8632,
+      "step": 11253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.148482226441161e-07,
+      "loss": 0.8665,
+      "step": 11254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1443402722352892e-07,
+      "loss": 0.9004,
+      "step": 11255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1402057574412816e-07,
+      "loss": 0.8932,
+      "step": 11256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1360786823702785e-07,
+      "loss": 0.8864,
+      "step": 11257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1319590473328534e-07,
+      "loss": 0.8615,
+      "step": 11258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1278468526390584e-07,
+      "loss": 0.8875,
+      "step": 11259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.123742098598335e-07,
+      "loss": 0.9281,
+      "step": 11260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1196447855195802e-07,
+      "loss": 0.8829,
+      "step": 11261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1155549137111698e-07,
+      "loss": 0.8827,
+      "step": 11262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1114724834808576e-07,
+      "loss": 0.8514,
+      "step": 11263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1073974951358867e-07,
+      "loss": 0.8691,
+      "step": 11264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1033299489829119e-07,
+      "loss": 0.8927,
+      "step": 11265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0992698453280437e-07,
+      "loss": 0.8559,
+      "step": 11266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0952171844768267e-07,
+      "loss": 0.916,
+      "step": 11267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.091171966734239e-07,
+      "loss": 0.8763,
+      "step": 11268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0871341924046929e-07,
+      "loss": 0.8873,
+      "step": 11269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0831038617920898e-07,
+      "loss": 0.8887,
+      "step": 11270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0790809751996978e-07,
+      "loss": 0.9094,
+      "step": 11271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0750655329302862e-07,
+      "loss": 0.8834,
+      "step": 11272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0710575352860131e-07,
+      "loss": 0.904,
+      "step": 11273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0670569825685262e-07,
+      "loss": 2.572,
+      "step": 11274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0630638750788625e-07,
+      "loss": 0.9033,
+      "step": 11275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0590782131175481e-07,
+      "loss": 0.8904,
+      "step": 11276
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.05509999698451e-07,
+      "loss": 0.9123,
+      "step": 11277
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0511292269791418e-07,
+      "loss": 0.8379,
+      "step": 11278
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.047165903400249e-07,
+      "loss": 0.829,
+      "step": 11279
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0432100265461043e-07,
+      "loss": 0.9491,
+      "step": 11280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0392615967144137e-07,
+      "loss": 0.8849,
+      "step": 11281
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0353206142023065e-07,
+      "loss": 0.8702,
+      "step": 11282
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0313870793063563e-07,
+      "loss": 0.7962,
+      "step": 11283
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0274609923226042e-07,
+      "loss": 0.9025,
+      "step": 11284
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0235423535464806e-07,
+      "loss": 0.9281,
+      "step": 11285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0196311632729272e-07,
+      "loss": 0.9038,
+      "step": 11286
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.015727421796231e-07,
+      "loss": 0.8346,
+      "step": 11287
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0118311294102013e-07,
+      "loss": 0.9221,
+      "step": 11288
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0079422864080479e-07,
+      "loss": 0.876,
+      "step": 11289
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0040608930824147e-07,
+      "loss": 0.8361,
+      "step": 11290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0001869497254235e-07,
+      "loss": 0.9462,
+      "step": 11291
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.963204566285856e-08,
+      "loss": 0.9223,
+      "step": 11292
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.924614140828792e-08,
+      "loss": 0.8748,
+      "step": 11293
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.886098223787277e-08,
+      "loss": 0.8487,
+      "step": 11294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.847656818059769e-08,
+      "loss": 0.8484,
+      "step": 11295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.809289926539179e-08,
+      "loss": 0.8799,
+      "step": 11296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.770997552112749e-08,
+      "loss": 0.8748,
+      "step": 11297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.732779697662398e-08,
+      "loss": 0.8869,
+      "step": 11298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.694636366064047e-08,
+      "loss": 0.8714,
+      "step": 11299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.656567560188179e-08,
+      "loss": 0.8316,
+      "step": 11300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.618573282899724e-08,
+      "loss": 0.8756,
+      "step": 11301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.58065353705806e-08,
+      "loss": 0.8946,
+      "step": 11302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.542808325516573e-08,
+      "loss": 0.8436,
+      "step": 11303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.505037651123539e-08,
+      "loss": 0.9195,
+      "step": 11304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.467341516721352e-08,
+      "loss": 0.9047,
+      "step": 11305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.429719925146851e-08,
+      "loss": 0.8944,
+      "step": 11306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.39217287923122e-08,
+      "loss": 2.3546,
+      "step": 11307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.354700381800086e-08,
+      "loss": 0.8486,
+      "step": 11308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.317302435673525e-08,
+      "loss": 0.8878,
+      "step": 11309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.279979043665732e-08,
+      "loss": 0.8893,
+      "step": 11310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.242730208585793e-08,
+      "loss": 0.851,
+      "step": 11311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.205555933236688e-08,
+      "loss": 0.915,
+      "step": 11312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.168456220415956e-08,
+      "loss": 0.8939,
+      "step": 11313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.131431072915586e-08,
+      "loss": 0.8853,
+      "step": 11314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.094480493521907e-08,
+      "loss": 0.8929,
+      "step": 11315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.057604485015691e-08,
+      "loss": 0.8189,
+      "step": 11316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.020803050172055e-08,
+      "loss": 0.8364,
+      "step": 11317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.984076191760449e-08,
+      "loss": 0.8677,
+      "step": 11318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.947423912544773e-08,
+      "loss": 0.8617,
+      "step": 11319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.910846215283376e-08,
+      "loss": 0.8673,
+      "step": 11320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.874343102728834e-08,
+      "loss": 0.8595,
+      "step": 11321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.837914577628171e-08,
+      "loss": 0.9091,
+      "step": 11322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.801560642722973e-08,
+      "loss": 0.8413,
+      "step": 11323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.76528130074894e-08,
+      "loss": 0.8697,
+      "step": 11324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.729076554436333e-08,
+      "loss": 0.8623,
+      "step": 11325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.69294640650975e-08,
+      "loss": 0.9002,
+      "step": 11326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.656890859688127e-08,
+      "loss": 0.9255,
+      "step": 11327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.620909916684961e-08,
+      "loss": 0.8926,
+      "step": 11328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.585003580207862e-08,
+      "loss": 2.4401,
+      "step": 11329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.549171852959004e-08,
+      "loss": 0.917,
+      "step": 11330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.513414737635006e-08,
+      "loss": 0.8861,
+      "step": 11331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.477732236926717e-08,
+      "loss": 0.8867,
+      "step": 11332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.442124353519432e-08,
+      "loss": 0.8245,
+      "step": 11333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.406591090092898e-08,
+      "loss": 0.8956,
+      "step": 11334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.371132449320973e-08,
+      "loss": 0.8775,
+      "step": 11335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.335748433872415e-08,
+      "loss": 0.8929,
+      "step": 11336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.300439046409759e-08,
+      "loss": 0.893,
+      "step": 11337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.265204289590434e-08,
+      "loss": 0.8877,
+      "step": 11338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.230044166065987e-08,
+      "loss": 0.9336,
+      "step": 11339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.1949586784823e-08,
+      "loss": 0.8859,
+      "step": 11340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.159947829479931e-08,
+      "loss": 0.8879,
+      "step": 11341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.125011621693546e-08,
+      "loss": 0.8938,
+      "step": 11342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.090150057752155e-08,
+      "loss": 0.8934,
+      "step": 11343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.055363140279437e-08,
+      "loss": 0.8431,
+      "step": 11344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.020650871893298e-08,
+      "loss": 0.8852,
+      "step": 11345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.98601325520576e-08,
+      "loss": 0.9432,
+      "step": 11346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951450292823959e-08,
+      "loss": 0.8795,
+      "step": 11347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.916961987348481e-08,
+      "loss": 0.9106,
+      "step": 11348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.882548341375029e-08,
+      "loss": 0.8832,
+      "step": 11349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.848209357493308e-08,
+      "loss": 0.9284,
+      "step": 11350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.813945038287473e-08,
+      "loss": 0.891,
+      "step": 11351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.77975538633624e-08,
+      "loss": 0.8653,
+      "step": 11352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.745640404212329e-08,
+      "loss": 0.9009,
+      "step": 11353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.711600094483352e-08,
+      "loss": 0.867,
+      "step": 11354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.677634459710815e-08,
+      "loss": 0.886,
+      "step": 11355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.643743502450895e-08,
+      "loss": 0.8781,
+      "step": 11356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.60992722525411e-08,
+      "loss": 0.8887,
+      "step": 11357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.576185630665201e-08,
+      "loss": 0.8602,
+      "step": 11358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.542518721223469e-08,
+      "loss": 0.8704,
+      "step": 11359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.508926499462444e-08,
+      "loss": 0.8995,
+      "step": 11360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.475408967910214e-08,
+      "loss": 0.9232,
+      "step": 11361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.441966129089207e-08,
+      "loss": 0.8948,
+      "step": 11362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.408597985516075e-08,
+      "loss": 0.8703,
+      "step": 11363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.375304539701922e-08,
+      "loss": 0.934,
+      "step": 11364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.34208579415241e-08,
+      "loss": 0.8817,
+      "step": 11365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.308941751367093e-08,
+      "loss": 0.8764,
+      "step": 11366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.275872413840534e-08,
+      "loss": 0.925,
+      "step": 11367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.242877784061408e-08,
+      "loss": 0.849,
+      "step": 11368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.209957864512618e-08,
+      "loss": 0.8481,
+      "step": 11369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.177112657671515e-08,
+      "loss": 0.8669,
+      "step": 11370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.144342166009899e-08,
+      "loss": 0.8713,
+      "step": 11371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.111646391994021e-08,
+      "loss": 0.8893,
+      "step": 11372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.079025338084355e-08,
+      "loss": 0.9007,
+      "step": 11373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.046479006735718e-08,
+      "loss": 0.8809,
+      "step": 11374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.014007400397482e-08,
+      "loss": 0.8954,
+      "step": 11375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.98161052151336e-08,
+      "loss": 0.8944,
+      "step": 11376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.94928837252129e-08,
+      "loss": 0.8577,
+      "step": 11377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.917040955853771e-08,
+      "loss": 0.8729,
+      "step": 11378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.884868273937528e-08,
+      "loss": 0.9017,
+      "step": 11379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.852770329193737e-08,
+      "loss": 0.8957,
+      "step": 11380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.820747124037908e-08,
+      "loss": 0.8767,
+      "step": 11381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.788798660880003e-08,
+      "loss": 0.9058,
+      "step": 11382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.756924942124321e-08,
+      "loss": 0.899,
+      "step": 11383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.7251259701695e-08,
+      "loss": 0.932,
+      "step": 11384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.693401747408513e-08,
+      "loss": 0.915,
+      "step": 11385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.661752276229005e-08,
+      "loss": 0.9159,
+      "step": 11386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.630177559012518e-08,
+      "loss": 0.9225,
+      "step": 11387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.598677598135261e-08,
+      "loss": 0.9157,
+      "step": 11388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.567252395968004e-08,
+      "loss": 0.9209,
+      "step": 11389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.535901954875301e-08,
+      "loss": 0.8591,
+      "step": 11390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.504626277216708e-08,
+      "loss": 0.9195,
+      "step": 11391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.473425365345787e-08,
+      "loss": 0.871,
+      "step": 11392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.442299221610548e-08,
+      "loss": 0.8939,
+      "step": 11393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.411247848353453e-08,
+      "loss": 0.8619,
+      "step": 11394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.380271247911184e-08,
+      "loss": 0.9294,
+      "step": 11395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.349369422615103e-08,
+      "loss": 0.8963,
+      "step": 11396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.31854237479057e-08,
+      "loss": 2.4952,
+      "step": 11397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.9002,
+      "step": 11398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.257112620830175e-08,
+      "loss": 0.8451,
+      "step": 11399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.22650991931717e-08,
+      "loss": 0.8373,
+      "step": 11400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.195982004521539e-08,
+      "loss": 0.9178,
+      "step": 11401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.16552887874089e-08,
+      "loss": 0.8497,
+      "step": 11402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.135150544266721e-08,
+      "loss": 0.8505,
+      "step": 11403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.104847003385206e-08,
+      "loss": 0.8969,
+      "step": 11404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.074618258376852e-08,
+      "loss": 0.8683,
+      "step": 11405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.044464311516618e-08,
+      "loss": 0.8646,
+      "step": 11406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.014385165073689e-08,
+      "loss": 0.884,
+      "step": 11407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.984380821311698e-08,
+      "loss": 0.8923,
+      "step": 11408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9544512824885046e-08,
+      "loss": 0.8958,
+      "step": 11409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.924596550856643e-08,
+      "loss": 0.8871,
+      "step": 11410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8948166286627584e-08,
+      "loss": 0.8885,
+      "step": 11411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.865111518148059e-08,
+      "loss": 0.8696,
+      "step": 11412
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.835481221547867e-08,
+      "loss": 0.8626,
+      "step": 11413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8059257410919555e-08,
+      "loss": 0.8843,
+      "step": 11414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7764450790046554e-08,
+      "loss": 0.8823,
+      "step": 11415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.747039237504637e-08,
+      "loss": 0.8797,
+      "step": 11416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.717708218804685e-08,
+      "loss": 0.9006,
+      "step": 11417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6884520251121455e-08,
+      "loss": 0.9246,
+      "step": 11418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.659270658628702e-08,
+      "loss": 0.8383,
+      "step": 11419
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6301641215504854e-08,
+      "loss": 0.8665,
+      "step": 11420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.601132416067856e-08,
+      "loss": 0.8738,
+      "step": 11421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5721755443655104e-08,
+      "loss": 2.4324,
+      "step": 11422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.543293508622816e-08,
+      "loss": 2.5423,
+      "step": 11423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5144863110130345e-08,
+      "loss": 0.8818,
+      "step": 11424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4857539537043204e-08,
+      "loss": 0.8574,
+      "step": 11425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.457096438858833e-08,
+      "loss": 0.8625,
+      "step": 11426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.428513768633181e-08,
+      "loss": 0.8732,
+      "step": 11427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4000059451783104e-08,
+      "loss": 0.8962,
+      "step": 11428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.371572970639727e-08,
+      "loss": 0.8624,
+      "step": 11429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3432148471570525e-08,
+      "loss": 0.9207,
+      "step": 11430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.314931576864468e-08,
+      "loss": 0.9063,
+      "step": 11431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2867231618902725e-08,
+      "loss": 0.8929,
+      "step": 11432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.258589604357544e-08,
+      "loss": 0.8844,
+      "step": 11433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2305309063832575e-08,
+      "loss": 0.866,
+      "step": 11434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.202547070079167e-08,
+      "loss": 0.8542,
+      "step": 11435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1746380975510323e-08,
+      "loss": 0.9182,
+      "step": 11436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1468039908991744e-08,
+      "loss": 0.9165,
+      "step": 11437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.119044752218472e-08,
+      "loss": 0.8865,
+      "step": 11438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.091360383597699e-08,
+      "loss": 0.8711,
+      "step": 11439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.063750887120411e-08,
+      "loss": 0.8797,
+      "step": 11440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.036216264864169e-08,
+      "loss": 0.8861,
+      "step": 11441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.008756518901314e-08,
+      "loss": 0.8709,
+      "step": 11442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.981371651298306e-08,
+      "loss": 0.9093,
+      "step": 11443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.954061664115939e-08,
+      "loss": 0.9171,
+      "step": 11444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9268265594094586e-08,
+      "loss": 0.8906,
+      "step": 11445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.899666339228448e-08,
+      "loss": 0.8721,
+      "step": 11446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.872581005616828e-08,
+      "loss": 0.9239,
+      "step": 11447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8455705606128554e-08,
+      "loss": 0.8518,
+      "step": 11448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8186350062493506e-08,
+      "loss": 0.8933,
+      "step": 11449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.791774344553357e-08,
+      "loss": 0.8807,
+      "step": 11450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.764988577546148e-08,
+      "loss": 0.9338,
+      "step": 11451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7382777072436656e-08,
+      "loss": 0.8963,
+      "step": 11452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7116417356558586e-08,
+      "loss": 0.882,
+      "step": 11453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.685080664787345e-08,
+      "loss": 0.8177,
+      "step": 11454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65859449663697e-08,
+      "loss": 0.9057,
+      "step": 11455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.632183233197807e-08,
+      "loss": 0.9293,
+      "step": 11456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6058468764577094e-08,
+      "loss": 0.8898,
+      "step": 11457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5795854283984255e-08,
+      "loss": 0.8798,
+      "step": 11458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.553398890996374e-08,
+      "loss": 0.8815,
+      "step": 11459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527287266222313e-08,
+      "loss": 0.9,
+      "step": 11460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.501250556041003e-08,
+      "loss": 0.8982,
+      "step": 11461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4752887624121e-08,
+      "loss": 0.9119,
+      "step": 11462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.449401887289151e-08,
+      "loss": 0.8474,
+      "step": 11463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4235899326204866e-08,
+      "loss": 0.8682,
+      "step": 11464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.397852900348554e-08,
+      "loss": 0.8791,
+      "step": 11465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3721907924100246e-08,
+      "loss": 2.5977,
+      "step": 11466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.346603610736355e-08,
+      "loss": 0.8836,
+      "step": 11467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.321091357253004e-08,
+      "loss": 0.9288,
+      "step": 11468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2956540338797705e-08,
+      "loss": 0.8909,
+      "step": 11469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.270291642531121e-08,
+      "loss": 0.8515,
+      "step": 11470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.245004185115753e-08,
+      "loss": 0.8491,
+      "step": 11471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2197916635364765e-08,
+      "loss": 0.9347,
+      "step": 11472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.194654079690885e-08,
+      "loss": 2.5279,
+      "step": 11473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.169591435470577e-08,
+      "loss": 0.9029,
+      "step": 11474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.144603732761709e-08,
+      "loss": 0.918,
+      "step": 11475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.119690973444668e-08,
+      "loss": 0.8628,
+      "step": 11476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0948531593942855e-08,
+      "loss": 0.8763,
+      "step": 11477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0700902924798443e-08,
+      "loss": 0.8767,
+      "step": 11478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.045402374564744e-08,
+      "loss": 0.8859,
+      "step": 11479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.020789407506942e-08,
+      "loss": 0.8777,
+      "step": 11480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9962513931588453e-08,
+      "loss": 0.856,
+      "step": 11481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.971788333366755e-08,
+      "loss": 0.8574,
+      "step": 11482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.947400229971865e-08,
+      "loss": 0.8443,
+      "step": 11483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.923087084809485e-08,
+      "loss": 0.8506,
+      "step": 11484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8988488997092624e-08,
+      "loss": 0.9225,
+      "step": 11485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8746856764951825e-08,
+      "loss": 0.8541,
+      "step": 11486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.850597416985902e-08,
+      "loss": 0.9247,
+      "step": 11487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8265841229938594e-08,
+      "loss": 2.5414,
+      "step": 11488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.802645796326498e-08,
+      "loss": 0.8163,
+      "step": 11489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.778782438785045e-08,
+      "loss": 0.8985,
+      "step": 11490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.754994052165506e-08,
+      "loss": 0.9267,
+      "step": 11491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7312806382580055e-08,
+      "loss": 0.8719,
+      "step": 11492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7076421988472275e-08,
+      "loss": 0.8931,
+      "step": 11493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.68407873571186e-08,
+      "loss": 0.8792,
+      "step": 11494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.660590250625373e-08,
+      "loss": 0.8698,
+      "step": 11495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.637176745355353e-08,
+      "loss": 0.9067,
+      "step": 11496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.613838221663724e-08,
+      "loss": 0.8447,
+      "step": 11497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5905746813068576e-08,
+      "loss": 0.8681,
+      "step": 11498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.567386126035577e-08,
+      "loss": 0.891,
+      "step": 11499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.544272557594819e-08,
+      "loss": 0.8632,
+      "step": 11500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.52123397772397e-08,
+      "loss": 0.9105,
+      "step": 11501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.498270388156977e-08,
+      "loss": 0.9063,
+      "step": 11502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4753817906217905e-08,
+      "loss": 0.8804,
+      "step": 11503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.452568186841032e-08,
+      "loss": 0.8709,
+      "step": 11504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4298295785314406e-08,
+      "loss": 0.8845,
+      "step": 11505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.407165967404202e-08,
+      "loss": 0.9019,
+      "step": 11506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384577355165064e-08,
+      "loss": 0.8768,
+      "step": 11507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.362063743513777e-08,
+      "loss": 0.8426,
+      "step": 11508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.339625134144542e-08,
+      "loss": 2.3767,
+      "step": 11509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.31726152874623e-08,
+      "loss": 0.8748,
+      "step": 11510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.294972929001605e-08,
+      "loss": 0.8653,
+      "step": 11511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.272759336587994e-08,
+      "loss": 0.8957,
+      "step": 11512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2506207531772806e-08,
+      "loss": 0.8241,
+      "step": 11513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2285571804352434e-08,
+      "loss": 0.8261,
+      "step": 11514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.206568620022554e-08,
+      "loss": 2.3553,
+      "step": 11515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1846550735938895e-08,
+      "loss": 0.8649,
+      "step": 11516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.162816542798264e-08,
+      "loss": 0.9005,
+      "step": 11517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1410530292791396e-08,
+      "loss": 2.4515,
+      "step": 11518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1193645346744296e-08,
+      "loss": 0.8568,
+      "step": 11519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.097751060616272e-08,
+      "loss": 0.8749,
+      "step": 11520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.076212608731255e-08,
+      "loss": 0.8854,
+      "step": 11521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.054749180640082e-08,
+      "loss": 0.8856,
+      "step": 11522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0333607779582386e-08,
+      "loss": 0.9121,
+      "step": 11523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.012047402295104e-08,
+      "loss": 0.8673,
+      "step": 11524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9908090552547287e-08,
+      "loss": 0.8929,
+      "step": 11525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9696457384353896e-08,
+      "loss": 0.8484,
+      "step": 11526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.948557453429701e-08,
+      "loss": 0.8795,
+      "step": 11527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.927544201824728e-08,
+      "loss": 0.8795,
+      "step": 11528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.906605985201871e-08,
+      "loss": 0.8847,
+      "step": 11529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8857428051366486e-08,
+      "loss": 0.9038,
+      "step": 11530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8649546631993595e-08,
+      "loss": 0.8101,
+      "step": 11531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8442415609541973e-08,
+      "loss": 2.3623,
+      "step": 11532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.823603499960137e-08,
+      "loss": 0.8949,
+      "step": 11533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8030404817701583e-08,
+      "loss": 0.8835,
+      "step": 11534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7825525079318016e-08,
+      "loss": 0.9337,
+      "step": 11535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7621395799869445e-08,
+      "loss": 0.9044,
+      "step": 11536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7418016994716913e-08,
+      "loss": 0.8645,
+      "step": 11537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7215388679165956e-08,
+      "loss": 0.915,
+      "step": 11538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7013510868464375e-08,
+      "loss": 0.9251,
+      "step": 11539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.681238357780669e-08,
+      "loss": 0.9098,
+      "step": 11540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6612006822327454e-08,
+      "loss": 0.9194,
+      "step": 11541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6412380617105715e-08,
+      "loss": 0.9002,
+      "step": 11542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6213504977166126e-08,
+      "loss": 0.8708,
+      "step": 11543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6015379917472273e-08,
+      "loss": 0.8873,
+      "step": 11544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5818005452937777e-08,
+      "loss": 0.8293,
+      "step": 11545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5621381598412985e-08,
+      "loss": 0.9217,
+      "step": 11546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5425508368696062e-08,
+      "loss": 0.8462,
+      "step": 11547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5230385778527432e-08,
+      "loss": 0.8945,
+      "step": 11548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.503601384259202e-08,
+      "loss": 0.933,
+      "step": 11549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4842392575515905e-08,
+      "loss": 0.8487,
+      "step": 11550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4649521991870762e-08,
+      "loss": 0.8731,
+      "step": 11551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4457402106171647e-08,
+      "loss": 0.9369,
+      "step": 11552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.426603293287477e-08,
+      "loss": 0.8971,
+      "step": 11553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4075414486384174e-08,
+      "loss": 0.8947,
+      "step": 11554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3885546781042824e-08,
+      "loss": 0.8868,
+      "step": 11555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.36964298311404e-08,
+      "loss": 0.9074,
+      "step": 11556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3508063650908852e-08,
+      "loss": 0.8637,
+      "step": 11557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3320448254523508e-08,
+      "loss": 0.9362,
+      "step": 11558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3133583656103075e-08,
+      "loss": 0.9182,
+      "step": 11559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2947469869710747e-08,
+      "loss": 0.8927,
+      "step": 11560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.276210690935199e-08,
+      "loss": 0.9083,
+      "step": 11561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.257749478897786e-08,
+      "loss": 0.886,
+      "step": 11562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2393633522479473e-08,
+      "loss": 0.8695,
+      "step": 11563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2210523123693538e-08,
+      "loss": 0.9098,
+      "step": 11564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2028163606401253e-08,
+      "loss": 0.868,
+      "step": 11565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1846554984324974e-08,
+      "loss": 0.8751,
+      "step": 11566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.166569727113266e-08,
+      "loss": 0.9062,
+      "step": 11567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1485590480434526e-08,
+      "loss": 0.8964,
+      "step": 11568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1306234625784184e-08,
+      "loss": 0.938,
+      "step": 11569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1127629720679722e-08,
+      "loss": 0.9286,
+      "step": 11570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0949775778561497e-08,
+      "loss": 0.9146,
+      "step": 11571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0772672812814366e-08,
+      "loss": 0.8765,
+      "step": 11572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0596320836765436e-08,
+      "loss": 0.8615,
+      "step": 11573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0420719863687433e-08,
+      "loss": 0.9168,
+      "step": 11574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0245869906794225e-08,
+      "loss": 0.8909,
+      "step": 11575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0071770979245286e-08,
+      "loss": 2.4701,
+      "step": 11576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.989842309414125e-08,
+      "loss": 0.8475,
+      "step": 11577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9725826264528348e-08,
+      "loss": 0.9011,
+      "step": 11578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9553980503395076e-08,
+      "loss": 0.9041,
+      "step": 11579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9382885823673316e-08,
+      "loss": 0.8789,
+      "step": 11580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.921254223823943e-08,
+      "loss": 0.8853,
+      "step": 11581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9042949759913164e-08,
+      "loss": 0.9006,
+      "step": 11582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8874108401456538e-08,
+      "loss": 0.8823,
+      "step": 11583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8706018175574936e-08,
+      "loss": 0.8877,
+      "step": 11584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8538679094919354e-08,
+      "loss": 0.8954,
+      "step": 11585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.837209117208305e-08,
+      "loss": 0.8984,
+      "step": 11586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8206254419601556e-08,
+      "loss": 0.9321,
+      "step": 11587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8041168849954883e-08,
+      "loss": 0.8655,
+      "step": 11588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.787683447556754e-08,
+      "loss": 0.9003,
+      "step": 11589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7713251308807412e-08,
+      "loss": 0.9265,
+      "step": 11590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7550419361982428e-08,
+      "loss": 0.8812,
+      "step": 11591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7388338647349458e-08,
+      "loss": 0.9169,
+      "step": 11592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7227009177103183e-08,
+      "loss": 0.9012,
+      "step": 11593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7066430963386115e-08,
+      "loss": 0.8527,
+      "step": 11594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6906604018281926e-08,
+      "loss": 0.893,
+      "step": 11595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6747528353819874e-08,
+      "loss": 0.8902,
+      "step": 11596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6589203981969282e-08,
+      "loss": 0.8825,
+      "step": 11597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6431630914646167e-08,
+      "loss": 0.8766,
+      "step": 11598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6274809163707716e-08,
+      "loss": 0.8965,
+      "step": 11599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6118738740956708e-08,
+      "loss": 0.9222,
+      "step": 11600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.596341965813708e-08,
+      "loss": 0.8855,
+      "step": 11601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5808851926938374e-08,
+      "loss": 0.9083,
+      "step": 11602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5655035558992394e-08,
+      "loss": 0.8607,
+      "step": 11603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5501970565875434e-08,
+      "loss": 0.9084,
+      "step": 11604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5349656959104953e-08,
+      "loss": 0.8964,
+      "step": 11605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5198094750144e-08,
+      "loss": 0.8748,
+      "step": 11606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5047283950397896e-08,
+      "loss": 2.3771,
+      "step": 11607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.489722457121756e-08,
+      "loss": 0.8975,
+      "step": 11608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4747916623893965e-08,
+      "loss": 0.8986,
+      "step": 11609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4599360119664784e-08,
+      "loss": 0.9488,
+      "step": 11610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4451555069708856e-08,
+      "loss": 2.5299,
+      "step": 11611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4304501485148392e-08,
+      "loss": 0.885,
+      "step": 11612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4158199377052318e-08,
+      "loss": 0.886,
+      "step": 11613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4012648756428493e-08,
+      "loss": 0.9225,
+      "step": 11614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3867849634230379e-08,
+      "loss": 0.9306,
+      "step": 11615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3723802021357035e-08,
+      "loss": 0.8517,
+      "step": 11616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.358050592864646e-08,
+      "loss": 2.4233,
+      "step": 11617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3437961366884466e-08,
+      "loss": 0.8708,
+      "step": 11618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3296168346796922e-08,
+      "loss": 0.8594,
+      "step": 11619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.315512687905418e-08,
+      "loss": 0.8811,
+      "step": 11620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3014836974271083e-08,
+      "loss": 0.8831,
+      "step": 11621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2875298643005852e-08,
+      "loss": 0.867,
+      "step": 11622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2736511895757864e-08,
+      "loss": 0.8838,
+      "step": 11623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2598476742973209e-08,
+      "loss": 0.8865,
+      "step": 11624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2461193195038023e-08,
+      "loss": 0.9298,
+      "step": 11625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2324661262285154e-08,
+      "loss": 0.8683,
+      "step": 11626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2188880954988602e-08,
+      "loss": 0.8396,
+      "step": 11627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2053852283366863e-08,
+      "loss": 0.9014,
+      "step": 11628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1919575257581806e-08,
+      "loss": 0.8567,
+      "step": 11629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1786049887737572e-08,
+      "loss": 0.881,
+      "step": 11630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1653276183882788e-08,
+      "loss": 0.9326,
+      "step": 11631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1521254156010576e-08,
+      "loss": 0.8556,
+      "step": 11632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1389983814055206e-08,
+      "loss": 0.8847,
+      "step": 11633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1259465167895444e-08,
+      "loss": 0.9072,
+      "step": 11634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1129698227354546e-08,
+      "loss": 0.8905,
+      "step": 11635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.100068300219692e-08,
+      "loss": 0.9172,
+      "step": 11636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.087241950213258e-08,
+      "loss": 0.8684,
+      "step": 11637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0744907736812694e-08,
+      "loss": 0.8865,
+      "step": 11638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0618147715835137e-08,
+      "loss": 0.8336,
+      "step": 11639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0492139448737838e-08,
+      "loss": 0.8981,
+      "step": 11640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.036688294500432e-08,
+      "loss": 0.9264,
+      "step": 11641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0242378214060378e-08,
+      "loss": 0.8928,
+      "step": 11642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0118625265275184e-08,
+      "loss": 0.8614,
+      "step": 11643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.995624107962399e-09,
+      "loss": 0.8893,
+      "step": 11644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.873374751379062e-09,
+      "loss": 0.8876,
+      "step": 11645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.751877204723371e-09,
+      "loss": 0.9049,
+      "step": 11646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.631131477141343e-09,
+      "loss": 0.8622,
+      "step": 11647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.511137577717932e-09,
+      "loss": 0.9043,
+      "step": 11648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.391895515482585e-09,
+      "loss": 0.9036,
+      "step": 11649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.273405299410343e-09,
+      "loss": 0.954,
+      "step": 11650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.155666938417407e-09,
+      "loss": 0.9404,
+      "step": 11651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.038680441365577e-09,
+      "loss": 0.9075,
+      "step": 11652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.922445817056701e-09,
+      "loss": 2.5428,
+      "step": 11653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.806963074239338e-09,
+      "loss": 0.8882,
+      "step": 11654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.692232221604314e-09,
+      "loss": 0.8925,
+      "step": 11655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.57825326778472e-09,
+      "loss": 0.8999,
+      "step": 11656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.465026221359251e-09,
+      "loss": 0.9028,
+      "step": 11657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.35255109084776e-09,
+      "loss": 0.8944,
+      "step": 11658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.240827884714586e-09,
+      "loss": 0.8875,
+      "step": 11659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.129856611368558e-09,
+      "loss": 0.9369,
+      "step": 11660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.019637279160775e-09,
+      "loss": 0.9009,
+      "step": 11661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.910169896384601e-09,
+      "loss": 0.896,
+      "step": 11662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.801454471279001e-09,
+      "loss": 0.8398,
+      "step": 11663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.693491012026322e-09,
+      "loss": 0.863,
+      "step": 11664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.586279526747842e-09,
+      "loss": 0.9156,
+      "step": 11665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.479820023515994e-09,
+      "loss": 0.9003,
+      "step": 11666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3741125103399254e-09,
+      "loss": 0.8781,
+      "step": 11667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.269156995175497e-09,
+      "loss": 0.8767,
+      "step": 11668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1649534859208335e-09,
+      "loss": 0.8673,
+      "step": 11669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.061501990418551e-09,
+      "loss": 0.9099,
+      "step": 11670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.9588025164524234e-09,
+      "loss": 0.8615,
+      "step": 11671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.856855071752933e-09,
+      "loss": 2.4324,
+      "step": 11672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.75565966399061e-09,
+      "loss": 0.8902,
+      "step": 11673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.655216300782697e-09,
+      "loss": 0.8805,
+      "step": 11674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.555524989686479e-09,
+      "loss": 0.8584,
+      "step": 11675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.456585738203736e-09,
+      "loss": 0.8588,
+      "step": 11676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.358398553782952e-09,
+      "loss": 0.8885,
+      "step": 11677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.260963443810441e-09,
+      "loss": 0.8974,
+      "step": 11678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.164280415620339e-09,
+      "loss": 0.8964,
+      "step": 11679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0683494764868235e-09,
+      "loss": 0.8683,
+      "step": 11680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.973170633631897e-09,
+      "loss": 0.8783,
+      "step": 11681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.878743894216499e-09,
+      "loss": 0.8995,
+      "step": 11682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.785069265346055e-09,
+      "loss": 0.9227,
+      "step": 11683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6921467540715925e-09,
+      "loss": 0.8888,
+      "step": 11684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5999763673852954e-09,
+      "loss": 0.8741,
+      "step": 11685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.508558112223838e-09,
+      "loss": 0.9063,
+      "step": 11686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.417891995467272e-09,
+      "loss": 0.8853,
+      "step": 11687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.327978023936808e-09,
+      "loss": 0.8897,
+      "step": 11688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.238816204401476e-09,
+      "loss": 0.9119,
+      "step": 11689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.150406543569242e-09,
+      "loss": 0.9087,
+      "step": 11690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.062749048094784e-09,
+      "loss": 0.9114,
+      "step": 11691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.975843724573936e-09,
+      "loss": 0.8935,
+      "step": 11692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.889690579547024e-09,
+      "loss": 0.8962,
+      "step": 11693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.804289619497748e-09,
+      "loss": 0.897,
+      "step": 11694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.719640850852081e-09,
+      "loss": 0.8797,
+      "step": 11695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635744279981591e-09,
+      "loss": 0.8582,
+      "step": 11696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.552599913199008e-09,
+      "loss": 2.3451,
+      "step": 11697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.470207756761546e-09,
+      "loss": 0.8873,
+      "step": 11698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.388567816869804e-09,
+      "loss": 0.8614,
+      "step": 11699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.307680099667755e-09,
+      "loss": 0.8921,
+      "step": 11700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.227544611242751e-09,
+      "loss": 0.9284,
+      "step": 11701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.148161357624414e-09,
+      "loss": 0.8772,
+      "step": 11702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.069530344787964e-09,
+      "loss": 0.9039,
+      "step": 11703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.99165157865089e-09,
+      "loss": 0.8877,
+      "step": 11704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9145250650729494e-09,
+      "loss": 0.8868,
+      "step": 11705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.838150809858388e-09,
+      "loss": 0.8903,
+      "step": 11706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762528818754829e-09,
+      "loss": 0.9649,
+      "step": 11707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.687659097454388e-09,
+      "loss": 0.893,
+      "step": 11708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6135416515903354e-09,
+      "loss": 0.9004,
+      "step": 11709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.540176486740432e-09,
+      "loss": 0.8546,
+      "step": 11710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.467563608426927e-09,
+      "loss": 2.574,
+      "step": 11711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.395703022112118e-09,
+      "loss": 0.8595,
+      "step": 11712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3245947332061212e-09,
+      "loss": 0.9127,
+      "step": 11713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2542387470591018e-09,
+      "loss": 0.8613,
+      "step": 11714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1846350689657137e-09,
+      "loss": 0.9142,
+      "step": 11715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1157837041639884e-09,
+      "loss": 0.9135,
+      "step": 11716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.047684657835337e-09,
+      "loss": 0.9202,
+      "step": 11717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9803379351056596e-09,
+      "loss": 0.8975,
+      "step": 11718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9137435410420134e-09,
+      "loss": 0.8788,
+      "step": 11719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8479014806559456e-09,
+      "loss": 0.8716,
+      "step": 11720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7828117589023816e-09,
+      "loss": 0.843,
+      "step": 11721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.718474380680736e-09,
+      "loss": 0.9053,
+      "step": 11722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.65488935083158e-09,
+      "loss": 0.9088,
+      "step": 11723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.592056674141086e-09,
+      "loss": 0.8571,
+      "step": 11724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.529976355336583e-09,
+      "loss": 0.8707,
+      "step": 11725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4686483990910005e-09,
+      "loss": 0.8867,
+      "step": 11726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.408072810018425e-09,
+      "loss": 0.9262,
+      "step": 11727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.348249592678542e-09,
+      "loss": 0.8888,
+      "step": 11728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2891787515721963e-09,
+      "loss": 0.902,
+      "step": 11729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.230860291145831e-09,
+      "loss": 0.9245,
+      "step": 11730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1732942157881574e-09,
+      "loss": 0.8493,
+      "step": 11731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.116480529831266e-09,
+      "loss": 0.856,
+      "step": 11732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.060419237550626e-09,
+      "loss": 0.8808,
+      "step": 11733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0051103431650844e-09,
+      "loss": 0.912,
+      "step": 11734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.950553850835757e-09,
+      "loss": 0.8982,
+      "step": 11735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8967497646704692e-09,
+      "loss": 0.87,
+      "step": 11736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.843698088717094e-09,
+      "loss": 0.8186,
+      "step": 11737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7913988269679939e-09,
+      "loss": 0.8758,
+      "step": 11738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.739851983358909e-09,
+      "loss": 0.9022,
+      "step": 11739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.689057561770069e-09,
+      "loss": 0.863,
+      "step": 11740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6390155660228613e-09,
+      "loss": 0.839,
+      "step": 11741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5897259998831626e-09,
+      "loss": 0.9121,
+      "step": 11742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5411888670613384e-09,
+      "loss": 0.9024,
+      "step": 11743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.493404171208912e-09,
+      "loss": 0.8775,
+      "step": 11744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4463719159230062e-09,
+      "loss": 0.908,
+      "step": 11745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4000921047419013e-09,
+      "loss": 0.872,
+      "step": 11746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3545647411483676e-09,
+      "loss": 0.9006,
+      "step": 11747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.309789828569663e-09,
+      "loss": 0.9052,
+      "step": 11748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2657673703753147e-09,
+      "loss": 0.8326,
+      "step": 11749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2224973698782284e-09,
+      "loss": 0.8785,
+      "step": 11750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1799798303335775e-09,
+      "loss": 0.8985,
+      "step": 11751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.138214754941025e-09,
+      "loss": 0.9669,
+      "step": 11752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0972021468447224e-09,
+      "loss": 0.8666,
+      "step": 11753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.05694200913109e-09,
+      "loss": 0.9097,
+      "step": 11754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0174343448299262e-09,
+      "loss": 0.9404,
+      "step": 11755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.786791569121879e-10,
+      "loss": 0.9267,
+      "step": 11756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.406764482977616e-10,
+      "loss": 0.8761,
+      "step": 11757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.034262218443613e-10,
+      "loss": 0.8785,
+      "step": 11758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.669284803564104e-10,
+      "loss": 0.9013,
+      "step": 11759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.311832265794905e-10,
+      "loss": 0.8989,
+      "step": 11760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.961904632036721e-10,
+      "loss": 0.9206,
+      "step": 11761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.619501928635143e-10,
+      "loss": 0.84,
+      "step": 11762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.284624181347344e-10,
+      "loss": 0.9037,
+      "step": 11763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.957271415386491e-10,
+      "loss": 0.9057,
+      "step": 11764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.637443655366227e-10,
+      "loss": 0.868,
+      "step": 11765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.325140925378393e-10,
+      "loss": 0.8801,
+      "step": 11766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.020363248915307e-10,
+      "loss": 0.8368,
+      "step": 11767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.723110648903074e-10,
+      "loss": 0.8726,
+      "step": 11768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.433383147734895e-10,
+      "loss": 0.8766,
+      "step": 11769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.151180767193342e-10,
+      "loss": 0.8548,
+      "step": 11770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.876503528528087e-10,
+      "loss": 0.8877,
+      "step": 11771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.609351452400379e-10,
+      "loss": 0.8938,
+      "step": 11772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3497245589163574e-10,
+      "loss": 0.9458,
+      "step": 11773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.097622867627049e-10,
+      "loss": 0.8815,
+      "step": 11774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8530463974839616e-10,
+      "loss": 0.9489,
+      "step": 11775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.615995166916797e-10,
+      "loss": 0.9162,
+      "step": 11776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.386469193744635e-10,
+      "loss": 0.9224,
+      "step": 11777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.164468495242545e-10,
+      "loss": 0.8704,
+      "step": 11778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.949993088130487e-10,
+      "loss": 0.8932,
+      "step": 11779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.743042988540001e-10,
+      "loss": 0.8609,
+      "step": 11780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.543618212036414e-10,
+      "loss": 0.8758,
+      "step": 11781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.351718773652145e-10,
+      "loss": 0.8535,
+      "step": 11782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1673446878089922e-10,
+      "loss": 0.907,
+      "step": 11783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9904959683847424e-10,
+      "loss": 0.902,
+      "step": 11784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8211726286909703e-10,
+      "loss": 0.9044,
+      "step": 11785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6593746814619338e-10,
+      "loss": 0.8785,
+      "step": 11786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5051021388878816e-10,
+      "loss": 0.8687,
+      "step": 11787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3583550125706447e-10,
+      "loss": 0.87,
+      "step": 11788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2191333135569416e-10,
+      "loss": 0.8863,
+      "step": 11789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0874370523272781e-10,
+      "loss": 0.8883,
+      "step": 11790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.632662387848435e-11,
+      "loss": 0.8882,
+      "step": 11791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.466208822666133e-11,
+      "loss": 0.9117,
+      "step": 11792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.37500991565554e-11,
+      "loss": 0.8671,
+      "step": 11793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.359065748973159e-11,
+      "loss": 0.8555,
+      "step": 11794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.418376398891312e-11,
+      "loss": 0.9038,
+      "step": 11795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.55294193635325e-11,
+      "loss": 0.8783,
+      "step": 11796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.762762426418043e-11,
+      "loss": 0.8869,
+      "step": 11797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.047837928593644e-11,
+      "loss": 0.8697,
+      "step": 11798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.408168496614849e-11,
+      "loss": 0.8697,
+      "step": 11799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8437541786653356e-11,
+      "loss": 0.8835,
+      "step": 11800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3545950171556244e-11,
+      "loss": 0.8653,
+      "step": 11801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.406910490561417e-12,
+      "loss": 0.8775,
+      "step": 11802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.020423053421099e-12,
+      "loss": 0.9045,
+      "step": 11803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.386488116596809e-12,
+      "loss": 0.8749,
+      "step": 11804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5051058765980232e-12,
+      "loss": 0.9089,
+      "step": 11805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7627647664351117e-13,
+      "loss": 0.8482,
+      "step": 11806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.6527,
+      "step": 11807
+    },
+    {
+      "epoch": 1.0,
+      "step": 11807,
+      "total_flos": 2.9413763845718016e+16,
+      "train_loss": 1.063088939773892,
+      "train_runtime": 195718.6128,
+      "train_samples_per_second": 7.722,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11807,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 2.9413763845718016e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}