diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28417 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 4732,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253522e-07,
+      "loss": 1.6517,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 1.6048,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225352112676057e-07,
+      "loss": 1.5814,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 1.6784,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.042253521126762e-07,
+      "loss": 1.6074,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 1.502,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.859154929577465e-07,
+      "loss": 1.543,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 1.6406,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.267605633802817e-06,
+      "loss": 1.5514,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 1.5462,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5492957746478873e-06,
+      "loss": 1.5221,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 1.5137,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8309859154929579e-06,
+      "loss": 1.4779,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 1.4668,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1126760563380285e-06,
+      "loss": 1.3509,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 1.3255,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3943661971830984e-06,
+      "loss": 1.3138,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 1.2174,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.676056338028169e-06,
+      "loss": 1.2507,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 1.2946,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9577464788732396e-06,
+      "loss": 1.1777,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 1.1826,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2394366197183104e-06,
+      "loss": 1.2337,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 1.1673,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5211267605633804e-06,
+      "loss": 1.1608,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 1.2201,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8028169014084508e-06,
+      "loss": 1.1719,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 1.1562,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0845070422535216e-06,
+      "loss": 1.1735,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 1.0811,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3661971830985915e-06,
+      "loss": 1.0918,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 1.0892,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.647887323943662e-06,
+      "loss": 1.1172,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 1.0837,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.929577464788733e-06,
+      "loss": 1.0443,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 1.0924,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.211267605633803e-06,
+      "loss": 1.082,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 1.1842,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.492957746478874e-06,
+      "loss": 1.112,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 1.0221,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.774647887323944e-06,
+      "loss": 1.0667,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 1.0768,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.056338028169015e-06,
+      "loss": 1.0811,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 1.0387,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3380281690140855e-06,
+      "loss": 1.1217,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 1.0404,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.619718309859155e-06,
+      "loss": 1.113,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 1.0599,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.901408450704225e-06,
+      "loss": 1.0498,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 1.0553,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.183098591549297e-06,
+      "loss": 1.0977,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 1.0755,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.464788732394367e-06,
+      "loss": 1.1257,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 1.0091,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.746478873239436e-06,
+      "loss": 1.0407,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 1.0361,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.028169014084509e-06,
+      "loss": 1.1061,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 1.0251,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.309859154929578e-06,
+      "loss": 1.0114,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 1.0954,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.591549295774648e-06,
+      "loss": 1.057,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 1.0146,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.87323943661972e-06,
+      "loss": 1.029,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 0.9876,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.15492957746479e-06,
+      "loss": 1.0273,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 1.0277,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.43661971830986e-06,
+      "loss": 1.0563,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 0.986,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.71830985915493e-06,
+      "loss": 0.9906,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 1.0179,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.068,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 0.9873,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0281690140845072e-05,
+      "loss": 1.0791,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 1.0267,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.056338028169014e-05,
+      "loss": 1.0573,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 0.9342,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0845070422535213e-05,
+      "loss": 1.0254,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 1.0114,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1126760563380284e-05,
+      "loss": 1.0081,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 1.0049,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1408450704225353e-05,
+      "loss": 0.974,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 1.0078,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1690140845070424e-05,
+      "loss": 1.0697,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 1.0456,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1971830985915493e-05,
+      "loss": 0.986,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 0.9912,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2253521126760564e-05,
+      "loss": 0.9896,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 1.0293,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2535211267605636e-05,
+      "loss": 0.9941,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 1.0199,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2816901408450705e-05,
+      "loss": 0.9782,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 1.041,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3098591549295776e-05,
+      "loss": 1.0234,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 1.0075,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3380281690140845e-05,
+      "loss": 1.0879,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 1.0449,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3661971830985916e-05,
+      "loss": 1.068,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 1.0075,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3943661971830987e-05,
+      "loss": 0.9404,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 0.9814,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4225352112676058e-05,
+      "loss": 1.0065,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 1.0264,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4507042253521129e-05,
+      "loss": 1.0153,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 1.0296,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.47887323943662e-05,
+      "loss": 0.9183,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 0.9896,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5070422535211269e-05,
+      "loss": 0.957,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 0.9622,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.535211267605634e-05,
+      "loss": 0.9915,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 1.0215,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.563380281690141e-05,
+      "loss": 1.0133,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 0.9906,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.591549295774648e-05,
+      "loss": 1.002,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 0.9837,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.619718309859155e-05,
+      "loss": 0.9574,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 0.9827,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6478873239436623e-05,
+      "loss": 0.9762,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 1.0459,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.676056338028169e-05,
+      "loss": 1.0225,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 1.0098,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.704225352112676e-05,
+      "loss": 1.0023,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 0.9544,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.732394366197183e-05,
+      "loss": 0.9886,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 1.015,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7605633802816902e-05,
+      "loss": 1.0202,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 1.0426,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7887323943661975e-05,
+      "loss": 0.9668,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 1.0003,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8169014084507044e-05,
+      "loss": 1.0417,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 1.0013,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8450704225352113e-05,
+      "loss": 0.9655,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 0.9854,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8732394366197186e-05,
+      "loss": 1.0026,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 0.9528,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9014084507042255e-05,
+      "loss": 0.9811,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 1.0938,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9295774647887327e-05,
+      "loss": 0.9818,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 0.9421,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9577464788732396e-05,
+      "loss": 0.9655,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 1.0088,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9859154929577465e-05,
+      "loss": 1.0381,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9785,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999765769006e-05,
+      "loss": 0.943,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999990630761334e-05,
+      "loss": 0.9486,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997891921711e-05,
+      "loss": 0.9674,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999962523062883e-05,
+      "loss": 0.9775,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999941442306328e-05,
+      "loss": 0.9814,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991567695732e-05,
+      "loss": 0.985,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999885227027934e-05,
+      "loss": 1.0465,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999850092532432e-05,
+      "loss": 0.9974,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981027348727e-05,
+      "loss": 0.9411,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999765769911108e-05,
+      "loss": 1.0,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999716581824788e-05,
+      "loss": 0.9766,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999662709251356e-05,
+      "loss": 0.9548,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999960415221605e-05,
+      "loss": 0.9769,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99995409107463e-05,
+      "loss": 1.0036,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999472984871734e-05,
+      "loss": 1.027,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999940037462417e-05,
+      "loss": 0.9684,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999323080037623e-05,
+      "loss": 0.9753,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999924110114831e-05,
+      "loss": 0.9525,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999154437994622e-05,
+      "loss": 0.9958,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999906309061717e-05,
+      "loss": 0.9896,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998967059058738e-05,
+      "loss": 0.9954,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998866343364317e-05,
+      "loss": 0.9766,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999876094358109e-05,
+      "loss": 1.027,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999865085975843e-05,
+      "loss": 0.9521,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998536091947907e-05,
+      "loss": 0.9557,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998416640203285e-05,
+      "loss": 0.9987,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998292504580528e-05,
+      "loss": 0.9948,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998163685137785e-05,
+      "loss": 0.9183,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99980301819354e-05,
+      "loss": 0.9521,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997891995035914e-05,
+      "loss": 0.9808,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999774912450407e-05,
+      "loss": 1.0046,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997601570406792e-05,
+      "loss": 0.9548,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99974493328132e-05,
+      "loss": 0.9346,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999729241179462e-05,
+      "loss": 0.9655,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997130807424556e-05,
+      "loss": 0.918,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999696451977872e-05,
+      "loss": 0.9352,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996793548935005e-05,
+      "loss": 0.93,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996617894973505e-05,
+      "loss": 0.932,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996437557976515e-05,
+      "loss": 0.9639,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999625253802851e-05,
+      "loss": 0.9108,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996062835216163e-05,
+      "loss": 1.0306,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995868449628346e-05,
+      "loss": 0.9118,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999566938135612e-05,
+      "loss": 0.9899,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999546563049274e-05,
+      "loss": 0.9788,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999525719713366e-05,
+      "loss": 0.93,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995044081376516e-05,
+      "loss": 0.9368,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994826283321155e-05,
+      "loss": 0.9945,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994603803069594e-05,
+      "loss": 0.9297,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994376640726066e-05,
+      "loss": 0.9626,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994144796396985e-05,
+      "loss": 0.9593,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 0.9235,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993667062218796e-05,
+      "loss": 0.9775,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999342117259349e-05,
+      "loss": 0.9596,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993170601430233e-05,
+      "loss": 0.9567,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992915348846403e-05,
+      "loss": 0.9609,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992655414961583e-05,
+      "loss": 0.9404,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992390799897533e-05,
+      "loss": 0.9691,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992121503778228e-05,
+      "loss": 0.9684,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991847526729816e-05,
+      "loss": 0.945,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999156886888064e-05,
+      "loss": 0.9349,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999128553036125e-05,
+      "loss": 0.8981,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999099751130437e-05,
+      "loss": 0.9066,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990704811844934e-05,
+      "loss": 0.9355,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990407432120055e-05,
+      "loss": 0.9535,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999010537226905e-05,
+      "loss": 0.9847,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989798632433417e-05,
+      "loss": 0.9834,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998948721275685e-05,
+      "loss": 0.9902,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998917111338525e-05,
+      "loss": 0.9548,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998885033446668e-05,
+      "loss": 0.9893,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988524876151425e-05,
+      "loss": 0.9639,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988194738591947e-05,
+      "loss": 0.9635,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99878599219429e-05,
+      "loss": 0.9307,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987520426361138e-05,
+      "loss": 0.9316,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987176252005697e-05,
+      "loss": 0.9801,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998682739903781e-05,
+      "loss": 0.9053,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986473867620905e-05,
+      "loss": 0.9176,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986115657920596e-05,
+      "loss": 0.8994,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985752770104693e-05,
+      "loss": 0.9144,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985385204343187e-05,
+      "loss": 0.8682,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985012960808275e-05,
+      "loss": 0.9202,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984636039674342e-05,
+      "loss": 0.8887,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984254441117952e-05,
+      "loss": 0.96,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983868165317883e-05,
+      "loss": 1.0052,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983477212455074e-05,
+      "loss": 0.89,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983081582712684e-05,
+      "loss": 0.9046,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982681276276044e-05,
+      "loss": 0.9382,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982276293332688e-05,
+      "loss": 0.9587,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998186663407233e-05,
+      "loss": 0.9538,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998145229868688e-05,
+      "loss": 0.8841,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981033287370443e-05,
+      "loss": 0.9355,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980609600319305e-05,
+      "loss": 0.9622,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998018123773195e-05,
+      "loss": 0.9587,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979748199809047e-05,
+      "loss": 0.9447,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997931048675346e-05,
+      "loss": 0.9183,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978868098770244e-05,
+      "loss": 0.9554,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997842103606663e-05,
+      "loss": 0.9922,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977969298852065e-05,
+      "loss": 0.9079,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977512887338162e-05,
+      "loss": 0.9554,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977051801738733e-05,
+      "loss": 0.9294,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976586042269776e-05,
+      "loss": 0.9176,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976115609149488e-05,
+      "loss": 0.9421,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 0.8984,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975160722838617e-05,
+      "loss": 0.9551,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974676270095362e-05,
+      "loss": 0.9466,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974187144595433e-05,
+      "loss": 0.8903,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997369334656796e-05,
+      "loss": 0.9056,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997319487624427e-05,
+      "loss": 0.8877,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997269173385788e-05,
+      "loss": 1.0544,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972183919644493e-05,
+      "loss": 0.9297,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971671433842e-05,
+      "loss": 0.9046,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997115427669048e-05,
+      "loss": 0.9072,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970632448432198e-05,
+      "loss": 0.9391,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970105949311614e-05,
+      "loss": 0.9349,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969574779575377e-05,
+      "loss": 0.944,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969038939472315e-05,
+      "loss": 0.9746,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968498429253453e-05,
+      "loss": 0.9215,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996795324917199e-05,
+      "loss": 0.9678,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967403399483336e-05,
+      "loss": 0.8779,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996684888044506e-05,
+      "loss": 0.944,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966289692316944e-05,
+      "loss": 0.9242,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965725835360944e-05,
+      "loss": 0.9355,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965157309841202e-05,
+      "loss": 0.971,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964584116024054e-05,
+      "loss": 0.9648,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996400625417802e-05,
+      "loss": 0.8652,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99634237245738e-05,
+      "loss": 0.887,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962836527484296e-05,
+      "loss": 0.9375,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962244663184582e-05,
+      "loss": 0.9551,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961648131951924e-05,
+      "loss": 0.9157,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996104693406578e-05,
+      "loss": 0.9411,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960441069807778e-05,
+      "loss": 0.9772,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995983053946175e-05,
+      "loss": 0.9561,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959215343313703e-05,
+      "loss": 0.862,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958595481651833e-05,
+      "loss": 0.8919,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957970954766528e-05,
+      "loss": 0.9788,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 0.9391,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956707906498046e-05,
+      "loss": 0.9271,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995606938570656e-05,
+      "loss": 0.9551,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995542620087502e-05,
+      "loss": 0.9362,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954778352304724e-05,
+      "loss": 0.9102,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954125840299165e-05,
+      "loss": 0.9743,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953468665164023e-05,
+      "loss": 0.9225,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995280682720716e-05,
+      "loss": 0.8965,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952140326738627e-05,
+      "loss": 0.9456,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951469164070647e-05,
+      "loss": 0.8766,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950793339517632e-05,
+      "loss": 0.9186,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950112853396188e-05,
+      "loss": 0.8978,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949427706025092e-05,
+      "loss": 0.9307,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948737897725307e-05,
+      "loss": 0.9229,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948043428819988e-05,
+      "loss": 0.8848,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947344299634464e-05,
+      "loss": 0.9251,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994664051049625e-05,
+      "loss": 0.8896,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945932061735047e-05,
+      "loss": 0.8932,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.931,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944501186673378e-05,
+      "loss": 0.9495,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943778761043223e-05,
+      "loss": 0.918,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943051677130696e-05,
+      "loss": 0.9147,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942319935276412e-05,
+      "loss": 0.9476,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941583535823162e-05,
+      "loss": 0.9268,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940842479115922e-05,
+      "loss": 0.9538,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994009676550185e-05,
+      "loss": 0.9427,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993934639533028e-05,
+      "loss": 0.9072,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993859136895274e-05,
+      "loss": 0.8923,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937831686722925e-05,
+      "loss": 0.971,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937067348996716e-05,
+      "loss": 0.9691,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993629835613218e-05,
+      "loss": 0.9212,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935524708489556e-05,
+      "loss": 0.915,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993474640643128e-05,
+      "loss": 0.9102,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933963450321944e-05,
+      "loss": 0.8542,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993317584052834e-05,
+      "loss": 0.9225,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.9033,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931586661366363e-05,
+      "loss": 0.8905,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930785092742462e-05,
+      "loss": 0.9033,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929978871923228e-05,
+      "loss": 0.9209,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992916799928635e-05,
+      "loss": 0.9443,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992835247521169e-05,
+      "loss": 0.9255,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927532300081286e-05,
+      "loss": 0.9821,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992670747427936e-05,
+      "loss": 0.8789,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925877998192318e-05,
+      "loss": 0.9251,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925043872208732e-05,
+      "loss": 0.9017,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992420509671936e-05,
+      "loss": 0.9036,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923361672117133e-05,
+      "loss": 0.8805,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992251359879717e-05,
+      "loss": 0.9147,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921660877156756e-05,
+      "loss": 0.9368,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992080350759536e-05,
+      "loss": 0.9421,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991994149051463e-05,
+      "loss": 0.8809,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991907482631838e-05,
+      "loss": 0.8408,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918203515412616e-05,
+      "loss": 0.8822,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917327558205517e-05,
+      "loss": 0.9111,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991644695510743e-05,
+      "loss": 0.8883,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915561706530882e-05,
+      "loss": 0.9863,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914671812890586e-05,
+      "loss": 0.9508,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991377727460342e-05,
+      "loss": 0.8958,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912878092088433e-05,
+      "loss": 0.8874,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991197426576687e-05,
+      "loss": 0.8524,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911065796062137e-05,
+      "loss": 0.9076,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991015268339981e-05,
+      "loss": 0.9023,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909234928207657e-05,
+      "loss": 0.929,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908312530915603e-05,
+      "loss": 0.9189,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990738549195576e-05,
+      "loss": 0.9069,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906453811762415e-05,
+      "loss": 0.8997,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990551749077202e-05,
+      "loss": 0.8411,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904576529423202e-05,
+      "loss": 0.8662,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990363092815677e-05,
+      "loss": 0.9209,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.877,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901725807645154e-05,
+      "loss": 0.9476,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900766289292442e-05,
+      "loss": 0.9124,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899802132807073e-05,
+      "loss": 0.8877,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989883333864071e-05,
+      "loss": 0.9193,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897859907247197e-05,
+      "loss": 0.9508,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896881839082554e-05,
+      "loss": 0.9079,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895899134604968e-05,
+      "loss": 0.9128,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894911794274797e-05,
+      "loss": 0.9365,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989391981855457e-05,
+      "loss": 0.8737,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892923207908997e-05,
+      "loss": 0.8997,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891921962804942e-05,
+      "loss": 0.9222,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890916083711463e-05,
+      "loss": 0.9255,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889905571099765e-05,
+      "loss": 0.8988,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888890425443238e-05,
+      "loss": 0.9183,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887870647217446e-05,
+      "loss": 0.9697,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886846236900102e-05,
+      "loss": 0.8662,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885817194971116e-05,
+      "loss": 0.8799,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884783521912554e-05,
+      "loss": 0.89,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883745218208647e-05,
+      "loss": 0.9066,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882702284345803e-05,
+      "loss": 0.9248,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881654720812594e-05,
+      "loss": 0.863,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988060252809977e-05,
+      "loss": 0.8711,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987954570670024e-05,
+      "loss": 0.9391,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878484257109085e-05,
+      "loss": 0.8978,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877418179823553e-05,
+      "loss": 0.8833,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876347475343062e-05,
+      "loss": 0.873,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875272144169195e-05,
+      "loss": 0.8516,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98741921868057e-05,
+      "loss": 0.9564,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987310760375851e-05,
+      "loss": 0.8939,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872018395535694e-05,
+      "loss": 0.9658,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870924562647512e-05,
+      "loss": 0.9036,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869826105606382e-05,
+      "loss": 0.9046,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868723024926893e-05,
+      "loss": 0.9082,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867615321125796e-05,
+      "loss": 0.8929,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866502994722e-05,
+      "loss": 0.901,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865386046236597e-05,
+      "loss": 0.9261,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986426447619283e-05,
+      "loss": 0.9036,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863138285116115e-05,
+      "loss": 0.902,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862007473534026e-05,
+      "loss": 0.8691,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986087204197631e-05,
+      "loss": 0.9287,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859731990974867e-05,
+      "loss": 0.8848,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858587321063777e-05,
+      "loss": 0.9131,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857438032779265e-05,
+      "loss": 0.9147,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985628412665973e-05,
+      "loss": 0.9147,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985512560324574e-05,
+      "loss": 0.8893,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853962463080013e-05,
+      "loss": 0.9326,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852794706707443e-05,
+      "loss": 0.8519,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851622334675065e-05,
+      "loss": 0.9661,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850445347532104e-05,
+      "loss": 0.9349,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849263745829934e-05,
+      "loss": 0.9137,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.877,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984688670096425e-05,
+      "loss": 0.8249,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845691258914293e-05,
+      "loss": 0.8945,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844491204532238e-05,
+      "loss": 0.9326,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984328653838026e-05,
+      "loss": 0.957,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984207726102269e-05,
+      "loss": 0.9163,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840863373026046e-05,
+      "loss": 0.8779,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839644874958976e-05,
+      "loss": 0.9102,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838421767392305e-05,
+      "loss": 0.8893,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983719405089901e-05,
+      "loss": 0.9639,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835961726054228e-05,
+      "loss": 0.9212,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983472479343526e-05,
+      "loss": 0.9271,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983348325362157e-05,
+      "loss": 0.9066,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983223710719475e-05,
+      "loss": 0.8701,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830986354738593e-05,
+      "loss": 0.9004,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.9069,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828471034084117e-05,
+      "loss": 0.807,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827206467064133e-05,
+      "loss": 0.9365,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982593729637147e-05,
+      "loss": 0.8734,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824663522600682e-05,
+      "loss": 0.9225,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823385146348485e-05,
+      "loss": 0.8924,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822102168213754e-05,
+      "loss": 0.9453,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820814588797508e-05,
+      "loss": 0.861,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819522408702937e-05,
+      "loss": 0.8737,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818225628535372e-05,
+      "loss": 0.8958,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816924248902304e-05,
+      "loss": 0.8682,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815618270413388e-05,
+      "loss": 0.8438,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981430769368042e-05,
+      "loss": 0.9004,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812992519317357e-05,
+      "loss": 0.9128,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811672747940303e-05,
+      "loss": 0.9193,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810348380167527e-05,
+      "loss": 0.8535,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809019416619445e-05,
+      "loss": 0.8577,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980768585791862e-05,
+      "loss": 0.8343,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806347704689778e-05,
+      "loss": 0.8802,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805004957559795e-05,
+      "loss": 0.8854,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803657617157693e-05,
+      "loss": 0.9398,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980230568411465e-05,
+      "loss": 0.887,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800949159063995e-05,
+      "loss": 0.9609,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799588042641214e-05,
+      "loss": 0.8962,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798222335483933e-05,
+      "loss": 0.9632,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796852038231932e-05,
+      "loss": 0.8857,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795477151527148e-05,
+      "loss": 0.8636,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979409767601366e-05,
+      "loss": 0.8743,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792713612337702e-05,
+      "loss": 0.9219,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791324961147652e-05,
+      "loss": 0.86,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789931723094046e-05,
+      "loss": 0.8551,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978853389882956e-05,
+      "loss": 0.8822,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978713148900902e-05,
+      "loss": 0.8503,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785724494289402e-05,
+      "loss": 0.8978,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978431291532983e-05,
+      "loss": 0.8747,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782896752791576e-05,
+      "loss": 0.9437,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.8444,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780050679634843e-05,
+      "loss": 0.8965,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778620770349637e-05,
+      "loss": 0.9535,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777186280152304e-05,
+      "loss": 0.8454,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775747209714847e-05,
+      "loss": 0.9469,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774303559711412e-05,
+      "loss": 0.8838,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97728553308183e-05,
+      "loss": 0.8346,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977140252371395e-05,
+      "loss": 0.9857,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769945139078942e-05,
+      "loss": 0.9209,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768483177596008e-05,
+      "loss": 0.8379,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767016639950027e-05,
+      "loss": 0.8704,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765545526828004e-05,
+      "loss": 0.9196,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976406983891911e-05,
+      "loss": 0.9251,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762589576914645e-05,
+      "loss": 0.8936,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976110474150806e-05,
+      "loss": 0.9535,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759615333394933e-05,
+      "loss": 0.8617,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758121353273004e-05,
+      "loss": 0.8981,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756622801842144e-05,
+      "loss": 0.9069,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975511967980437e-05,
+      "loss": 0.9196,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975361198786383e-05,
+      "loss": 0.9176,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975209972672683e-05,
+      "loss": 0.8779,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750582897101798e-05,
+      "loss": 0.8717,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749061499699315e-05,
+      "loss": 0.9382,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97475355352321e-05,
+      "loss": 0.9069,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746005004415004e-05,
+      "loss": 0.8926,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744469907965027e-05,
+      "loss": 0.8535,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742930246601305e-05,
+      "loss": 0.9398,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741386021045105e-05,
+      "loss": 0.8776,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973983723201984e-05,
+      "loss": 0.8568,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973828388025106e-05,
+      "loss": 0.873,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973672596646645e-05,
+      "loss": 0.8896,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735163491395836e-05,
+      "loss": 0.8887,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733596455771176e-05,
+      "loss": 0.916,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732024860326566e-05,
+      "loss": 0.8706,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.9209,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728867992924567e-05,
+      "loss": 0.8828,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972728272244605e-05,
+      "loss": 0.8893,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725692895105328e-05,
+      "loss": 0.9225,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724098511647173e-05,
+      "loss": 0.8691,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722499572818496e-05,
+      "loss": 0.9049,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720896079368338e-05,
+      "loss": 0.8929,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719288032047874e-05,
+      "loss": 0.8682,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717675431610413e-05,
+      "loss": 0.9251,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716058278811402e-05,
+      "loss": 0.8838,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714436574408408e-05,
+      "loss": 0.9339,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971281031916114e-05,
+      "loss": 0.943,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971117951383144e-05,
+      "loss": 0.887,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709544159183273e-05,
+      "loss": 0.8753,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707904255982748e-05,
+      "loss": 0.8581,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706259804998093e-05,
+      "loss": 0.89,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704610806999668e-05,
+      "loss": 0.8538,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 0.8561,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701299173053616e-05,
+      "loss": 0.903,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699636538657364e-05,
+      "loss": 0.8486,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697969360350098e-05,
+      "loss": 0.8828,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696297638912817e-05,
+      "loss": 0.9242,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969462137512867e-05,
+      "loss": 0.8919,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692940569782914e-05,
+      "loss": 0.8721,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969125522366295e-05,
+      "loss": 0.8343,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968956533755829e-05,
+      "loss": 0.8652,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968787091226059e-05,
+      "loss": 0.8988,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686171948563617e-05,
+      "loss": 0.9561,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684468447263276e-05,
+      "loss": 0.9261,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968276040915759e-05,
+      "loss": 0.8662,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681047835046708e-05,
+      "loss": 0.8773,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679330725732915e-05,
+      "loss": 0.9079,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96776090820206e-05,
+      "loss": 0.8747,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675882904716294e-05,
+      "loss": 0.8828,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967415219462864e-05,
+      "loss": 0.9199,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672416952568416e-05,
+      "loss": 0.9173,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967067717934852e-05,
+      "loss": 0.9183,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668932875783958e-05,
+      "loss": 0.9157,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667184042691877e-05,
+      "loss": 0.891,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665430680891536e-05,
+      "loss": 0.9378,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663672791204328e-05,
+      "loss": 0.8766,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661910374453743e-05,
+      "loss": 0.8802,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660143431465416e-05,
+      "loss": 0.9225,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658371963067088e-05,
+      "loss": 0.8428,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656595970088627e-05,
+      "loss": 0.8809,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654815453362016e-05,
+      "loss": 0.9193,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653030413721366e-05,
+      "loss": 0.8545,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965124085200289e-05,
+      "loss": 0.9071,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964944676904494e-05,
+      "loss": 0.8646,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964764816568797e-05,
+      "loss": 0.8893,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645845042774555e-05,
+      "loss": 0.8656,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964403740114939e-05,
+      "loss": 0.9056,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642225241659294e-05,
+      "loss": 0.8906,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640408565153187e-05,
+      "loss": 0.8828,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638587372482115e-05,
+      "loss": 0.8304,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963676166449924e-05,
+      "loss": 0.9033,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634931442059833e-05,
+      "loss": 0.9027,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633096706021287e-05,
+      "loss": 0.8613,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96312574572431e-05,
+      "loss": 0.8862,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629413696586903e-05,
+      "loss": 0.9056,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962756542491641e-05,
+      "loss": 0.8802,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625712643097487e-05,
+      "loss": 0.8851,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962385535199807e-05,
+      "loss": 0.8851,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621993552488245e-05,
+      "loss": 0.8831,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962012724544019e-05,
+      "loss": 0.8343,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 0.9372,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616381112228666e-05,
+      "loss": 0.8711,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614501287820125e-05,
+      "loss": 0.8411,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 0.8955,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610728127800604e-05,
+      "loss": 0.889,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960883479395721e-05,
+      "loss": 0.9137,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606936958739963e-05,
+      "loss": 0.9202,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960503462303793e-05,
+      "loss": 0.8636,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603127787742273e-05,
+      "loss": 0.8704,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601216453746285e-05,
+      "loss": 0.8294,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959930062194534e-05,
+      "loss": 0.8848,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597380293236947e-05,
+      "loss": 0.8958,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595455468520695e-05,
+      "loss": 0.8789,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593526148698293e-05,
+      "loss": 0.8232,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591592334673565e-05,
+      "loss": 0.8529,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589654027352412e-05,
+      "loss": 0.8841,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958771122764287e-05,
+      "loss": 0.8555,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585763936455066e-05,
+      "loss": 0.9202,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583812154701227e-05,
+      "loss": 0.9059,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581855883295685e-05,
+      "loss": 0.8695,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957989512315489e-05,
+      "loss": 0.9049,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577929875197376e-05,
+      "loss": 0.9141,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957596014034379e-05,
+      "loss": 0.9163,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573985919516875e-05,
+      "loss": 0.846,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572007213641483e-05,
+      "loss": 0.89,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957002402364456e-05,
+      "loss": 0.9264,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568036350455147e-05,
+      "loss": 0.8464,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956604419500441e-05,
+      "loss": 0.8952,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564047558225585e-05,
+      "loss": 0.8288,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562046441054026e-05,
+      "loss": 0.9089,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956004084442718e-05,
+      "loss": 0.8633,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955803076928459e-05,
+      "loss": 0.8916,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556016216567897e-05,
+      "loss": 0.9274,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553997187220856e-05,
+      "loss": 0.8838,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551973682189292e-05,
+      "loss": 0.8395,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549945702421144e-05,
+      "loss": 0.87,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547913248866447e-05,
+      "loss": 0.8844,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954587632247732e-05,
+      "loss": 0.8971,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543834924207993e-05,
+      "loss": 0.8932,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541789055014786e-05,
+      "loss": 0.9046,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95397387158561e-05,
+      "loss": 0.9268,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953768390769245e-05,
+      "loss": 0.902,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535624631486436e-05,
+      "loss": 0.8844,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533560888202742e-05,
+      "loss": 0.8551,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953149267880816e-05,
+      "loss": 0.9414,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529420004271568e-05,
+      "loss": 0.8581,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952734286556393e-05,
+      "loss": 0.8428,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525261263658315e-05,
+      "loss": 0.8301,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523175199529864e-05,
+      "loss": 0.8499,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521084674155828e-05,
+      "loss": 0.8994,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518989688515533e-05,
+      "loss": 0.9049,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95168902435904e-05,
+      "loss": 0.8932,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514786340363944e-05,
+      "loss": 0.8809,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951267797982176e-05,
+      "loss": 0.8831,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.8639,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950844789074305e-05,
+      "loss": 0.915,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506326164188153e-05,
+      "loss": 0.9098,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504199984280802e-05,
+      "loss": 0.8558,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502069352017026e-05,
+      "loss": 0.8389,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949993426839495e-05,
+      "loss": 0.8623,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497794734414782e-05,
+      "loss": 0.8656,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495650751078802e-05,
+      "loss": 0.9001,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493502319391396e-05,
+      "loss": 0.9082,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491349440359014e-05,
+      "loss": 0.9336,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489192114990202e-05,
+      "loss": 0.8906,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487030344295586e-05,
+      "loss": 0.8968,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484864129287872e-05,
+      "loss": 0.9255,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948269347098185e-05,
+      "loss": 0.8796,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480518370394386e-05,
+      "loss": 0.902,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478338828544437e-05,
+      "loss": 0.8597,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476154846453037e-05,
+      "loss": 0.8665,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473966425143292e-05,
+      "loss": 0.9082,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471773565640405e-05,
+      "loss": 0.8926,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469576268971635e-05,
+      "loss": 0.891,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946737453616634e-05,
+      "loss": 0.8545,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465168368255946e-05,
+      "loss": 0.8711,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946295776627396e-05,
+      "loss": 0.8232,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946074273125596e-05,
+      "loss": 0.8831,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458523264239612e-05,
+      "loss": 0.8135,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945629936626465e-05,
+      "loss": 0.8835,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945407103837288e-05,
+      "loss": 0.8802,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94518382816082e-05,
+      "loss": 0.8906,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944960109701656e-05,
+      "loss": 0.9342,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447359485646002e-05,
+      "loss": 0.9134,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445113448546636e-05,
+      "loss": 0.7992,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442862986770645e-05,
+      "loss": 0.8763,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440608101372283e-05,
+      "loss": 0.8766,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438348793407882e-05,
+      "loss": 0.8577,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436085063935837e-05,
+      "loss": 0.8903,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943381691401662e-05,
+      "loss": 0.9036,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431544344712776e-05,
+      "loss": 0.8851,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429267357088914e-05,
+      "loss": 0.8724,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942698595221172e-05,
+      "loss": 0.8167,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942470013114994e-05,
+      "loss": 0.8525,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94224098949744e-05,
+      "loss": 0.9167,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420115244757985e-05,
+      "loss": 0.8851,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941781618157565e-05,
+      "loss": 0.8714,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415512706504425e-05,
+      "loss": 0.8096,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413204820623392e-05,
+      "loss": 0.8971,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410892525013717e-05,
+      "loss": 0.8975,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408575820758616e-05,
+      "loss": 0.9059,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940625470894338e-05,
+      "loss": 0.8896,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403929190655358e-05,
+      "loss": 0.819,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401599266983973e-05,
+      "loss": 0.8174,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93992649390207e-05,
+      "loss": 0.8149,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.8747,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394583074594738e-05,
+      "loss": 0.8411,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939223554032532e-05,
+      "loss": 0.9329,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938988360615057e-05,
+      "loss": 0.8587,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938752727317227e-05,
+      "loss": 0.9251,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938516654249428e-05,
+      "loss": 0.9463,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382801415222516e-05,
+      "loss": 0.8942,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938043189246494e-05,
+      "loss": 0.8294,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378057975331594e-05,
+      "loss": 0.7619,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375679664934556e-05,
+      "loss": 0.8457,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373296962387988e-05,
+      "loss": 0.8372,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937090986880808e-05,
+      "loss": 0.8555,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368518385313108e-05,
+      "loss": 0.9277,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366122513023385e-05,
+      "loss": 0.8372,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363722253061287e-05,
+      "loss": 0.8219,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936131760655124e-05,
+      "loss": 0.8245,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358908574619735e-05,
+      "loss": 0.8949,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356495158395317e-05,
+      "loss": 0.8786,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935407735900857e-05,
+      "loss": 0.8786,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351655177592143e-05,
+      "loss": 0.806,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349228615280736e-05,
+      "loss": 0.96,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346797673211108e-05,
+      "loss": 0.8324,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344362352522055e-05,
+      "loss": 0.8864,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341922654354435e-05,
+      "loss": 0.9196,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339478579851156e-05,
+      "loss": 0.8877,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337030130157166e-05,
+      "loss": 0.8656,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933457730641948e-05,
+      "loss": 0.8607,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933212010978715e-05,
+      "loss": 0.8945,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329658541411277e-05,
+      "loss": 0.9193,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327192602445016e-05,
+      "loss": 0.9548,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.8506,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932224761736416e-05,
+      "loss": 0.7868,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319768573566104e-05,
+      "loss": 0.8665,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317285163810732e-05,
+      "loss": 0.8626,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9314797389261426e-05,
+      "loss": 0.9095,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312305251083613e-05,
+      "loss": 0.9193,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930980875044477e-05,
+      "loss": 0.8617,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307307888514407e-05,
+      "loss": 0.9232,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304802666464084e-05,
+      "loss": 0.9092,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302293085467404e-05,
+      "loss": 0.8376,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929977914670001e-05,
+      "loss": 0.903,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297260851339588e-05,
+      "loss": 0.86,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294738200565856e-05,
+      "loss": 0.8258,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929221119556059e-05,
+      "loss": 0.8571,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289679837507593e-05,
+      "loss": 0.887,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287144127592704e-05,
+      "loss": 0.8431,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284604067003812e-05,
+      "loss": 0.8691,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282059656930843e-05,
+      "loss": 0.8356,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927951089856575e-05,
+      "loss": 0.8691,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927695779310253e-05,
+      "loss": 0.8607,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274400341737214e-05,
+      "loss": 0.8714,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 0.8636,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269272406094617e-05,
+      "loss": 0.863,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266701924219576e-05,
+      "loss": 0.8721,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926412710124693e-05,
+      "loss": 0.848,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926154793838288e-05,
+      "loss": 0.819,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258964436835668e-05,
+      "loss": 0.8542,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256376597815565e-05,
+      "loss": 0.9004,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253784422534876e-05,
+      "loss": 0.9186,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925118791220794e-05,
+      "loss": 0.8695,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924858706805112e-05,
+      "loss": 0.8571,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245981891282812e-05,
+      "loss": 0.8356,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924337238312344e-05,
+      "loss": 0.8695,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924075854479547e-05,
+      "loss": 0.861,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238140377523375e-05,
+      "loss": 0.8477,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923551788253367e-05,
+      "loss": 0.8281,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232891061054896e-05,
+      "loss": 0.9245,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230259914317618e-05,
+      "loss": 0.8428,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227624443554425e-05,
+      "loss": 0.8548,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922498464999994e-05,
+      "loss": 0.8581,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222340534890803e-05,
+      "loss": 0.8594,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219692099465683e-05,
+      "loss": 0.8477,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921703934496527e-05,
+      "loss": 0.9092,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921438227263228e-05,
+      "loss": 0.8841,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921172088371145e-05,
+      "loss": 0.8669,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920905517944954e-05,
+      "loss": 0.9049,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206385161095328e-05,
+      "loss": 0.8981,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203710829899617e-05,
+      "loss": 0.8444,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201032187115235e-05,
+      "loss": 0.8525,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198349233997015e-05,
+      "loss": 0.8258,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195661971801825e-05,
+      "loss": 0.8226,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919297040178855e-05,
+      "loss": 0.834,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190274525218077e-05,
+      "loss": 0.8665,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187574343353324e-05,
+      "loss": 0.8786,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184869857459233e-05,
+      "loss": 0.86,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182161068802742e-05,
+      "loss": 0.9049,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917944797865282e-05,
+      "loss": 0.8639,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917673058828045e-05,
+      "loss": 0.849,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917400889895862e-05,
+      "loss": 0.8682,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917128291196234e-05,
+      "loss": 0.8747,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168552628568632e-05,
+      "loss": 0.8285,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165818050056528e-05,
+      "loss": 0.8773,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163079177707077e-05,
+      "loss": 0.8109,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916033601280334e-05,
+      "loss": 0.8525,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157588556630373e-05,
+      "loss": 0.917,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154836810475266e-05,
+      "loss": 0.9043,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152080775627105e-05,
+      "loss": 0.9577,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149320453376988e-05,
+      "loss": 0.9134,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914655584501802e-05,
+      "loss": 0.8678,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143786951845312e-05,
+      "loss": 0.8737,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914101377515599e-05,
+      "loss": 0.86,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138236316249178e-05,
+      "loss": 0.89,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.8135,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913266855698963e-05,
+      "loss": 0.8566,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129878259245182e-05,
+      "loss": 0.9365,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127083684499805e-05,
+      "loss": 0.8561,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912428483406266e-05,
+      "loss": 0.8727,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121481709244903e-05,
+      "loss": 0.8258,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118674311359685e-05,
+      "loss": 0.904,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911586264172217e-05,
+      "loss": 0.849,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113046701649517e-05,
+      "loss": 0.8968,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110226492460886e-05,
+      "loss": 0.8348,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107402015477435e-05,
+      "loss": 0.8307,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104573272022333e-05,
+      "loss": 0.8151,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910174026342073e-05,
+      "loss": 0.8337,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909890299099979e-05,
+      "loss": 0.8421,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909606145608866e-05,
+      "loss": 0.8477,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093215660018494e-05,
+      "loss": 0.9095,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909036560412244e-05,
+      "loss": 0.8975,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087511289735646e-05,
+      "loss": 0.7938,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084652718195237e-05,
+      "loss": 0.8561,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081789890840357e-05,
+      "loss": 0.8994,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078922809012127e-05,
+      "loss": 0.9082,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076051474053666e-05,
+      "loss": 0.8822,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073175887310086e-05,
+      "loss": 0.8219,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070296050128486e-05,
+      "loss": 0.8535,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067411963857967e-05,
+      "loss": 0.8743,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064523629849613e-05,
+      "loss": 0.8766,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906163104945649e-05,
+      "loss": 0.8919,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058734224033673e-05,
+      "loss": 0.8337,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055833154938208e-05,
+      "loss": 0.8997,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905292784352914e-05,
+      "loss": 0.8031,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050018291167492e-05,
+      "loss": 0.9274,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904710449921628e-05,
+      "loss": 0.862,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044186469040508e-05,
+      "loss": 0.8704,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041264202007158e-05,
+      "loss": 0.8029,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038337699485207e-05,
+      "loss": 0.8304,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035406962845604e-05,
+      "loss": 0.8529,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903247199346129e-05,
+      "loss": 0.8597,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029532792707187e-05,
+      "loss": 0.8743,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90265893619602e-05,
+      "loss": 0.8883,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902364170259921e-05,
+      "loss": 0.8958,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020689816005086e-05,
+      "loss": 0.8737,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017733703560677e-05,
+      "loss": 0.8164,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014773366650807e-05,
+      "loss": 0.8405,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901180880666228e-05,
+      "loss": 0.8864,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008840024983883e-05,
+      "loss": 0.8962,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005867023006374e-05,
+      "loss": 0.8343,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002889802122495e-05,
+      "loss": 0.7799,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899990836372696e-05,
+      "loss": 0.8678,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996922709216456e-05,
+      "loss": 0.9163,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899393283998965e-05,
+      "loss": 0.8691,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990938757447184e-05,
+      "loss": 0.8926,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987940462991673e-05,
+      "loss": 0.821,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984937958027697e-05,
+      "loss": 0.9124,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981931243961823e-05,
+      "loss": 0.8405,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978920322202582e-05,
+      "loss": 0.9007,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975905194160473e-05,
+      "loss": 0.8506,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897288586124797e-05,
+      "loss": 0.8704,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969862324879513e-05,
+      "loss": 0.8783,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966834586471517e-05,
+      "loss": 0.9502,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963802647442368e-05,
+      "loss": 0.9128,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89607665092124e-05,
+      "loss": 0.9137,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895772617320394e-05,
+      "loss": 0.8708,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954681640841267e-05,
+      "loss": 0.8551,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 0.8737,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894857999276023e-05,
+      "loss": 0.8571,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945522879900262e-05,
+      "loss": 0.8763,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942461576402858e-05,
+      "loss": 0.8893,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893939608370212e-05,
+      "loss": 0.8415,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.8555,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893325253643689e-05,
+      "loss": 0.8763,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930174484750413e-05,
+      "loss": 0.847,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927092249616637e-05,
+      "loss": 0.8314,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892400583247948e-05,
+      "loss": 0.8649,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920915234784805e-05,
+      "loss": 0.8499,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8917820457980444e-05,
+      "loss": 0.8226,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891472150351618e-05,
+      "loss": 0.9316,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891161837284375e-05,
+      "loss": 0.8688,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890851106741686e-05,
+      "loss": 0.8438,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905399588691165e-05,
+      "loss": 0.8737,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890228393812427e-05,
+      "loss": 0.846,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899164117175745e-05,
+      "loss": 0.8503,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88960401273071e-05,
+      "loss": 0.833,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892911969981806e-05,
+      "loss": 0.8398,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888977964666529e-05,
+      "loss": 0.8411,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888664315882493e-05,
+      "loss": 0.8519,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883502507930044e-05,
+      "loss": 0.8688,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880357695451907e-05,
+      "loss": 0.8701,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877208722863753e-05,
+      "loss": 0.8135,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874055591640746e-05,
+      "loss": 0.8936,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887089830326001e-05,
+      "loss": 0.876,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886773685920062e-05,
+      "loss": 0.8428,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886457126094359e-05,
+      "loss": 0.832,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861401509971878e-05,
+      "loss": 0.8571,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858227607770398e-05,
+      "loss": 0.8379,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855049555826004e-05,
+      "loss": 0.8158,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851867355627482e-05,
+      "loss": 0.8724,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884868100866558e-05,
+      "loss": 0.8135,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884549051643298e-05,
+      "loss": 0.8818,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842295880424305e-05,
+      "loss": 0.8887,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839097102136123e-05,
+      "loss": 0.901,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835894183066934e-05,
+      "loss": 0.875,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832687124717187e-05,
+      "loss": 0.8317,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 0.8721,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826260596187505e-05,
+      "loss": 0.9043,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882304112901815e-05,
+      "loss": 0.8717,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819817528589405e-05,
+      "loss": 0.9219,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88165897964114e-05,
+      "loss": 0.8812,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813357933996218e-05,
+      "loss": 0.8369,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810121942857848e-05,
+      "loss": 0.8376,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806881824512234e-05,
+      "loss": 0.8241,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803637580477254e-05,
+      "loss": 0.8112,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800389212272708e-05,
+      "loss": 0.862,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879713672142033e-05,
+      "loss": 0.8376,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793880109443797e-05,
+      "loss": 0.8473,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87906193778687e-05,
+      "loss": 0.8724,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787354528222573e-05,
+      "loss": 0.9209,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878408556203487e-05,
+      "loss": 0.8818,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878081248083698e-05,
+      "loss": 0.8268,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777535286162217e-05,
+      "loss": 0.9245,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877425397954582e-05,
+      "loss": 0.8454,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877096856252496e-05,
+      "loss": 0.82,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767679036638724e-05,
+      "loss": 0.9196,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764385403428137e-05,
+      "loss": 0.8805,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761087664436137e-05,
+      "loss": 0.8968,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875778582120759e-05,
+      "loss": 0.8304,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754479875289283e-05,
+      "loss": 0.8685,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751169828229927e-05,
+      "loss": 0.8942,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747855681580156e-05,
+      "loss": 0.8828,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744537436892517e-05,
+      "loss": 0.8415,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741215095721483e-05,
+      "loss": 0.8844,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737888659623444e-05,
+      "loss": 0.8402,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734558130156713e-05,
+      "loss": 0.8656,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731223508881512e-05,
+      "loss": 0.8662,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727884797359984e-05,
+      "loss": 0.7992,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724541997156194e-05,
+      "loss": 0.7712,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872119510983611e-05,
+      "loss": 0.8958,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717844136967626e-05,
+      "loss": 0.8688,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871448908012054e-05,
+      "loss": 0.9023,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711129940866577e-05,
+      "loss": 0.8796,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870776672077936e-05,
+      "loss": 0.8372,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870439942143443e-05,
+      "loss": 0.874,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701028044409246e-05,
+      "loss": 0.9144,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697652591283157e-05,
+      "loss": 0.8962,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694273063637444e-05,
+      "loss": 0.9036,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690889463055285e-05,
+      "loss": 0.807,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687501791121766e-05,
+      "loss": 0.8092,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684110049423888e-05,
+      "loss": 0.8477,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680714239550547e-05,
+      "loss": 0.8298,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677314363092555e-05,
+      "loss": 0.9053,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867391042164262e-05,
+      "loss": 0.8291,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670502416795368e-05,
+      "loss": 0.8822,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866709035014731e-05,
+      "loss": 0.8714,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663674223296876e-05,
+      "loss": 0.7822,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.9121,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656829795392076e-05,
+      "loss": 0.9027,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653401497544063e-05,
+      "loss": 0.8467,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864996914590638e-05,
+      "loss": 0.8548,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864653274208695e-05,
+      "loss": 0.7933,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643092287695604e-05,
+      "loss": 0.8021,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639647784344058e-05,
+      "loss": 0.8358,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636199233645934e-05,
+      "loss": 0.874,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632746637216745e-05,
+      "loss": 0.8968,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86292899966739e-05,
+      "loss": 0.7799,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625829313636707e-05,
+      "loss": 0.8626,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862236458972636e-05,
+      "loss": 0.807,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618895826565957e-05,
+      "loss": 0.8636,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615423025780475e-05,
+      "loss": 0.766,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611946188996794e-05,
+      "loss": 0.8291,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860846531784368e-05,
+      "loss": 0.835,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604980413951782e-05,
+      "loss": 0.9131,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860149147895366e-05,
+      "loss": 0.8643,
+      "step": 924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597998514483724e-05,
+      "loss": 0.9102,
+      "step": 925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594501522178318e-05,
+      "loss": 0.8255,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591000503675635e-05,
+      "loss": 0.8366,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858749546061578e-05,
+      "loss": 0.8271,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583986394640723e-05,
+      "loss": 0.8464,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580473307394334e-05,
+      "loss": 0.8356,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576956200522356e-05,
+      "loss": 0.8219,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573435075672422e-05,
+      "loss": 0.8286,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856990993449405e-05,
+      "loss": 0.8382,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856638077863863e-05,
+      "loss": 0.8213,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8562847609759435e-05,
+      "loss": 0.7972,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559310429511625e-05,
+      "loss": 0.8265,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555769239552232e-05,
+      "loss": 0.8434,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855222404154017e-05,
+      "loss": 0.8695,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854867483713623e-05,
+      "loss": 0.791,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854512162800308e-05,
+      "loss": 0.8652,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854156441580526e-05,
+      "loss": 0.8522,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538003202209186e-05,
+      "loss": 0.8564,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853443798888316e-05,
+      "loss": 0.8216,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530868777497348e-05,
+      "loss": 0.877,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527295569723784e-05,
+      "loss": 0.8535,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852371836723638e-05,
+      "loss": 0.8115,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520137171710923e-05,
+      "loss": 0.8267,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516551984825063e-05,
+      "loss": 0.8206,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851296280825833e-05,
+      "loss": 0.8193,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509369643692115e-05,
+      "loss": 0.8978,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505772492809676e-05,
+      "loss": 0.8226,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.8226,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849856623883851e-05,
+      "loss": 0.8691,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849495713912564e-05,
+      "loss": 0.8617,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849134405984826e-05,
+      "loss": 0.8213,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487727002698955e-05,
+      "loss": 0.8385,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8484105969372184e-05,
+      "loss": 0.9342,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.8298,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476851980973368e-05,
+      "loss": 0.8014,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473219029299537e-05,
+      "loss": 0.8359,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469582108244672e-05,
+      "loss": 0.8665,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465941219512533e-05,
+      "loss": 0.8506,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462296364808742e-05,
+      "loss": 0.8636,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458647545840766e-05,
+      "loss": 0.8232,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454994764317943e-05,
+      "loss": 0.8314,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845133802195146e-05,
+      "loss": 0.8096,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447677320454367e-05,
+      "loss": 0.8786,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444012661541556e-05,
+      "loss": 0.8122,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844034404692979e-05,
+      "loss": 0.8854,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8436671478337666e-05,
+      "loss": 0.8311,
+      "step": 970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432994957485645e-05,
+      "loss": 0.8607,
+      "step": 971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429314486096042e-05,
+      "loss": 0.8356,
+      "step": 972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425630065893014e-05,
+      "loss": 0.8623,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421941698602572e-05,
+      "loss": 0.9333,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418249385952575e-05,
+      "loss": 0.875,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414553129672734e-05,
+      "loss": 0.8271,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410852931494606e-05,
+      "loss": 0.8529,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407148793151586e-05,
+      "loss": 0.8678,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840344071637893e-05,
+      "loss": 0.8229,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399728702913723e-05,
+      "loss": 0.8776,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839601275449491e-05,
+      "loss": 0.9118,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839229287286327e-05,
+      "loss": 0.8369,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838856905976142e-05,
+      "loss": 0.8714,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384841316933833e-05,
+      "loss": 0.8438,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381109646126806e-05,
+      "loss": 0.8063,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377374049088494e-05,
+      "loss": 0.8604,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373634527568877e-05,
+      "loss": 0.9176,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836989108331978e-05,
+      "loss": 0.8307,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366143718094866e-05,
+      "loss": 0.7502,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836239243364963e-05,
+      "loss": 0.8652,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358637231741406e-05,
+      "loss": 0.7443,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.8643,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351115082574513e-05,
+      "loss": 0.8932,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347348138839685e-05,
+      "loss": 0.8538,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343577284689548e-05,
+      "loss": 0.8473,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339802521890608e-05,
+      "loss": 0.8906,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336023852211197e-05,
+      "loss": 0.8252,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332241277421477e-05,
+      "loss": 0.8431,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328454799293442e-05,
+      "loss": 0.861,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832466441960091e-05,
+      "loss": 0.8467,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320870140119535e-05,
+      "loss": 0.8481,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831707196262679e-05,
+      "loss": 0.8525,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313269888901974e-05,
+      "loss": 0.8766,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309463920726217e-05,
+      "loss": 0.8786,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830565405988247e-05,
+      "loss": 0.8283,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301840308155507e-05,
+      "loss": 0.8675,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829802266733193e-05,
+      "loss": 0.8258,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294201139200152e-05,
+      "loss": 0.8714,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 0.8268,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286546428174787e-05,
+      "loss": 0.8727,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282713248867142e-05,
+      "loss": 0.8548,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827887618942318e-05,
+      "loss": 0.8691,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275035251640417e-05,
+      "loss": 0.8488,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271190437318186e-05,
+      "loss": 0.8037,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267341748257636e-05,
+      "loss": 0.8389,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263489186261734e-05,
+      "loss": 0.8236,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259632753135257e-05,
+      "loss": 0.8896,
+      "step": 1017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82557724506848e-05,
+      "loss": 0.9036,
+      "step": 1018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251908280718763e-05,
+      "loss": 0.8034,
+      "step": 1019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248040245047365e-05,
+      "loss": 0.8968,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244168345482635e-05,
+      "loss": 0.8633,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824029258383841e-05,
+      "loss": 0.9261,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823641296193034e-05,
+      "loss": 0.8626,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232529481575874e-05,
+      "loss": 0.8903,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822864214459428e-05,
+      "loss": 0.8717,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8224750952806626e-05,
+      "loss": 0.8743,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220855908035783e-05,
+      "loss": 0.8864,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821695701210644e-05,
+      "loss": 0.8551,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213054266845076e-05,
+      "loss": 0.8711,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209147674079983e-05,
+      "loss": 0.7786,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205237235641243e-05,
+      "loss": 0.8454,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201322953360758e-05,
+      "loss": 0.8102,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197404829072214e-05,
+      "loss": 0.8112,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193482864611102e-05,
+      "loss": 0.8242,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818955706181472e-05,
+      "loss": 0.8366,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818562742252215e-05,
+      "loss": 0.8444,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181693948574285e-05,
+      "loss": 0.7799,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177756641813805e-05,
+      "loss": 0.8167,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173815504085187e-05,
+      "loss": 0.7839,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169870537234707e-05,
+      "loss": 0.8604,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816592174311043e-05,
+      "loss": 0.9238,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816196912356222e-05,
+      "loss": 0.8288,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158012680441723e-05,
+      "loss": 0.888,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154052415602382e-05,
+      "loss": 0.8936,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815008833089944e-05,
+      "loss": 0.8665,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814612042818991e-05,
+      "loss": 0.9004,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814214870933261e-05,
+      "loss": 0.835,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138173176188133e-05,
+      "loss": 0.8877,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134193830618873e-05,
+      "loss": 0.8848,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130210674488994e-05,
+      "loss": 0.8346,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812622370966446e-05,
+      "loss": 0.8607,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812223293801301e-05,
+      "loss": 0.8115,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811823836140416e-05,
+      "loss": 0.9212,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114239981709234e-05,
+      "loss": 0.8558,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110237800801312e-05,
+      "loss": 0.903,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810623182055526e-05,
+      "loss": 0.8188,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102222042847735e-05,
+      "loss": 0.9473,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809820846955716e-05,
+      "loss": 0.8913,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809419110256375e-05,
+      "loss": 0.8045,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.8317,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086144994998104e-05,
+      "loss": 0.8438,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082116258195173e-05,
+      "loss": 0.8652,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807808373522799e-05,
+      "loss": 0.8216,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807404742798564e-05,
+      "loss": 0.8584,
+      "step": 1064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070007338358973e-05,
+      "loss": 0.8483,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065963468240625e-05,
+      "loss": 0.8158,
+      "step": 1066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061915819524995e-05,
+      "loss": 0.8626,
+      "step": 1067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805786439410825e-05,
+      "loss": 0.8229,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053809193888326e-05,
+      "loss": 0.846,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804975022076494e-05,
+      "loss": 0.8005,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045687476639555e-05,
+      "loss": 0.8366,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041620963415418e-05,
+      "loss": 0.8727,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037550682997538e-05,
+      "loss": 0.8708,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033476637292682e-05,
+      "loss": 0.8411,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029398828209387e-05,
+      "loss": 0.8574,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025317257657954e-05,
+      "loss": 0.8444,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802123192755044e-05,
+      "loss": 0.8516,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017142839800667e-05,
+      "loss": 0.792,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013049996324225e-05,
+      "loss": 0.8301,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008953399038442e-05,
+      "loss": 0.82,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004853049862427e-05,
+      "loss": 0.8145,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800074895071704e-05,
+      "loss": 0.874,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799664110352489e-05,
+      "loss": 0.8704,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992529510210347e-05,
+      "loss": 0.8024,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798841417269954e-05,
+      "loss": 0.8184,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984295092920344e-05,
+      "loss": 0.8571,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 0.8205,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797604571427708e-05,
+      "loss": 0.7964,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971915419277525e-05,
+      "loss": 0.8115,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967781389738625e-05,
+      "loss": 0.8107,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796364362759701e-05,
+      "loss": 0.8291,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795950213479107e-05,
+      "loss": 0.9245,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955356913260933e-05,
+      "loss": 0.8454,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795120796494848e-05,
+      "loss": 0.8252,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794705529179733e-05,
+      "loss": 0.8011,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794289889575286e-05,
+      "loss": 0.8193,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938738778762182e-05,
+      "loss": 0.8636,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793457494277415e-05,
+      "loss": 0.8465,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930407389739364e-05,
+      "loss": 0.9004,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926236121610163e-05,
+      "loss": 0.8652,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792206114034063e-05,
+      "loss": 0.8063,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917882447886585e-05,
+      "loss": 0.8242,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791370004620558e-05,
+      "loss": 0.8174,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790951393725692e-05,
+      "loss": 0.8473,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905324123001634e-05,
+      "loss": 0.7799,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901130605402494e-05,
+      "loss": 0.8381,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896933386423998e-05,
+      "loss": 0.8411,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892732468032385e-05,
+      "loss": 0.8301,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888527852195627e-05,
+      "loss": 0.8421,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788431954088343e-05,
+      "loss": 0.8506,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 0.833,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787589183972017e-05,
+      "loss": 0.9053,
+      "step": 1112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871672453817167e-05,
+      "loss": 0.8636,
+      "step": 1113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867449380334834e-05,
+      "loss": 0.8835,
+      "step": 1114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786322262125152e-05,
+      "loss": 0.8779,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858992178547306e-05,
+      "loss": 0.8294,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785475805420399e-05,
+      "loss": 0.8796,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850520250205095e-05,
+      "loss": 0.8164,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784627876853588e-05,
+      "loss": 0.9147,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842033611183308e-05,
+      "loss": 0.9115,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837784780136078e-05,
+      "loss": 0.8265,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833532277384607e-05,
+      "loss": 0.9274,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782927610492103e-05,
+      "loss": 0.903,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825016264739202e-05,
+      "loss": 0.8444,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820752758834696e-05,
+      "loss": 0.8073,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816485589204802e-05,
+      "loss": 0.8161,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812214757848523e-05,
+      "loss": 0.8291,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807940266766595e-05,
+      "loss": 0.7985,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803662117961437e-05,
+      "loss": 0.874,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779938031343721e-05,
+      "loss": 0.8467,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795094855199777e-05,
+      "loss": 0.834,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790805745256703e-05,
+      "loss": 0.8291,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786512985617284e-05,
+      "loss": 0.835,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778221657829251e-05,
+      "loss": 0.8874,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777791652529508e-05,
+      "loss": 0.8359,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773612828639415e-05,
+      "loss": 0.8122,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769305490341623e-05,
+      "loss": 0.918,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764994512419534e-05,
+      "loss": 0.8291,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776067989689268e-05,
+      "loss": 0.8398,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775636164578229e-05,
+      "loss": 0.8665,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77520397611113e-05,
+      "loss": 0.804,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747714244904348e-05,
+      "loss": 0.8311,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774338509918778e-05,
+      "loss": 0.8708,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739052325989628e-05,
+      "loss": 0.8291,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734715927339642e-05,
+      "loss": 0.8053,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730375905269248e-05,
+      "loss": 0.8737,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772603226181159e-05,
+      "loss": 0.8734,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721684999001496e-05,
+      "loss": 0.8711,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717334118875494e-05,
+      "loss": 0.8538,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771297962347181e-05,
+      "loss": 0.8975,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770862151483035e-05,
+      "loss": 0.8223,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704259794992734e-05,
+      "loss": 0.8555,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769989446600225e-05,
+      "loss": 0.8923,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695525529903895e-05,
+      "loss": 0.84,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691152988744354e-05,
+      "loss": 0.8906,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768677684457199e-05,
+      "loss": 0.8405,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768239709943686e-05,
+      "loss": 0.8636,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767801375539071e-05,
+      "loss": 0.873,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767362681448697e-05,
+      "loss": 0.8171,
+      "step": 1159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669236278780758e-05,
+      "loss": 0.821,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664842150328865e-05,
+      "loss": 0.7773,
+      "step": 1161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.8662,
+      "step": 1162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765604312342367e-05,
+      "loss": 0.8304,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651638229092373e-05,
+      "loss": 0.863,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647229750259415e-05,
+      "loss": 0.8447,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642817688990006e-05,
+      "loss": 0.9245,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638402047351025e-05,
+      "loss": 0.932,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763398282741103e-05,
+      "loss": 0.8766,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629560031240264e-05,
+      "loss": 0.8613,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762513366091064e-05,
+      "loss": 0.7884,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620703718495735e-05,
+      "loss": 0.875,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616270206070814e-05,
+      "loss": 0.8555,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611833125712808e-05,
+      "loss": 0.8604,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607392479500325e-05,
+      "loss": 0.846,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602948269513634e-05,
+      "loss": 0.8483,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598500497834678e-05,
+      "loss": 0.8877,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594049166547073e-05,
+      "loss": 0.8555,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589594277736095e-05,
+      "loss": 0.8551,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585135833488692e-05,
+      "loss": 0.9206,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580673835893473e-05,
+      "loss": 0.8398,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576208287040717e-05,
+      "loss": 0.8079,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571739189022365e-05,
+      "loss": 0.8076,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567266543932016e-05,
+      "loss": 0.8473,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7562790353864934e-05,
+      "loss": 0.8317,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755831062091805e-05,
+      "loss": 0.7689,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553827347189937e-05,
+      "loss": 0.8088,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549340534780852e-05,
+      "loss": 0.849,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544850185792688e-05,
+      "loss": 0.9007,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540356302329005e-05,
+      "loss": 0.8281,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535858886495014e-05,
+      "loss": 0.8936,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753135794039759e-05,
+      "loss": 0.8525,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526853466145248e-05,
+      "loss": 0.8558,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752234546584816e-05,
+      "loss": 0.8877,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517833941618166e-05,
+      "loss": 0.8311,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513318895568734e-05,
+      "loss": 0.8464,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508800329814993e-05,
+      "loss": 0.877,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750427824647372e-05,
+      "loss": 0.8167,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749975264766334e-05,
+      "loss": 0.8633,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749522353550392e-05,
+      "loss": 0.8411,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749069091211718e-05,
+      "loss": 0.7925,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748615477962648e-05,
+      "loss": 0.8265,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481615140156837e-05,
+      "loss": 0.819,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747707199583488e-05,
+      "loss": 0.8464,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747252534878891e-05,
+      "loss": 0.8835,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467975201148862e-05,
+      "loss": 0.7975,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463421555046298e-05,
+      "loss": 0.875,
+      "step": 1206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458864412614436e-05,
+      "loss": 0.8317,
+      "step": 1207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454303775988118e-05,
+      "loss": 0.8613,
+      "step": 1208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449739647303835e-05,
+      "loss": 0.8252,
+      "step": 1209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445172028699703e-05,
+      "loss": 0.8613,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440600922315478e-05,
+      "loss": 0.8516,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743602633029255e-05,
+      "loss": 0.8564,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.8311,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742686669790431e-05,
+      "loss": 0.8255,
+      "step": 1214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742228166182994e-05,
+      "loss": 0.863,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417693148698743e-05,
+      "loss": 0.8265,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413101160660267e-05,
+      "loss": 0.8776,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740850569986568e-05,
+      "loss": 0.7988,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740390676846778e-05,
+      "loss": 0.8304,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399304368620996e-05,
+      "loss": 0.8496,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394698502481377e-05,
+      "loss": 0.834,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.8294,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738547637995594e-05,
+      "loss": 0.8154,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738086012789034e-05,
+      "loss": 0.8271,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737624041817233e-05,
+      "loss": 0.862,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371617252966066e-05,
+      "loss": 0.8213,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366990634437328e-05,
+      "loss": 0.8652,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362360564753507e-05,
+      "loss": 0.8031,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735772704608362e-05,
+      "loss": 0.9095,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735309008059829e-05,
+      "loss": 0.8262,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348449670469758e-05,
+      "loss": 0.8646,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343805817871885e-05,
+      "loss": 0.8639,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339158524980138e-05,
+      "loss": 0.8281,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334507793971592e-05,
+      "loss": 0.8438,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329853627024945e-05,
+      "loss": 0.902,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325196026320496e-05,
+      "loss": 0.8359,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320534994040148e-05,
+      "loss": 0.8486,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315870532367423e-05,
+      "loss": 0.8008,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731120264348744e-05,
+      "loss": 0.7881,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306531329586933e-05,
+      "loss": 0.758,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730185659285423e-05,
+      "loss": 0.8506,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729717843547927e-05,
+      "loss": 0.8704,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729249685965359e-05,
+      "loss": 0.8354,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287811867570333e-05,
+      "loss": 0.7767,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283123461424237e-05,
+      "loss": 0.8164,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278431643411643e-05,
+      "loss": 0.8327,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273736415730488e-05,
+      "loss": 0.8379,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726903778058031e-05,
+      "loss": 0.8242,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264335740162244e-05,
+      "loss": 0.8883,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259630296679008e-05,
+      "loss": 0.8037,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254921452334934e-05,
+      "loss": 0.8509,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725020920933593e-05,
+      "loss": 0.8382,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245493569889503e-05,
+      "loss": 0.835,
+      "step": 1253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724077453620475e-05,
+      "loss": 0.8397,
+      "step": 1254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236052110492366e-05,
+      "loss": 0.8802,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231326294964618e-05,
+      "loss": 0.8809,
+      "step": 1256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226597091835377e-05,
+      "loss": 0.8109,
+      "step": 1257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221864503320093e-05,
+      "loss": 0.8779,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217128531635806e-05,
+      "loss": 0.8112,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721238917900114e-05,
+      "loss": 0.8047,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207646447636296e-05,
+      "loss": 0.8158,
+      "step": 1261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202900339763066e-05,
+      "loss": 0.8268,
+      "step": 1262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719815085760482e-05,
+      "loss": 0.849,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 0.9072,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188641779334675e-05,
+      "loss": 0.8958,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183882187677413e-05,
+      "loss": 0.8203,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717911923064442e-05,
+      "loss": 0.7669,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174352910466958e-05,
+      "loss": 0.8887,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169583229377866e-05,
+      "loss": 0.8083,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716481018961156e-05,
+      "loss": 0.9056,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160033793404024e-05,
+      "loss": 0.8444,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155254042992827e-05,
+      "loss": 0.8376,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150470940617087e-05,
+      "loss": 0.861,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145684488517518e-05,
+      "loss": 0.7861,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140894688936383e-05,
+      "loss": 0.7721,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136101544117526e-05,
+      "loss": 0.8617,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713130505630635e-05,
+      "loss": 0.8418,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126505227749826e-05,
+      "loss": 0.8548,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121702060696495e-05,
+      "loss": 0.8581,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116895557396457e-05,
+      "loss": 0.8477,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711208572010137e-05,
+      "loss": 0.8211,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710727255106447e-05,
+      "loss": 0.8135,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102456052540546e-05,
+      "loss": 0.7718,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709763622678593e-05,
+      "loss": 0.8617,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092813076058536e-05,
+      "loss": 0.7871,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087986602617824e-05,
+      "loss": 0.8971,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083156808724817e-05,
+      "loss": 0.8363,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078323696642093e-05,
+      "loss": 0.8857,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073487268633768e-05,
+      "loss": 0.8008,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706864752696554e-05,
+      "loss": 0.8516,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063804473904632e-05,
+      "loss": 0.8219,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058958111719836e-05,
+      "loss": 0.8792,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054108442681487e-05,
+      "loss": 0.8796,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049255469061476e-05,
+      "loss": 0.8434,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044399193133225e-05,
+      "loss": 0.8317,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039539617171724e-05,
+      "loss": 0.8205,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70346767434535e-05,
+      "loss": 0.8529,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702981057425662e-05,
+      "loss": 0.8579,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70249411118607e-05,
+      "loss": 0.7829,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70200683585469e-05,
+      "loss": 0.8226,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015192316597918e-05,
+      "loss": 0.7904,
+      "step": 1301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010312988297993e-05,
+      "loss": 0.8792,
+      "step": 1302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700543037593291e-05,
+      "loss": 0.8652,
+      "step": 1303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000544481789984e-05,
+      "loss": 0.819,
+      "step": 1304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699565530815807e-05,
+      "loss": 0.7933,
+      "step": 1305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699076285732756e-05,
+      "loss": 0.8451,
+      "step": 1306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985867131590383e-05,
+      "loss": 0.7959,
+      "step": 1307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6980968133240002e-05,
+      "loss": 0.8828,
+      "step": 1308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976065864571407e-05,
+      "loss": 0.7891,
+      "step": 1309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6971160327881126e-05,
+      "loss": 0.8392,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966251525467216e-05,
+      "loss": 0.8411,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696133945962927e-05,
+      "loss": 0.8166,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956424132668394e-05,
+      "loss": 0.7933,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951505546887238e-05,
+      "loss": 0.8206,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 0.8249,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941658608082296e-05,
+      "loss": 0.8164,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6936730259671423e-05,
+      "loss": 0.8464,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931798661666103e-05,
+      "loss": 0.7858,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926863816376596e-05,
+      "loss": 0.818,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921925726114694e-05,
+      "loss": 0.8112,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916984393193704e-05,
+      "loss": 0.8337,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691203981992845e-05,
+      "loss": 0.8301,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907092008635278e-05,
+      "loss": 0.8529,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902140961632054e-05,
+      "loss": 0.8551,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689718668123815e-05,
+      "loss": 0.7783,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892229169774463e-05,
+      "loss": 0.8971,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887268429563387e-05,
+      "loss": 0.7946,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882304462928853e-05,
+      "loss": 0.8083,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877337272196287e-05,
+      "loss": 0.7772,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687236685969263e-05,
+      "loss": 0.7809,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867393227746327e-05,
+      "loss": 0.8125,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.8867,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857436314847128e-05,
+      "loss": 0.8662,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852453038558667e-05,
+      "loss": 0.8343,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684746655215643e-05,
+      "loss": 0.8467,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842476857976397e-05,
+      "loss": 0.8522,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837483958356054e-05,
+      "loss": 0.8079,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683248785563438e-05,
+      "loss": 0.8477,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827488552151855e-05,
+      "loss": 0.8525,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822486050250472e-05,
+      "loss": 0.8177,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817480352273706e-05,
+      "loss": 0.8773,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681247146056654e-05,
+      "loss": 0.805,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680745937747545e-05,
+      "loss": 0.8848,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802444105348405e-05,
+      "loss": 0.7485,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679742564653487e-05,
+      "loss": 0.8172,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792404003385797e-05,
+      "loss": 0.8187,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787379178253642e-05,
+      "loss": 0.7682,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678235117349234e-05,
+      "loss": 0.8747,
+      "step": 1348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777319991457325e-05,
+      "loss": 0.7852,
+      "step": 1349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772285634505507e-05,
+      "loss": 0.8613,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67672481049953e-05,
+      "loss": 0.8053,
+      "step": 1351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676220740528659e-05,
+      "loss": 0.8337,
+      "step": 1352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757163537740752e-05,
+      "loss": 0.8288,
+      "step": 1353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752116504720644e-05,
+      "loss": 0.8968,
+      "step": 1354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747066308590614e-05,
+      "loss": 0.9079,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674201295171649e-05,
+      "loss": 0.8073,
+      "step": 1356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 0.8353,
+      "step": 1357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731896765206648e-05,
+      "loss": 0.7969,
+      "step": 1358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726833940309978e-05,
+      "loss": 0.8698,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721767964147307e-05,
+      "loss": 0.8288,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671669883909185e-05,
+      "loss": 0.873,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67116265675183e-05,
+      "loss": 0.8607,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706551151802822e-05,
+      "loss": 0.7822,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670147259432306e-05,
+      "loss": 0.8698,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669639089745812e-05,
+      "loss": 0.8252,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.8623,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6686218095096506e-05,
+      "loss": 0.7969,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681126994365407e-05,
+      "loss": 0.8831,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667603276378027e-05,
+      "loss": 0.8538,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667093540572755e-05,
+      "loss": 0.8366,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665834922595166e-05,
+      "loss": 0.8545,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660731316772503e-05,
+      "loss": 0.8477,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655624590650402e-05,
+      "loss": 0.8255,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650514746621173e-05,
+      "loss": 0.8322,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664540178707858e-05,
+      "loss": 0.8206,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640285714417857e-05,
+      "loss": 0.8864,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663516653103568e-05,
+      "loss": 0.8538,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66300442393302e-05,
+      "loss": 0.8545,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624918841701014e-05,
+      "loss": 0.8281,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619790340549175e-05,
+      "loss": 0.7982,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661465873827719e-05,
+      "loss": 0.8877,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660952403728902e-05,
+      "loss": 0.8548,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6604386239990077e-05,
+      "loss": 0.8766,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659924534878723e-05,
+      "loss": 0.8164,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659410136608878e-05,
+      "loss": 0.8613,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588954294304495e-05,
+      "loss": 0.8031,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583804135845582e-05,
+      "loss": 0.8161,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578650893124686e-05,
+      "loss": 0.8223,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573494568555917e-05,
+      "loss": 0.8167,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568335164554813e-05,
+      "loss": 0.8138,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563172683538357e-05,
+      "loss": 0.8356,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655800712792498e-05,
+      "loss": 0.819,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655283850013454e-05,
+      "loss": 0.7676,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547666802588344e-05,
+      "loss": 0.8288,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542492037709146e-05,
+      "loss": 0.874,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537314207921117e-05,
+      "loss": 0.9242,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653213331564987e-05,
+      "loss": 0.806,
+      "step": 1397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652694936332247e-05,
+      "loss": 0.8639,
+      "step": 1398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652176235336739e-05,
+      "loss": 0.8219,
+      "step": 1399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516572288214555e-05,
+      "loss": 0.8851,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511379170295305e-05,
+      "loss": 0.902,
+      "step": 1401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650618300204242e-05,
+      "loss": 0.7985,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500983785890118e-05,
+      "loss": 0.8161,
+      "step": 1403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495781524274018e-05,
+      "loss": 0.8327,
+      "step": 1404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649057621963119e-05,
+      "loss": 0.8942,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485367874400124e-05,
+      "loss": 0.9124,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648015649102073e-05,
+      "loss": 0.903,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474942071934336e-05,
+      "loss": 0.834,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646972461958371e-05,
+      "loss": 0.8392,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464504136413025e-05,
+      "loss": 0.8675,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459280624867876e-05,
+      "loss": 0.8385,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454054087395284e-05,
+      "loss": 0.8171,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644882452644368e-05,
+      "loss": 0.8796,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443591944462917e-05,
+      "loss": 0.8473,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438356343904258e-05,
+      "loss": 0.8011,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433117727220388e-05,
+      "loss": 0.8747,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.8639,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422631455294787e-05,
+      "loss": 0.8408,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417383804965476e-05,
+      "loss": 0.8704,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412133148335786e-05,
+      "loss": 0.874,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640687948786545e-05,
+      "loss": 0.8128,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401622826015616e-05,
+      "loss": 0.8161,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396363165248823e-05,
+      "loss": 0.806,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391100508029026e-05,
+      "loss": 0.7965,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638583485682157e-05,
+      "loss": 0.8271,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380566214093226e-05,
+      "loss": 0.759,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637529458231215e-05,
+      "loss": 0.8372,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370019963947888e-05,
+      "loss": 0.9102,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364742361471416e-05,
+      "loss": 0.8366,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635946177735508e-05,
+      "loss": 0.8483,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354178214072636e-05,
+      "loss": 0.8582,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634889167409923e-05,
+      "loss": 0.8434,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343602159911407e-05,
+      "loss": 0.8151,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63383096739871e-05,
+      "loss": 0.8815,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633301421880565e-05,
+      "loss": 0.8548,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632771579684776e-05,
+      "loss": 0.8372,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322414410595548e-05,
+      "loss": 0.8343,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631711006253251e-05,
+      "loss": 0.8656,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311802755143534e-05,
+      "loss": 0.8724,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306492490914887e-05,
+      "loss": 0.7878,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630117927233423e-05,
+      "loss": 0.8496,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295863101890603e-05,
+      "loss": 0.8783,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6290543982074424e-05,
+      "loss": 0.875,
+      "step": 1443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628522191537751e-05,
+      "loss": 0.8216,
+      "step": 1444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279896904293034e-05,
+      "loss": 0.8044,
+      "step": 1445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274568951315575e-05,
+      "loss": 0.8529,
+      "step": 1446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626923805894107e-05,
+      "loss": 0.7757,
+      "step": 1447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626390422966683e-05,
+      "loss": 0.8206,
+      "step": 1448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258567465991564e-05,
+      "loss": 0.8099,
+      "step": 1449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625322777041534e-05,
+      "loss": 0.7812,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247885145439602e-05,
+      "loss": 0.891,
+      "step": 1451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624253959356717e-05,
+      "loss": 0.8623,
+      "step": 1452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623719111730222e-05,
+      "loss": 0.8509,
+      "step": 1453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623183971915032e-05,
+      "loss": 0.8649,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62264854016184e-05,
+      "loss": 0.7985,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221128167214742e-05,
+      "loss": 0.863,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215768018449015e-05,
+      "loss": 0.7969,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210404957832244e-05,
+      "loss": 0.7839,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205038987876815e-05,
+      "loss": 0.7824,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6199670111096487e-05,
+      "loss": 0.8597,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194298330006374e-05,
+      "loss": 0.8828,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188923647122946e-05,
+      "loss": 0.7956,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183546064964043e-05,
+      "loss": 0.8669,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178165586048857e-05,
+      "loss": 0.8021,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617278221289793e-05,
+      "loss": 0.8454,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167395948033176e-05,
+      "loss": 0.8447,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162006793977858e-05,
+      "loss": 0.8175,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 0.7806,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615121982839532e-05,
+      "loss": 0.8698,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145822021921386e-05,
+      "loss": 0.8457,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6140421336363447e-05,
+      "loss": 0.7899,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613501777425152e-05,
+      "loss": 0.8262,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129611338116968e-05,
+      "loss": 0.8118,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61242020304925e-05,
+      "loss": 0.8457,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118789853912175e-05,
+      "loss": 0.805,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611337481091139e-05,
+      "loss": 0.8379,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610795690402688e-05,
+      "loss": 0.8737,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102536135796734e-05,
+      "loss": 0.8473,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609711250876038e-05,
+      "loss": 0.8675,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6091686025458577e-05,
+      "loss": 0.7871,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608625668843343e-05,
+      "loss": 0.8089,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080824500228367e-05,
+      "loss": 0.8743,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607538946338817e-05,
+      "loss": 0.8226,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069951580458946e-05,
+      "loss": 0.8006,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064510853988137e-05,
+      "loss": 0.8105,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059067286524514e-05,
+      "loss": 0.8512,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605362088061818e-05,
+      "loss": 0.8353,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048171638820574e-05,
+      "loss": 0.835,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042719563684458e-05,
+      "loss": 0.8766,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037264657763918e-05,
+      "loss": 0.8265,
+      "step": 1490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031806923614376e-05,
+      "loss": 0.8525,
+      "step": 1491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.8402,
+      "step": 1492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020882980856555e-05,
+      "loss": 0.834,
+      "step": 1493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015416777365734e-05,
+      "loss": 0.82,
+      "step": 1494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600994775588081e-05,
+      "loss": 0.849,
+      "step": 1495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600447591896381e-05,
+      "loss": 0.832,
+      "step": 1496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999001269178082e-05,
+      "loss": 0.8239,
+      "step": 1497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993523809088292e-05,
+      "loss": 0.8148,
+      "step": 1498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988043541260417e-05,
+      "loss": 0.8047,
+      "step": 1499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982560468261764e-05,
+      "loss": 0.8887,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977074592660936e-05,
+      "loss": 0.7716,
+      "step": 1501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.7982,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966094443933776e-05,
+      "loss": 0.8665,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960600175951224e-05,
+      "loss": 0.8092,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5955103115654062e-05,
+      "loss": 0.876,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949603265617452e-05,
+      "loss": 0.8265,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594410062841787e-05,
+      "loss": 0.8421,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938595206633088e-05,
+      "loss": 0.8092,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5933087002842188e-05,
+      "loss": 0.8594,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592757601962555e-05,
+      "loss": 0.7225,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592206225956487e-05,
+      "loss": 0.8864,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916545725243124e-05,
+      "loss": 0.8158,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911026419244607e-05,
+      "loss": 0.8548,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905504344154897e-05,
+      "loss": 0.8376,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589997950256088e-05,
+      "loss": 0.7225,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589445189705074e-05,
+      "loss": 0.8464,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888921530213938e-05,
+      "loss": 0.8236,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588338840464125e-05,
+      "loss": 0.7627,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.7907,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5872313887657734e-05,
+      "loss": 0.8337,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866772501434898e-05,
+      "loss": 0.848,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861228366852148e-05,
+      "loss": 0.7959,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855681486506705e-05,
+      "loss": 0.8262,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585013186299707e-05,
+      "loss": 0.8304,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844579498923033e-05,
+      "loss": 0.8193,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839024396885657e-05,
+      "loss": 0.8447,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833466559487305e-05,
+      "loss": 0.8467,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582790598933161e-05,
+      "loss": 0.7972,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822342689023485e-05,
+      "loss": 0.8265,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816776661169133e-05,
+      "loss": 0.757,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581120790837602e-05,
+      "loss": 0.8249,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805636433252892e-05,
+      "loss": 0.7712,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580006223840978e-05,
+      "loss": 0.9102,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579448532645798e-05,
+      "loss": 0.8558,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788905700010058e-05,
+      "loss": 0.8079,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783323361679865e-05,
+      "loss": 0.7879,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777738314082514e-05,
+      "loss": 0.8226,
+      "step": 1537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577215055983438e-05,
+      "loss": 0.8219,
+      "step": 1538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576656010155312e-05,
+      "loss": 0.818,
+      "step": 1539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760966941857646e-05,
+      "loss": 0.8162,
+      "step": 1540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575537108336815e-05,
+      "loss": 0.807,
+      "step": 1541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574977252870607e-05,
+      "loss": 0.8034,
+      "step": 1542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744171280494115e-05,
+      "loss": 0.8014,
+      "step": 1543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738567341356265e-05,
+      "loss": 0.8561,
+      "step": 1544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732960713917746e-05,
+      "loss": 0.8545,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727351400805054e-05,
+      "loss": 0.9014,
+      "step": 1546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5721739404645937e-05,
+      "loss": 0.8229,
+      "step": 1547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57161247280694e-05,
+      "loss": 0.8271,
+      "step": 1548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571050737370571e-05,
+      "loss": 0.7744,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570488734418638e-05,
+      "loss": 0.8617,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569926464214418e-05,
+      "loss": 0.8031,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693639270213138e-05,
+      "loss": 0.834,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568801123102852e-05,
+      "loss": 0.7822,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568238052722685e-05,
+      "loss": 0.793,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676747161445903e-05,
+      "loss": 0.8285,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567111113632469e-05,
+      "loss": 0.7788,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665472454503484e-05,
+      "loss": 0.8271,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659831118623783e-05,
+      "loss": 0.8021,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654187131328343e-05,
+      "loss": 0.7562,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648540495261157e-05,
+      "loss": 0.8337,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642891213067463e-05,
+      "loss": 0.8389,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637239287393725e-05,
+      "loss": 0.8239,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563158472088767e-05,
+      "loss": 0.8558,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625927516198235e-05,
+      "loss": 0.8418,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620267675975606e-05,
+      "loss": 0.8822,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614605202871213e-05,
+      "loss": 0.8001,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56089400995377e-05,
+      "loss": 0.7822,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603272368628956e-05,
+      "loss": 0.833,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597602012800098e-05,
+      "loss": 0.7743,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 0.8099,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558625343700865e-05,
+      "loss": 0.8102,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580575222362435e-05,
+      "loss": 0.8171,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5574894393428856e-05,
+      "loss": 0.8563,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569210952869166e-05,
+      "loss": 0.8216,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556352490334584e-05,
+      "loss": 0.8799,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557836247522574e-05,
+      "loss": 0.8499,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5552144988064292e-05,
+      "loss": 0.8291,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546451127637125e-05,
+      "loss": 0.846,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554075466890844e-05,
+      "loss": 0.8643,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535055614546805e-05,
+      "loss": 0.7738,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552935396722201e-05,
+      "loss": 0.8411,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552364972960506e-05,
+      "loss": 0.9072,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517942904368175e-05,
+      "loss": 0.786,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551223349418479e-05,
+      "loss": 0.8112,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506521501729536e-05,
+      "loss": 0.834,
+      "step": 1585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500806929678276e-05,
+      "loss": 0.8226,
+      "step": 1586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495089780708062e-05,
+      "loss": 0.8307,
+      "step": 1587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489370057497165e-05,
+      "loss": 0.7852,
+      "step": 1588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483647762725055e-05,
+      "loss": 0.8267,
+      "step": 1589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477922899072412e-05,
+      "loss": 0.89,
+      "step": 1590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472195469221115e-05,
+      "loss": 0.8467,
+      "step": 1591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466465475854246e-05,
+      "loss": 0.8083,
+      "step": 1592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5460732921656093e-05,
+      "loss": 0.8737,
+      "step": 1593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545499780931214e-05,
+      "loss": 0.8138,
+      "step": 1594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449260141509063e-05,
+      "loss": 0.8291,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544351992093475e-05,
+      "loss": 0.8812,
+      "step": 1596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437777150278268e-05,
+      "loss": 0.7972,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543203183222989e-05,
+      "loss": 0.874,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426283969481082e-05,
+      "loss": 0.8467,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420533564724495e-05,
+      "loss": 0.8503,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414780620653975e-05,
+      "loss": 0.8568,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540902513996456e-05,
+      "loss": 0.7832,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403267125352475e-05,
+      "loss": 0.8532,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539750657951513e-05,
+      "loss": 0.8441,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391743505151122e-05,
+      "loss": 0.807,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5385977904960228e-05,
+      "loss": 0.7998,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538020978164341e-05,
+      "loss": 0.8483,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374439137902832e-05,
+      "loss": 0.8464,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368665976441802e-05,
+      "loss": 0.821,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362890299964834e-05,
+      "loss": 0.7581,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357112111177616e-05,
+      "loss": 0.8021,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351331412787004e-05,
+      "loss": 0.8094,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534554820750104e-05,
+      "loss": 0.7777,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339762498028937e-05,
+      "loss": 0.8084,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333974287081073e-05,
+      "loss": 0.7829,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328183577369007e-05,
+      "loss": 0.8695,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322390371605473e-05,
+      "loss": 0.8239,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316594672504362e-05,
+      "loss": 0.8027,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531079648278074e-05,
+      "loss": 0.8083,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304995805150834e-05,
+      "loss": 0.7594,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529919264233205e-05,
+      "loss": 0.8278,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293386997042943e-05,
+      "loss": 0.8226,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528757887200324e-05,
+      "loss": 0.8136,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528176826993382e-05,
+      "loss": 0.8278,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527595519355674e-05,
+      "loss": 0.8226,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52701396455952e-05,
+      "loss": 0.9066,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526432162877356e-05,
+      "loss": 0.8262,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258501145817344e-05,
+      "loss": 0.8226,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252678199453224e-05,
+      "loss": 0.8146,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246852792409033e-05,
+      "loss": 0.8047,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524102492741375e-05,
+      "loss": 0.8646,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235194607197508e-05,
+      "loss": 0.8034,
+      "step": 1632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229361834491591e-05,
+      "loss": 0.8844,
+      "step": 1633
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223526612028432e-05,
+      "loss": 0.8613,
+      "step": 1634
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217688942541611e-05,
+      "loss": 0.7733,
+      "step": 1635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211848828765852e-05,
+      "loss": 0.8154,
+      "step": 1636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206006273437031e-05,
+      "loss": 0.8356,
+      "step": 1637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200161279292154e-05,
+      "loss": 0.8343,
+      "step": 1638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194313849069389e-05,
+      "loss": 0.8353,
+      "step": 1639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518846398550803e-05,
+      "loss": 0.7793,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518261169134852e-05,
+      "loss": 0.7461,
+      "step": 1641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176756969332428e-05,
+      "loss": 0.7458,
+      "step": 1642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170899822202473e-05,
+      "loss": 0.8197,
+      "step": 1643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165040252702502e-05,
+      "loss": 0.8099,
+      "step": 1644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159178263577512e-05,
+      "loss": 0.8239,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153313857573615e-05,
+      "loss": 0.7897,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147447037438055e-05,
+      "loss": 0.8073,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141577805919225e-05,
+      "loss": 0.7708,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135706165766634e-05,
+      "loss": 0.8421,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129832119730919e-05,
+      "loss": 0.7939,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123955670563853e-05,
+      "loss": 0.821,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118076821018322e-05,
+      "loss": 0.8135,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112195573848347e-05,
+      "loss": 0.8353,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510631193180907e-05,
+      "loss": 0.7998,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100425897656754e-05,
+      "loss": 0.8467,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509453747414878e-05,
+      "loss": 0.7975,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088646664043652e-05,
+      "loss": 0.793,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082753470100987e-05,
+      "loss": 0.8428,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076857895081527e-05,
+      "loss": 0.8545,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070959941747126e-05,
+      "loss": 0.7786,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065059612860742e-05,
+      "loss": 0.8275,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059156911186465e-05,
+      "loss": 0.9023,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053251839489482e-05,
+      "loss": 0.8193,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047344400536095e-05,
+      "loss": 0.7946,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041434597093715e-05,
+      "loss": 0.8493,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035522431930859e-05,
+      "loss": 0.8516,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502960790781715e-05,
+      "loss": 0.8447,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023691027523324e-05,
+      "loss": 0.8115,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501777179382121e-05,
+      "loss": 0.8968,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011850209483741e-05,
+      "loss": 0.8242,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005926277284959e-05,
+      "loss": 0.7731,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9049,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994071380405102e-05,
+      "loss": 0.7669,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988140421277595e-05,
+      "loss": 0.8184,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982207125395907e-05,
+      "loss": 0.7782,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497627149553956e-05,
+      "loss": 0.7943,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970333534489179e-05,
+      "loss": 0.8581,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964393245026467e-05,
+      "loss": 0.7731,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958450629934223e-05,
+      "loss": 0.7812,
+      "step": 1679
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952505691996338e-05,
+      "loss": 0.779,
+      "step": 1680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946558433997792e-05,
+      "loss": 0.8249,
+      "step": 1681
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494060885872464e-05,
+      "loss": 0.7832,
+      "step": 1682
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934656968964047e-05,
+      "loss": 0.8171,
+      "step": 1683
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928702767504235e-05,
+      "loss": 0.8548,
+      "step": 1684
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922746257134523e-05,
+      "loss": 0.8382,
+      "step": 1685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916787440645315e-05,
+      "loss": 0.8232,
+      "step": 1686
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910826320828085e-05,
+      "loss": 0.7689,
+      "step": 1687
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904862900475391e-05,
+      "loss": 0.7956,
+      "step": 1688
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898897182380872e-05,
+      "loss": 0.8587,
+      "step": 1689
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892929169339237e-05,
+      "loss": 0.8226,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886958864146277e-05,
+      "loss": 0.8118,
+      "step": 1691
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488098626959885e-05,
+      "loss": 0.8268,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487501138849489e-05,
+      "loss": 0.8415,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869034223633399e-05,
+      "loss": 0.8278,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486305477781446e-05,
+      "loss": 0.8438,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857073053839206e-05,
+      "loss": 0.8076,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851089054509852e-05,
+      "loss": 0.8639,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845102782629676e-05,
+      "loss": 0.8854,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839114241003017e-05,
+      "loss": 0.8109,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483312343243528e-05,
+      "loss": 0.8464,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827130359732929e-05,
+      "loss": 0.8014,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821135025703491e-05,
+      "loss": 0.8105,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815137433155554e-05,
+      "loss": 0.8438,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480913758489876e-05,
+      "loss": 0.8031,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480313548374381e-05,
+      "loss": 0.8717,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797131132502464e-05,
+      "loss": 0.8151,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791124533987529e-05,
+      "loss": 0.8005,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785115691012866e-05,
+      "loss": 0.8438,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779104606393396e-05,
+      "loss": 0.777,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773091282945076e-05,
+      "loss": 0.8063,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767075723484928e-05,
+      "loss": 0.8626,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761057930831002e-05,
+      "loss": 0.7756,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755037907802414e-05,
+      "loss": 0.7643,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749015657219315e-05,
+      "loss": 0.8418,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47429911819029e-05,
+      "loss": 0.8118,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.47369644846754e-05,
+      "loss": 0.8236,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730935568360103e-05,
+      "loss": 0.8421,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724904435781322e-05,
+      "loss": 0.8175,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718871089764413e-05,
+      "loss": 0.8018,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712835533135774e-05,
+      "loss": 0.8304,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706797768722826e-05,
+      "loss": 0.835,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470075779935404e-05,
+      "loss": 0.86,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469471562785891e-05,
+      "loss": 0.8493,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688671257067962e-05,
+      "loss": 0.8496,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682624689812753e-05,
+      "loss": 0.835,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676575928925869e-05,
+      "loss": 0.8389,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670524977240929e-05,
+      "loss": 0.8568,
+      "step": 1727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664471837592573e-05,
+      "loss": 0.7712,
+      "step": 1728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658416512816468e-05,
+      "loss": 0.835,
+      "step": 1729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652359005749295e-05,
+      "loss": 0.7946,
+      "step": 1730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646299319228772e-05,
+      "loss": 0.7952,
+      "step": 1731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640237456093636e-05,
+      "loss": 0.7782,
+      "step": 1732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634173419183633e-05,
+      "loss": 0.875,
+      "step": 1733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628107211339537e-05,
+      "loss": 0.8405,
+      "step": 1734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622038835403135e-05,
+      "loss": 0.8356,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4615968294217226e-05,
+      "loss": 0.8014,
+      "step": 1736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4609895590625635e-05,
+      "loss": 0.8158,
+      "step": 1737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603820727473192e-05,
+      "loss": 0.8451,
+      "step": 1738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597743707605734e-05,
+      "loss": 0.707,
+      "step": 1739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591664533870118e-05,
+      "loss": 0.8027,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585583209114205e-05,
+      "loss": 0.8268,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579499736186864e-05,
+      "loss": 0.8128,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573414117937972e-05,
+      "loss": 0.8285,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567326357218408e-05,
+      "loss": 0.8037,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561236456880057e-05,
+      "loss": 0.7975,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555144419775808e-05,
+      "loss": 0.8418,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549050248759546e-05,
+      "loss": 0.8454,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542953946686161e-05,
+      "loss": 0.7451,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536855516411534e-05,
+      "loss": 0.8281,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530754960792554e-05,
+      "loss": 0.8001,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524652282687096e-05,
+      "loss": 0.79,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518547484954033e-05,
+      "loss": 0.8317,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451244057045323e-05,
+      "loss": 0.8665,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4506331542045545e-05,
+      "loss": 0.8424,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500220402592829e-05,
+      "loss": 0.7656,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449410715495791e-05,
+      "loss": 0.7832,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487991802004625e-05,
+      "loss": 0.7822,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448187434659777e-05,
+      "loss": 0.791,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475754791603152e-05,
+      "loss": 0.8369,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469633139887542e-05,
+      "loss": 0.7767,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4463509394318707e-05,
+      "loss": 0.8647,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 0.8219,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451255633097296e-05,
+      "loss": 0.8561,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445125623185148e-05,
+      "loss": 0.8356,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443899353090061e-05,
+      "loss": 0.8691,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432859359116337e-05,
+      "loss": 0.8607,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442672311070595e-05,
+      "loss": 0.82,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420584788544059e-05,
+      "loss": 0.8092,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414444395506226e-05,
+      "loss": 0.8385,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408301934468996e-05,
+      "loss": 0.8327,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402157408309876e-05,
+      "loss": 0.8268,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439601081990734e-05,
+      "loss": 0.818,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389862172140841e-05,
+      "loss": 0.8711,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 0.7676,
+      "step": 1774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377558710038522e-05,
+      "loss": 0.7653,
+      "step": 1775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437140390146641e-05,
+      "loss": 0.8148,
+      "step": 1776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365247045057732e-05,
+      "loss": 0.7744,
+      "step": 1777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435908814369675e-05,
+      "loss": 0.8753,
+      "step": 1778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352927200268667e-05,
+      "loss": 0.7896,
+      "step": 1779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346764217659652e-05,
+      "loss": 0.8236,
+      "step": 1780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340599198756828e-05,
+      "loss": 0.8597,
+      "step": 1781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334432146448272e-05,
+      "loss": 0.8587,
+      "step": 1782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328263063623018e-05,
+      "loss": 0.8454,
+      "step": 1783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322091953171043e-05,
+      "loss": 0.7948,
+      "step": 1784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315918817983274e-05,
+      "loss": 0.8353,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309743660951597e-05,
+      "loss": 0.8249,
+      "step": 1786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303566484968836e-05,
+      "loss": 0.8717,
+      "step": 1787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429738729292876e-05,
+      "loss": 0.8255,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429120608772609e-05,
+      "loss": 0.7586,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285022872256484e-05,
+      "loss": 0.9062,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278837649416543e-05,
+      "loss": 0.8096,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427265042210381e-05,
+      "loss": 0.8356,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266461193216765e-05,
+      "loss": 0.8285,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260269965654824e-05,
+      "loss": 0.8952,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254076742318347e-05,
+      "loss": 0.8473,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247881526108622e-05,
+      "loss": 0.7829,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241684319927869e-05,
+      "loss": 0.8577,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235485126679244e-05,
+      "loss": 0.859,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229283949266837e-05,
+      "loss": 0.8376,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4223080790595662e-05,
+      "loss": 0.8229,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216875653571665e-05,
+      "loss": 0.7939,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210668541101713e-05,
+      "loss": 0.8031,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204459456093602e-05,
+      "loss": 0.8685,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198248401456057e-05,
+      "loss": 0.8236,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192035380098716e-05,
+      "loss": 0.7018,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185820394932148e-05,
+      "loss": 0.8607,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179603448867836e-05,
+      "loss": 0.8672,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417338454481818e-05,
+      "loss": 0.7829,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167163685696504e-05,
+      "loss": 0.7666,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160940874417041e-05,
+      "loss": 0.8548,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154716113894942e-05,
+      "loss": 0.8148,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148489407046274e-05,
+      "loss": 0.9215,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142260756788006e-05,
+      "loss": 0.7469,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136030166038031e-05,
+      "loss": 0.8001,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129797637715141e-05,
+      "loss": 0.79,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123563174739036e-05,
+      "loss": 0.8239,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411732678003033e-05,
+      "loss": 0.7669,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111088456510531e-05,
+      "loss": 0.8014,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104848207102062e-05,
+      "loss": 0.8044,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409860603472824e-05,
+      "loss": 0.8971,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092361942313284e-05,
+      "loss": 0.7894,
+      "step": 1821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086115932782316e-05,
+      "loss": 0.8685,
+      "step": 1822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079868009061355e-05,
+      "loss": 0.7832,
+      "step": 1823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073618174077315e-05,
+      "loss": 0.8206,
+      "step": 1824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.7373,
+      "step": 1825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061112782032126e-05,
+      "loss": 0.818,
+      "step": 1826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054857230829284e-05,
+      "loss": 0.8721,
+      "step": 1827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048599780079957e-05,
+      "loss": 0.8278,
+      "step": 1828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042340432715528e-05,
+      "loss": 0.7926,
+      "step": 1829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036079191668266e-05,
+      "loss": 0.8226,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029816059871317e-05,
+      "loss": 0.8281,
+      "step": 1831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023551040258726e-05,
+      "loss": 0.8451,
+      "step": 1832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017284135765414e-05,
+      "loss": 0.8372,
+      "step": 1833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011015349327188e-05,
+      "loss": 0.8206,
+      "step": 1834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004744683880738e-05,
+      "loss": 0.8096,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998472142363628e-05,
+      "loss": 0.7988,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399219772771431e-05,
+      "loss": 0.8363,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985921442872105e-05,
+      "loss": 0.8197,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397964329077722e-05,
+      "loss": 0.8245,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973363274370722e-05,
+      "loss": 0.8158,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967081396594563e-05,
+      "loss": 0.8447,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.7853,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954512068705425e-05,
+      "loss": 0.9183,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948224624480697e-05,
+      "loss": 0.8477,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394193533066281e-05,
+      "loss": 0.7917,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935644190198061e-05,
+      "loss": 0.8245,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929351206033607e-05,
+      "loss": 0.8044,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923056381117479e-05,
+      "loss": 0.8392,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916759718398555e-05,
+      "loss": 0.8424,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910461220826585e-05,
+      "loss": 0.7839,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3904160891352176e-05,
+      "loss": 0.8011,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897858732926794e-05,
+      "loss": 0.8369,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891554748502763e-05,
+      "loss": 0.8089,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885248941033252e-05,
+      "loss": 0.8454,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878941313472298e-05,
+      "loss": 0.8581,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872631868774789e-05,
+      "loss": 0.8428,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866320609896449e-05,
+      "loss": 0.8122,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860007539793872e-05,
+      "loss": 0.7516,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853692661424485e-05,
+      "loss": 0.847,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384737597774657e-05,
+      "loss": 0.8223,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841057491719261e-05,
+      "loss": 0.8158,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834737206302519e-05,
+      "loss": 0.8333,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382841512445716e-05,
+      "loss": 0.8379,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382209124914484e-05,
+      "loss": 0.8037,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815765583328053e-05,
+      "loss": 0.7863,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809438129970136e-05,
+      "loss": 0.8115,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803108892035259e-05,
+      "loss": 0.8268,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796777872488427e-05,
+      "loss": 0.8255,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790445074295481e-05,
+      "loss": 0.7686,
+      "step": 1869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784110500423104e-05,
+      "loss": 0.8242,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777774153838799e-05,
+      "loss": 0.8076,
+      "step": 1871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771436037510897e-05,
+      "loss": 0.8314,
+      "step": 1872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376509615440857e-05,
+      "loss": 0.8044,
+      "step": 1873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758754507501817e-05,
+      "loss": 0.8743,
+      "step": 1874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375241109976145e-05,
+      "loss": 0.8503,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 0.7555,
+      "step": 1876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739719013667297e-05,
+      "loss": 0.8752,
+      "step": 1877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733370341259265e-05,
+      "loss": 0.6862,
+      "step": 1878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372701991990914e-05,
+      "loss": 0.8389,
+      "step": 1879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720667752591857e-05,
+      "loss": 0.8848,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714313842283158e-05,
+      "loss": 0.821,
+      "step": 1881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707958191959609e-05,
+      "loss": 0.8464,
+      "step": 1882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701600804598593e-05,
+      "loss": 0.8372,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369524168317831e-05,
+      "loss": 0.7995,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688880830677758e-05,
+      "loss": 0.7835,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682518250076755e-05,
+      "loss": 0.7487,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367615394435593e-05,
+      "loss": 0.8037,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669787916496722e-05,
+      "loss": 0.8079,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663420169481369e-05,
+      "loss": 0.7865,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365705070629292e-05,
+      "loss": 0.9105,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650679529915225e-05,
+      "loss": 0.8148,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644306643332939e-05,
+      "loss": 0.8034,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637932049531517e-05,
+      "loss": 0.8216,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631555751497214e-05,
+      "loss": 0.8239,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625177752217082e-05,
+      "loss": 0.835,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618798054678972e-05,
+      "loss": 0.8604,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612416661871532e-05,
+      "loss": 0.8411,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606033576784197e-05,
+      "loss": 0.8379,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599648802407205e-05,
+      "loss": 0.8128,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593262341731577e-05,
+      "loss": 0.8301,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358687419774913e-05,
+      "loss": 0.828,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580484373452462e-05,
+      "loss": 0.877,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574092871834966e-05,
+      "loss": 0.7806,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567699695890817e-05,
+      "loss": 0.7599,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561304848614974e-05,
+      "loss": 0.8197,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355490833300318e-05,
+      "loss": 0.8158,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548510152051963e-05,
+      "loss": 0.8014,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542110308758625e-05,
+      "loss": 0.8154,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535708806121246e-05,
+      "loss": 0.8451,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529305647138689e-05,
+      "loss": 0.8216,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522900834810594e-05,
+      "loss": 0.7764,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516494372137368e-05,
+      "loss": 0.7985,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510086262120198e-05,
+      "loss": 0.8291,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503676507761036e-05,
+      "loss": 0.8454,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497265112062613e-05,
+      "loss": 0.8464,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3490852078028423e-05,
+      "loss": 0.8929,
+      "step": 1916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484437408662725e-05,
+      "loss": 0.8538,
+      "step": 1917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478021106970552e-05,
+      "loss": 0.8125,
+      "step": 1918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471603175957693e-05,
+      "loss": 0.8805,
+      "step": 1919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346518361863071e-05,
+      "loss": 0.8135,
+      "step": 1920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345876243799692e-05,
+      "loss": 0.859,
+      "step": 1921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34523396370644e-05,
+      "loss": 0.75,
+      "step": 1922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445915218841989e-05,
+      "loss": 0.8623,
+      "step": 1923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439489186339283e-05,
+      "loss": 0.8483,
+      "step": 1924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433061542566634e-05,
+      "loss": 0.7327,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426632290535149e-05,
+      "loss": 0.8275,
+      "step": 1926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.8122,
+      "step": 1927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413768973743863e-05,
+      "loss": 0.8073,
+      "step": 1928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407334915010037e-05,
+      "loss": 0.8359,
+      "step": 1929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400899260069325e-05,
+      "loss": 0.8203,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339446201193658e-05,
+      "loss": 0.7708,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388023173627413e-05,
+      "loss": 0.8298,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381582748158173e-05,
+      "loss": 0.8945,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375140738545954e-05,
+      "loss": 0.8236,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368697147808596e-05,
+      "loss": 0.8216,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362251978964675e-05,
+      "loss": 0.8509,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355805235033503e-05,
+      "loss": 0.8145,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334935691903514e-05,
+      "loss": 0.7777,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342907033990375e-05,
+      "loss": 0.8574,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336455582920738e-05,
+      "loss": 0.8434,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330002568848479e-05,
+      "loss": 0.7664,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323547994796597e-05,
+      "loss": 0.8158,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317091863788816e-05,
+      "loss": 0.8148,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310634178849583e-05,
+      "loss": 0.8359,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330417494300408e-05,
+      "loss": 0.7791,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297714159278213e-05,
+      "loss": 0.7562,
+      "step": 1946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291251830698615e-05,
+      "loss": 0.807,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328478796029264e-05,
+      "loss": 0.8454,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278322551088365e-05,
+      "loss": 0.8148,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271855606114593e-05,
+      "loss": 0.8398,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265387128400833e-05,
+      "loss": 0.8607,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258917120977327e-05,
+      "loss": 0.8945,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252445586875031e-05,
+      "loss": 0.8078,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245972529125609e-05,
+      "loss": 0.8154,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239497950761438e-05,
+      "loss": 0.7215,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233021854815617e-05,
+      "loss": 0.8089,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322654424432195e-05,
+      "loss": 0.7946,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322006512231495e-05,
+      "loss": 0.8363,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213584491829843e-05,
+      "loss": 0.8232,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207102355902553e-05,
+      "loss": 0.7894,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200618717569716e-05,
+      "loss": 0.8392,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194133579868672e-05,
+      "loss": 0.7664,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187646945837462e-05,
+      "loss": 0.8086,
+      "step": 1963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181158818514824e-05,
+      "loss": 0.7773,
+      "step": 1964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31746692009402e-05,
+      "loss": 0.792,
+      "step": 1965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168178096153732e-05,
+      "loss": 0.8346,
+      "step": 1966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161685507196251e-05,
+      "loss": 0.8978,
+      "step": 1967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155191437109294e-05,
+      "loss": 0.7907,
+      "step": 1968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314869588893508e-05,
+      "loss": 0.8066,
+      "step": 1969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314219886571653e-05,
+      "loss": 0.8258,
+      "step": 1970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135700370497253e-05,
+      "loss": 0.8317,
+      "step": 1971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129200406321545e-05,
+      "loss": 0.8597,
+      "step": 1972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122698976234393e-05,
+      "loss": 0.8112,
+      "step": 1973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311619608328147e-05,
+      "loss": 0.8239,
+      "step": 1974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109691730509131e-05,
+      "loss": 0.8268,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103185920964425e-05,
+      "loss": 0.8516,
+      "step": 1976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096678657695072e-05,
+      "loss": 0.8652,
+      "step": 1977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.8245,
+      "step": 1978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083659782176723e-05,
+      "loss": 0.8467,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077148176026579e-05,
+      "loss": 0.8402,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070635128349482e-05,
+      "loss": 0.8285,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3064120642196549e-05,
+      "loss": 0.7835,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057604720619566e-05,
+      "loss": 0.7786,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051087366670994e-05,
+      "loss": 0.7959,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304456858340397e-05,
+      "loss": 0.8154,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038048373872297e-05,
+      "loss": 0.8057,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031526741130435e-05,
+      "loss": 0.7801,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025003688233533e-05,
+      "loss": 0.7578,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018479218237388e-05,
+      "loss": 0.807,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011953334198465e-05,
+      "loss": 0.8363,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005426039173899e-05,
+      "loss": 0.7614,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299889733622147e-05,
+      "loss": 0.8962,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992367228399633e-05,
+      "loss": 0.8411,
+      "step": 1993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985835718767496e-05,
+      "loss": 0.7822,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297930281038482e-05,
+      "loss": 0.7965,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972768506312028e-05,
+      "loss": 0.7995,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966232809610189e-05,
+      "loss": 0.8369,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959695723341028e-05,
+      "loss": 0.806,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295315725056693e-05,
+      "loss": 0.8574,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294661739435091e-05,
+      "loss": 0.8525,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940076157756647e-05,
+      "loss": 0.8415,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933533543848462e-05,
+      "loss": 0.8363,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292698955569132e-05,
+      "loss": 0.7484,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2920444196350829e-05,
+      "loss": 0.848,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913897468893249e-05,
+      "loss": 0.8216,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907349376385461e-05,
+      "loss": 0.8311,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900799921895004e-05,
+      "loss": 0.7646,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894249108490052e-05,
+      "loss": 0.7933,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2887696939239407e-05,
+      "loss": 0.7612,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288114341721251e-05,
+      "loss": 0.8265,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2874588545479439e-05,
+      "loss": 0.835,
+      "step": 2011
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.8392,
+      "step": 2012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286147476517824e-05,
+      "loss": 0.7926,
+      "step": 2013
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854915862753424e-05,
+      "loss": 0.8125,
+      "step": 2014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2848355622909038e-05,
+      "loss": 0.7591,
+      "step": 2015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284179404871832e-05,
+      "loss": 0.8242,
+      "step": 2016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283523114325511e-05,
+      "loss": 0.7611,
+      "step": 2017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828666909593883e-05,
+      "loss": 0.8307,
+      "step": 2018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822101350809728e-05,
+      "loss": 0.8018,
+      "step": 2019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2815534469978363e-05,
+      "loss": 0.8337,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808966270176121e-05,
+      "loss": 0.8193,
+      "step": 2021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802396754479958e-05,
+      "loss": 0.7959,
+      "step": 2022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279582592596744e-05,
+      "loss": 0.8135,
+      "step": 2023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789253787716747e-05,
+      "loss": 0.8421,
+      "step": 2024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782680342806679e-05,
+      "loss": 0.8441,
+      "step": 2025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776105594316647e-05,
+      "loss": 0.8519,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2769529545326669e-05,
+      "loss": 0.8574,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762952198917371e-05,
+      "loss": 0.7777,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.8255,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749793626166377e-05,
+      "loss": 0.8512,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743212405988975e-05,
+      "loss": 0.7656,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 0.8092,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730046113445602e-05,
+      "loss": 0.7992,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723461047247545e-05,
+      "loss": 0.8311,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2716874705211511e-05,
+      "loss": 0.8822,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710287090422948e-05,
+      "loss": 0.7952,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703698205967907e-05,
+      "loss": 0.763,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697108054933027e-05,
+      "loss": 0.8229,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690516640405548e-05,
+      "loss": 0.8014,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683923965473293e-05,
+      "loss": 0.8187,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677330033224681e-05,
+      "loss": 0.8376,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670734846748717e-05,
+      "loss": 0.8258,
+      "step": 2042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664138409134998e-05,
+      "loss": 0.833,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26575407234737e-05,
+      "loss": 0.8044,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650941792855593e-05,
+      "loss": 0.8438,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644341620372025e-05,
+      "loss": 0.8311,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637740209114918e-05,
+      "loss": 0.806,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631137562176792e-05,
+      "loss": 0.8213,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624533682650733e-05,
+      "loss": 0.7962,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617928573630405e-05,
+      "loss": 0.7881,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611322238210053e-05,
+      "loss": 0.7689,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260471467948449e-05,
+      "loss": 0.8255,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598105900549111e-05,
+      "loss": 0.8395,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591495904499878e-05,
+      "loss": 0.8802,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584884694433321e-05,
+      "loss": 0.7926,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578272273446536e-05,
+      "loss": 0.846,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25716586446372e-05,
+      "loss": 0.8063,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565043811103541e-05,
+      "loss": 0.7658,
+      "step": 2058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558427775944357e-05,
+      "loss": 0.8281,
+      "step": 2059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255181054225901e-05,
+      "loss": 0.7806,
+      "step": 2060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545192113147425e-05,
+      "loss": 0.7812,
+      "step": 2061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538572491710079e-05,
+      "loss": 0.8626,
+      "step": 2062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253195168104802e-05,
+      "loss": 0.8073,
+      "step": 2063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525329684262838e-05,
+      "loss": 0.7572,
+      "step": 2064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518706504456693e-05,
+      "loss": 0.7868,
+      "step": 2065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512082144732292e-05,
+      "loss": 0.7773,
+      "step": 2066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505456608192889e-05,
+      "loss": 0.8089,
+      "step": 2067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498829897942308e-05,
+      "loss": 0.8538,
+      "step": 2068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492202017084904e-05,
+      "loss": 0.8086,
+      "step": 2069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485572968725588e-05,
+      "loss": 0.7871,
+      "step": 2070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478942755969814e-05,
+      "loss": 0.7617,
+      "step": 2071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247231138192359e-05,
+      "loss": 0.7767,
+      "step": 2072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465678849693459e-05,
+      "loss": 0.8122,
+      "step": 2073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459045162386514e-05,
+      "loss": 0.7249,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245241032311038e-05,
+      "loss": 0.7484,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445774334973229e-05,
+      "loss": 0.7965,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2439137201083772e-05,
+      "loss": 0.8291,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432498924551256e-05,
+      "loss": 0.7975,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425859508485449e-05,
+      "loss": 0.8825,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 0.8302,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412577270195784e-05,
+      "loss": 0.7552,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405934454194146e-05,
+      "loss": 0.7262,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239929051110367e-05,
+      "loss": 0.8232,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239264544403679e-05,
+      "loss": 0.7819,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385999256106466e-05,
+      "loss": 0.8008,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379351950426188e-05,
+      "loss": 0.8643,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372703530109967e-05,
+      "loss": 0.862,
+      "step": 2087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236605399827233e-05,
+      "loss": 0.7702,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235940335802833e-05,
+      "loss": 0.7814,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352751612493548e-05,
+      "loss": 0.8044,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346098764784062e-05,
+      "loss": 0.8073,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339444818016488e-05,
+      "loss": 0.8125,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233278977530794e-05,
+      "loss": 0.8314,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2326133639776054e-05,
+      "loss": 0.8265,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2319476414538982e-05,
+      "loss": 0.8034,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312818102715375e-05,
+      "loss": 0.7812,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306158707424402e-05,
+      "loss": 0.772,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299498231785739e-05,
+      "loss": 0.8001,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292836678919556e-05,
+      "loss": 0.7985,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286174051946547e-05,
+      "loss": 0.7754,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279510353987895e-05,
+      "loss": 0.7591,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227284558816529e-05,
+      "loss": 0.7529,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226617975760092e-05,
+      "loss": 0.8135,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259512865417478e-05,
+      "loss": 0.8232,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252844914738145e-05,
+      "loss": 0.807,
+      "step": 2105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246175908686607e-05,
+      "loss": 0.8105,
+      "step": 2106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239505850387032e-05,
+      "loss": 0.804,
+      "step": 2107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232834742964093e-05,
+      "loss": 0.8304,
+      "step": 2108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226162589542954e-05,
+      "loss": 0.8234,
+      "step": 2109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219489393249263e-05,
+      "loss": 0.904,
+      "step": 2110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212815157209154e-05,
+      "loss": 0.7616,
+      "step": 2111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220613988454926e-05,
+      "loss": 0.7842,
+      "step": 2112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199463578396688e-05,
+      "loss": 0.8236,
+      "step": 2113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192786241879033e-05,
+      "loss": 0.7962,
+      "step": 2114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218610787812438e-05,
+      "loss": 0.7832,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179428490261278e-05,
+      "loss": 0.835,
+      "step": 2116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2172748081418775e-05,
+      "loss": 0.7754,
+      "step": 2117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166066654726387e-05,
+      "loss": 0.8643,
+      "step": 2118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159384213314105e-05,
+      "loss": 0.7199,
+      "step": 2119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152700760312401e-05,
+      "loss": 0.8454,
+      "step": 2120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146016298852222e-05,
+      "loss": 0.7935,
+      "step": 2121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139330832064975e-05,
+      "loss": 0.8548,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132644363082555e-05,
+      "loss": 0.8216,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125956895037318e-05,
+      "loss": 0.8294,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119268431062085e-05,
+      "loss": 0.7365,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211257897429015e-05,
+      "loss": 0.8034,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210588852785527e-05,
+      "loss": 0.7959,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099197094891659e-05,
+      "loss": 0.7653,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092504678534007e-05,
+      "loss": 0.8018,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085811281917453e-05,
+      "loss": 0.7917,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.791,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072421560450497e-05,
+      "loss": 0.8571,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065725241872676e-05,
+      "loss": 0.8099,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20590279555811e-05,
+      "loss": 0.7871,
+      "step": 2134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052329704713195e-05,
+      "loss": 0.8135,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045630492406835e-05,
+      "loss": 0.8447,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038930321800346e-05,
+      "loss": 0.8421,
+      "step": 2137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032229196032508e-05,
+      "loss": 0.7949,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025527118242534e-05,
+      "loss": 0.7881,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018824091570103e-05,
+      "loss": 0.8444,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012120119155324e-05,
+      "loss": 0.8197,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005415204138753e-05,
+      "loss": 0.765,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1998709349661388e-05,
+      "loss": 0.793,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992002558864668e-05,
+      "loss": 0.7803,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985294834890465e-05,
+      "loss": 0.846,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19785861808811e-05,
+      "loss": 0.798,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197187659997932e-05,
+      "loss": 0.8105,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965166095328302e-05,
+      "loss": 0.8197,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958454670071671e-05,
+      "loss": 0.8428,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951742327353471e-05,
+      "loss": 0.797,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1945029070318178e-05,
+      "loss": 0.8714,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938314902110701e-05,
+      "loss": 0.848,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1931599825876372e-05,
+      "loss": 0.8636,
+      "step": 2153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924883844760944e-05,
+      "loss": 0.8307,
+      "step": 2154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918166961910607e-05,
+      "loss": 0.7923,
+      "step": 2155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911449180471962e-05,
+      "loss": 0.8158,
+      "step": 2156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190473050359203e-05,
+      "loss": 0.7321,
+      "step": 2157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898010934418261e-05,
+      "loss": 0.846,
+      "step": 2158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891290476098514e-05,
+      "loss": 0.7731,
+      "step": 2159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884569131781069e-05,
+      "loss": 0.8555,
+      "step": 2160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877846904614626e-05,
+      "loss": 0.8372,
+      "step": 2161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871123797748285e-05,
+      "loss": 0.7959,
+      "step": 2162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864399814331568e-05,
+      "loss": 0.7637,
+      "step": 2163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857674957514411e-05,
+      "loss": 0.8063,
+      "step": 2164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1850949230447146e-05,
+      "loss": 0.8327,
+      "step": 2165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184422263628053e-05,
+      "loss": 0.8379,
+      "step": 2166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837495178165706e-05,
+      "loss": 0.8063,
+      "step": 2167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830766859254234e-05,
+      "loss": 0.7673,
+      "step": 2168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824037682698086e-05,
+      "loss": 0.7861,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817307651649616e-05,
+      "loss": 0.8291,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181057676926159e-05,
+      "loss": 0.8249,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803845038687171e-05,
+      "loss": 0.8242,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797112463079918e-05,
+      "loss": 0.8154,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790379045593787e-05,
+      "loss": 0.7907,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783644789383132e-05,
+      "loss": 0.8062,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177690969760269e-05,
+      "loss": 0.8737,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770173773407594e-05,
+      "loss": 0.8333,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763437019953377e-05,
+      "loss": 0.8255,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756699440395947e-05,
+      "loss": 0.8089,
+      "step": 2179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749961037891605e-05,
+      "loss": 0.8281,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743221815597034e-05,
+      "loss": 0.8392,
+      "step": 2181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.8079,
+      "step": 2182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172974092426587e-05,
+      "loss": 0.875,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722999261544565e-05,
+      "loss": 0.7962,
+      "step": 2184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716256791663599e-05,
+      "loss": 0.7624,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709513517781563e-05,
+      "loss": 0.8708,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702769443057425e-05,
+      "loss": 0.821,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696024570650528e-05,
+      "loss": 0.7923,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689278903720592e-05,
+      "loss": 0.8307,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16825324454277e-05,
+      "loss": 0.8044,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1675785198932308e-05,
+      "loss": 0.7952,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669037167395256e-05,
+      "loss": 0.8096,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662288353977732e-05,
+      "loss": 0.735,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16555387618413e-05,
+      "loss": 0.8079,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648788394147891e-05,
+      "loss": 0.7891,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164203725405979e-05,
+      "loss": 0.8034,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163528534473965e-05,
+      "loss": 0.7793,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162853266935049e-05,
+      "loss": 0.7734,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621779231055677e-05,
+      "loss": 0.8128,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1615025033018937e-05,
+      "loss": 0.7891,
+      "step": 2200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608270078404363e-05,
+      "loss": 0.8135,
+      "step": 2201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160151437037639e-05,
+      "loss": 0.7253,
+      "step": 2202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159475791209981e-05,
+      "loss": 0.7757,
+      "step": 2203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588000706739768e-05,
+      "loss": 0.8024,
+      "step": 2204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581242757461758e-05,
+      "loss": 0.7962,
+      "step": 2205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574484067431617e-05,
+      "loss": 0.7383,
+      "step": 2206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567724639815546e-05,
+      "loss": 0.7419,
+      "step": 2207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156096447778007e-05,
+      "loss": 0.7539,
+      "step": 2208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554203584492072e-05,
+      "loss": 0.8076,
+      "step": 2209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547441963118771e-05,
+      "loss": 0.8021,
+      "step": 2210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154067961682773e-05,
+      "loss": 0.7982,
+      "step": 2211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533916548786856e-05,
+      "loss": 0.807,
+      "step": 2212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527152762164384e-05,
+      "loss": 0.7738,
+      "step": 2213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152038826012889e-05,
+      "loss": 0.745,
+      "step": 2214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513623045849291e-05,
+      "loss": 0.7923,
+      "step": 2215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506857122494832e-05,
+      "loss": 0.8187,
+      "step": 2216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500090493235088e-05,
+      "loss": 0.7321,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149332316123997e-05,
+      "loss": 0.8151,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486555129679712e-05,
+      "loss": 0.79,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479786401724884e-05,
+      "loss": 0.8421,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473016980546377e-05,
+      "loss": 0.861,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466246869315407e-05,
+      "loss": 0.7646,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459476071203511e-05,
+      "loss": 0.7604,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145270458938255e-05,
+      "loss": 0.835,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445932427024716e-05,
+      "loss": 0.8441,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439159587302495e-05,
+      "loss": 0.8099,
+      "step": 2226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432386073388718e-05,
+      "loss": 0.79,
+      "step": 2227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425611888456509e-05,
+      "loss": 0.7378,
+      "step": 2228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141883703567932e-05,
+      "loss": 0.7894,
+      "step": 2229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412061518230916e-05,
+      "loss": 0.7819,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405285339285364e-05,
+      "loss": 0.8184,
+      "step": 2231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398508502017047e-05,
+      "loss": 0.7998,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391731009600655e-05,
+      "loss": 0.8438,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384952865211188e-05,
+      "loss": 0.7633,
+      "step": 2234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378174072023951e-05,
+      "loss": 0.8337,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371394633214548e-05,
+      "loss": 0.835,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136461455195888e-05,
+      "loss": 0.8783,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357833831433176e-05,
+      "loss": 0.7816,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135105247481393e-05,
+      "loss": 0.7979,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344270485277958e-05,
+      "loss": 0.8024,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133748786600236e-05,
+      "loss": 0.8294,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133070462016454e-05,
+      "loss": 0.7751,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323920750942184e-05,
+      "loss": 0.7725,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317136261513281e-05,
+      "loss": 0.7884,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310351155056107e-05,
+      "loss": 0.8122,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303565434749225e-05,
+      "loss": 0.8115,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129677910377149e-05,
+      "loss": 0.8379,
+      "step": 2247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289992165302036e-05,
+      "loss": 0.7988,
+      "step": 2248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283204622520285e-05,
+      "loss": 0.8105,
+      "step": 2249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127641647860595e-05,
+      "loss": 0.8088,
+      "step": 2250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269627736739009e-05,
+      "loss": 0.7249,
+      "step": 2251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1262838400099733e-05,
+      "loss": 0.7643,
+      "step": 2252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256048471868675e-05,
+      "loss": 0.8415,
+      "step": 2253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249257955226649e-05,
+      "loss": 0.8333,
+      "step": 2254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242466853354759e-05,
+      "loss": 0.8337,
+      "step": 2255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235675169434372e-05,
+      "loss": 0.8192,
+      "step": 2256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228882906647142e-05,
+      "loss": 0.8148,
+      "step": 2257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222090068174983e-05,
+      "loss": 0.8271,
+      "step": 2258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121529665720008e-05,
+      "loss": 0.8519,
+      "step": 2259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208502676904887e-05,
+      "loss": 0.749,
+      "step": 2260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1201708130472127e-05,
+      "loss": 0.807,
+      "step": 2261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119491302108479e-05,
+      "loss": 0.7721,
+      "step": 2262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118811735192612e-05,
+      "loss": 0.8415,
+      "step": 2263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181321126179637e-05,
+      "loss": 0.86,
+      "step": 2264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174524347029106e-05,
+      "loss": 0.765,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167727017658562e-05,
+      "loss": 0.7856,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160929141252303e-05,
+      "loss": 0.8197,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115413072099487e-05,
+      "loss": 0.7871,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147331760071064e-05,
+      "loss": 0.7604,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140532261665937e-05,
+      "loss": 0.8493,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133732228964802e-05,
+      "loss": 0.8135,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1126931665153213e-05,
+      "loss": 0.827,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120130573416976e-05,
+      "loss": 0.7314,
+      "step": 2273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113328956942139e-05,
+      "loss": 0.8151,
+      "step": 2274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1106526818915008e-05,
+      "loss": 0.818,
+      "step": 2275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099724162522125e-05,
+      "loss": 0.8262,
+      "step": 2276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092920990950276e-05,
+      "loss": 0.771,
+      "step": 2277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086117307386481e-05,
+      "loss": 0.8311,
+      "step": 2278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1079313115018016e-05,
+      "loss": 0.7347,
+      "step": 2279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1072508417032382e-05,
+      "loss": 0.7865,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065703216617322e-05,
+      "loss": 0.7376,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058897516960817e-05,
+      "loss": 0.807,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052091321251073e-05,
+      "loss": 0.8643,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.8063,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038477454425885e-05,
+      "loss": 0.8408,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031669789688017e-05,
+      "loss": 0.8105,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102486164165207e-05,
+      "loss": 0.8584,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10180530135074e-05,
+      "loss": 0.8356,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011243908443591e-05,
+      "loss": 0.7793,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004434329650453e-05,
+      "loss": 0.8242,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099762428031801e-05,
+      "loss": 0.8363,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990813763636511e-05,
+      "loss": 0.7819,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984002782796427e-05,
+      "loss": 0.8695,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977191340988447e-05,
+      "loss": 0.7943,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0970379441403463e-05,
+      "loss": 0.8271,
+      "step": 2295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09635670872326e-05,
+      "loss": 0.7842,
+      "step": 2296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956754281667182e-05,
+      "loss": 0.8171,
+      "step": 2297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094994102789875e-05,
+      "loss": 0.7497,
+      "step": 2298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943127329119057e-05,
+      "loss": 0.738,
+      "step": 2299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093631318852006e-05,
+      "loss": 0.7998,
+      "step": 2300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929498609293925e-05,
+      "loss": 0.8379,
+      "step": 2301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 0.7581,
+      "step": 2302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091586814772993e-05,
+      "loss": 0.806,
+      "step": 2303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090905227177742e-05,
+      "loss": 0.7744,
+      "step": 2304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090223596996848e-05,
+      "loss": 0.8333,
+      "step": 2305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895419245496282e-05,
+      "loss": 0.8678,
+      "step": 2306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888602101554202e-05,
+      "loss": 0.7744,
+      "step": 2307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881784541335818e-05,
+      "loss": 0.7839,
+      "step": 2308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0874966568034893e-05,
+      "loss": 0.8102,
+      "step": 2309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868148184845391e-05,
+      "loss": 0.7624,
+      "step": 2310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0861329394961461e-05,
+      "loss": 0.7917,
+      "step": 2311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854510201577451e-05,
+      "loss": 0.8206,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847690607887894e-05,
+      "loss": 0.8226,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840870617087513e-05,
+      "loss": 0.8171,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834050232371205e-05,
+      "loss": 0.7596,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827229456934068e-05,
+      "loss": 0.7926,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082040829397138e-05,
+      "loss": 0.7389,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813586746678584e-05,
+      "loss": 0.821,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806764818251328e-05,
+      "loss": 0.8789,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799942511885417e-05,
+      "loss": 0.8132,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793119830776849e-05,
+      "loss": 0.7832,
+      "step": 2321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786296778121787e-05,
+      "loss": 0.7874,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779473357116577e-05,
+      "loss": 0.7461,
+      "step": 2323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772649570957723e-05,
+      "loss": 0.7559,
+      "step": 2324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765825422841916e-05,
+      "loss": 0.7614,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759000915966011e-05,
+      "loss": 0.8382,
+      "step": 2326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752176053527025e-05,
+      "loss": 0.7692,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0745350838722153e-05,
+      "loss": 0.8153,
+      "step": 2328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073852527474874e-05,
+      "loss": 0.7731,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731699364804312e-05,
+      "loss": 0.8743,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724873112086544e-05,
+      "loss": 0.8447,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718046519793276e-05,
+      "loss": 0.8574,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071121959112251e-05,
+      "loss": 0.75,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704392329272396e-05,
+      "loss": 0.8203,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697564737441254e-05,
+      "loss": 0.8086,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0690736818827547e-05,
+      "loss": 0.8125,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06839085766299e-05,
+      "loss": 0.7998,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677080014047076e-05,
+      "loss": 0.8548,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670251134278005e-05,
+      "loss": 0.8291,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663421940521753e-05,
+      "loss": 0.7933,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656592435977542e-05,
+      "loss": 0.8447,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649762623844733e-05,
+      "loss": 0.8086,
+      "step": 2342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642932507322832e-05,
+      "loss": 0.8405,
+      "step": 2343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063610208961149e-05,
+      "loss": 0.8558,
+      "step": 2344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06292713739105e-05,
+      "loss": 0.8115,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062244036341979e-05,
+      "loss": 0.7826,
+      "step": 2346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615609061339431e-05,
+      "loss": 0.7808,
+      "step": 2347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608777470869625e-05,
+      "loss": 0.8001,
+      "step": 2348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601945595210717e-05,
+      "loss": 0.847,
+      "step": 2349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0595113437563175e-05,
+      "loss": 0.7839,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588281001127612e-05,
+      "loss": 0.8538,
+      "step": 2351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.7897,
+      "step": 2352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574615304695486e-05,
+      "loss": 0.7975,
+      "step": 2353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567782051100786e-05,
+      "loss": 0.7806,
+      "step": 2354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560948531521776e-05,
+      "loss": 0.7923,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05541147491597e-05,
+      "loss": 0.7793,
+      "step": 2356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054728070721593e-05,
+      "loss": 0.8275,
+      "step": 2357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540446408891949e-05,
+      "loss": 0.7949,
+      "step": 2358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053361185738937e-05,
+      "loss": 0.7982,
+      "step": 2359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052677705590992e-05,
+      "loss": 0.7487,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519942007655444e-05,
+      "loss": 0.8245,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0513106715827897e-05,
+      "loss": 0.7588,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506271183629362e-05,
+      "loss": 0.7399,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499435414262018e-05,
+      "loss": 0.7357,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049259941092817e-05,
+      "loss": 0.8288,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485763176830221e-05,
+      "loss": 0.8079,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478926715170687e-05,
+      "loss": 0.8203,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472090029152196e-05,
+      "loss": 0.7764,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046525312197747e-05,
+      "loss": 0.7513,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458415996849337e-05,
+      "loss": 0.8314,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451578656970738e-05,
+      "loss": 0.8112,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444741105544705e-05,
+      "loss": 0.7536,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437903345774367e-05,
+      "loss": 0.8506,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431065380862959e-05,
+      "loss": 0.8408,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424227214013802e-05,
+      "loss": 0.8372,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417388848430323e-05,
+      "loss": 0.7764,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410550287316035e-05,
+      "loss": 0.7568,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403711533874541e-05,
+      "loss": 0.7751,
+      "step": 2378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396872591309542e-05,
+      "loss": 0.8164,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390033462824817e-05,
+      "loss": 0.8325,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038319415162424e-05,
+      "loss": 0.7848,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376354660911772e-05,
+      "loss": 0.7783,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369514993891451e-05,
+      "loss": 0.7308,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362675153767398e-05,
+      "loss": 0.7799,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355835143743825e-05,
+      "loss": 0.8249,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 0.7754,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342154626815321e-05,
+      "loss": 0.7832,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335314126319196e-05,
+      "loss": 0.734,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328473468741147e-05,
+      "loss": 0.8656,
+      "step": 2389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321632657285763e-05,
+      "loss": 0.8096,
+      "step": 2390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0314791695157707e-05,
+      "loss": 0.7446,
+      "step": 2391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0307950585561705e-05,
+      "loss": 0.7533,
+      "step": 2392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301109331702561e-05,
+      "loss": 0.7572,
+      "step": 2393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029426793678514e-05,
+      "loss": 0.7985,
+      "step": 2394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287426404014377e-05,
+      "loss": 0.8861,
+      "step": 2395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028058473659527e-05,
+      "loss": 0.8454,
+      "step": 2396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273742937732877e-05,
+      "loss": 0.777,
+      "step": 2397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266901010632324e-05,
+      "loss": 0.793,
+      "step": 2398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260058958498794e-05,
+      "loss": 0.8304,
+      "step": 2399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253216784537527e-05,
+      "loss": 0.7617,
+      "step": 2400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0246374491953823e-05,
+      "loss": 0.7757,
+      "step": 2401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0239532083953032e-05,
+      "loss": 0.7804,
+      "step": 2402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232689563740563e-05,
+      "loss": 0.7638,
+      "step": 2403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225846934521881e-05,
+      "loss": 0.806,
+      "step": 2404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219004199502496e-05,
+      "loss": 0.8044,
+      "step": 2405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021216136188797e-05,
+      "loss": 0.7728,
+      "step": 2406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205318424883906e-05,
+      "loss": 0.7192,
+      "step": 2407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198475391695966e-05,
+      "loss": 0.7607,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019163226552985e-05,
+      "loss": 0.765,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01847890495913e-05,
+      "loss": 0.8281,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177945747086106e-05,
+      "loss": 0.7793,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171102361220093e-05,
+      "loss": 0.8542,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164258895199129e-05,
+      "loss": 0.8522,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157415352229115e-05,
+      "loss": 0.7764,
+      "step": 2414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150571735515992e-05,
+      "loss": 0.7096,
+      "step": 2415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143728048265735e-05,
+      "loss": 0.7524,
+      "step": 2416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013688429368435e-05,
+      "loss": 0.8096,
+      "step": 2417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130040474977878e-05,
+      "loss": 0.791,
+      "step": 2418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123196595352385e-05,
+      "loss": 0.8086,
+      "step": 2419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116352658013973e-05,
+      "loss": 0.8317,
+      "step": 2420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109508666168759e-05,
+      "loss": 0.8174,
+      "step": 2421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01026646230229e-05,
+      "loss": 0.8066,
+      "step": 2422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095820531782567e-05,
+      "loss": 0.7829,
+      "step": 2423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008897639565396e-05,
+      "loss": 0.8405,
+      "step": 2424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008213221784329e-05,
+      "loss": 0.8353,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075288001556798e-05,
+      "loss": 0.776,
+      "step": 2426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006844375000074e-05,
+      "loss": 0.8141,
+      "step": 2427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0061599466381388e-05,
+      "loss": 0.7848,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054755153905027e-05,
+      "loss": 0.833,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047910815777956e-05,
+      "loss": 0.8506,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004106645520649e-05,
+      "loss": 0.7611,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034222075396954e-05,
+      "loss": 0.8027,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002737767955567e-05,
+      "loss": 0.7606,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002053327088899e-05,
+      "loss": 0.7669,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001368885260325e-05,
+      "loss": 0.8073,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00068444279048e-05,
+      "loss": 0.7731,
+      "step": 2436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7627,
+      "step": 2437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993155572095199e-06,
+      "loss": 0.7283,
+      "step": 2438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986311147396755e-06,
+      "loss": 0.8734,
+      "step": 2439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979466729111014e-06,
+      "loss": 0.791,
+      "step": 2440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97262232044433e-06,
+      "loss": 0.8245,
+      "step": 2441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965777924603053e-06,
+      "loss": 0.779,
+      "step": 2442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958933544793513e-06,
+      "loss": 0.7965,
+      "step": 2443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.952089184222045e-06,
+      "loss": 0.8584,
+      "step": 2444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945244846094978e-06,
+      "loss": 0.7464,
+      "step": 2445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938400533618615e-06,
+      "loss": 0.804,
+      "step": 2446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931556249999262e-06,
+      "loss": 0.8249,
+      "step": 2447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924711998443205e-06,
+      "loss": 0.8337,
+      "step": 2448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917867782156712e-06,
+      "loss": 0.804,
+      "step": 2449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911023604346043e-06,
+      "loss": 0.8402,
+      "step": 2450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904179468217438e-06,
+      "loss": 0.901,
+      "step": 2451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897335376977104e-06,
+      "loss": 0.7578,
+      "step": 2452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890491333831243e-06,
+      "loss": 0.7399,
+      "step": 2453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883647341986032e-06,
+      "loss": 0.8171,
+      "step": 2454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876803404647617e-06,
+      "loss": 0.8057,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869959525022124e-06,
+      "loss": 0.7793,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863115706315652e-06,
+      "loss": 0.8031,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856271951734268e-06,
+      "loss": 0.8294,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84942826448401e-06,
+      "loss": 0.8451,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84258464777089e-06,
+      "loss": 0.8014,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835741104800876e-06,
+      "loss": 0.7842,
+      "step": 2461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.828897638779909e-06,
+      "loss": 0.8454,
+      "step": 2462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822054252913897e-06,
+      "loss": 0.8278,
+      "step": 2463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815210950408703e-06,
+      "loss": 0.8001,
+      "step": 2464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808367734470153e-06,
+      "loss": 0.7585,
+      "step": 2465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.801524608304036e-06,
+      "loss": 0.8359,
+      "step": 2466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794681575116097e-06,
+      "loss": 0.8138,
+      "step": 2467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787838638112034e-06,
+      "loss": 0.8105,
+      "step": 2468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780995800497504e-06,
+      "loss": 0.7646,
+      "step": 2469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774153065478122e-06,
+      "loss": 0.8379,
+      "step": 2470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767310436259438e-06,
+      "loss": 0.8063,
+      "step": 2471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760467916046971e-06,
+      "loss": 0.7972,
+      "step": 2472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753625508046183e-06,
+      "loss": 0.7699,
+      "step": 2473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746783215462476e-06,
+      "loss": 0.8167,
+      "step": 2474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.739941041501208e-06,
+      "loss": 0.779,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733098989367677e-06,
+      "loss": 0.8148,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726257062267124e-06,
+      "loss": 0.8141,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719415263404732e-06,
+      "loss": 0.8063,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712573595985628e-06,
+      "loss": 0.82,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705732063214864e-06,
+      "loss": 0.805,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69889066829744e-06,
+      "loss": 0.8226,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692049414438298e-06,
+      "loss": 0.8359,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.685208304842298e-06,
+      "loss": 0.7069,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678367342714238e-06,
+      "loss": 0.8512,
+      "step": 2484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671526531258856e-06,
+      "loss": 0.7796,
+      "step": 2485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664685873680807e-06,
+      "loss": 0.8275,
+      "step": 2486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65784537318468e-06,
+      "loss": 0.8193,
+      "step": 2487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 0.7972,
+      "step": 2488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644164856256179e-06,
+      "loss": 0.7847,
+      "step": 2489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637324846232603e-06,
+      "loss": 0.7957,
+      "step": 2490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630485006108554e-06,
+      "loss": 0.7673,
+      "step": 2491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62364533908823e-06,
+      "loss": 0.7775,
+      "step": 2492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61680584837576e-06,
+      "loss": 0.8392,
+      "step": 2493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609966537175186e-06,
+      "loss": 0.8138,
+      "step": 2494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603127408690461e-06,
+      "loss": 0.7975,
+      "step": 2495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59628846612546e-06,
+      "loss": 0.8102,
+      "step": 2496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58944971268397e-06,
+      "loss": 0.8558,
+      "step": 2497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58261115156968e-06,
+      "loss": 0.7923,
+      "step": 2498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575772785986201e-06,
+      "loss": 0.7526,
+      "step": 2499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568934619137048e-06,
+      "loss": 0.7458,
+      "step": 2500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562096654225636e-06,
+      "loss": 0.7663,
+      "step": 2501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555258894455298e-06,
+      "loss": 0.7998,
+      "step": 2502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548421343029264e-06,
+      "loss": 0.8024,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.541584003150664e-06,
+      "loss": 0.8177,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534746878022533e-06,
+      "loss": 0.8089,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527909970847809e-06,
+      "loss": 0.8148,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521073284829315e-06,
+      "loss": 0.7549,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51423682316978e-06,
+      "loss": 0.7926,
+      "step": 2508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507400589071833e-06,
+      "loss": 0.849,
+      "step": 2509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500564585737984e-06,
+      "loss": 0.7521,
+      "step": 2510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493728816370641e-06,
+      "loss": 0.8102,
+      "step": 2511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.486893284172103e-06,
+      "loss": 0.7638,
+      "step": 2512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48005799234456e-06,
+      "loss": 0.7972,
+      "step": 2513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473222944090082e-06,
+      "loss": 0.8493,
+      "step": 2514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46638814261063e-06,
+      "loss": 0.7477,
+      "step": 2515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459553591108054e-06,
+      "loss": 0.7918,
+      "step": 2516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452719292784074e-06,
+      "loss": 0.8066,
+      "step": 2517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445885250840301e-06,
+      "loss": 0.8581,
+      "step": 2518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43905146847823e-06,
+      "loss": 0.7858,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432217948899217e-06,
+      "loss": 0.8151,
+      "step": 2520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425384695304514e-06,
+      "loss": 0.7855,
+      "step": 2521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.7897,
+      "step": 2522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41171899887239e-06,
+      "loss": 0.8311,
+      "step": 2523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404886562436826e-06,
+      "loss": 0.7565,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398054404789288e-06,
+      "loss": 0.8519,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391222529130379e-06,
+      "loss": 0.7965,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.384390938660572e-06,
+      "loss": 0.848,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377559636580213e-06,
+      "loss": 0.7666,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370728626089502e-06,
+      "loss": 0.7728,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363897910388512e-06,
+      "loss": 0.832,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357067492677171e-06,
+      "loss": 0.8271,
+      "step": 2531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350237376155269e-06,
+      "loss": 0.7406,
+      "step": 2532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343407564022458e-06,
+      "loss": 0.7767,
+      "step": 2533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33657805947825e-06,
+      "loss": 0.7464,
+      "step": 2534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329748865721999e-06,
+      "loss": 0.8359,
+      "step": 2535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322919985952926e-06,
+      "loss": 0.7523,
+      "step": 2536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316091423370105e-06,
+      "loss": 0.7962,
+      "step": 2537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.309263181172456e-06,
+      "loss": 0.7992,
+      "step": 2538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302435262558748e-06,
+      "loss": 0.7969,
+      "step": 2539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295607670727605e-06,
+      "loss": 0.82,
+      "step": 2540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288780408877493e-06,
+      "loss": 0.7852,
+      "step": 2541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281953480206725e-06,
+      "loss": 0.7331,
+      "step": 2542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27512688791346e-06,
+      "loss": 0.7646,
+      "step": 2543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268300635195691e-06,
+      "loss": 0.7949,
+      "step": 2544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261474725251261e-06,
+      "loss": 0.8343,
+      "step": 2545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.254649161277853e-06,
+      "loss": 0.7992,
+      "step": 2546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247823946472978e-06,
+      "loss": 0.7578,
+      "step": 2547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240999084033992e-06,
+      "loss": 0.8232,
+      "step": 2548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234174577158089e-06,
+      "loss": 0.7845,
+      "step": 2549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227350429042278e-06,
+      "loss": 0.7493,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220526642883427e-06,
+      "loss": 0.7467,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213703221878217e-06,
+      "loss": 0.8034,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206880169223154e-06,
+      "loss": 0.7835,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200057488114585e-06,
+      "loss": 0.7539,
+      "step": 2554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193235181748677e-06,
+      "loss": 0.8325,
+      "step": 2555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18641325332142e-06,
+      "loss": 0.7988,
+      "step": 2556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179591706028626e-06,
+      "loss": 0.8164,
+      "step": 2557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172770543065932e-06,
+      "loss": 0.7773,
+      "step": 2558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165949767628796e-06,
+      "loss": 0.7816,
+      "step": 2559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15912938291249e-06,
+      "loss": 0.8154,
+      "step": 2560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152309392112105e-06,
+      "loss": 0.7869,
+      "step": 2561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14548979842255e-06,
+      "loss": 0.7858,
+      "step": 2562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.138670605038542e-06,
+      "loss": 0.7839,
+      "step": 2563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131851815154614e-06,
+      "loss": 0.8102,
+      "step": 2564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12503343196511e-06,
+      "loss": 0.834,
+      "step": 2565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118215458664185e-06,
+      "loss": 0.7985,
+      "step": 2566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111397898445798e-06,
+      "loss": 0.8177,
+      "step": 2567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104580754503721e-06,
+      "loss": 0.8096,
+      "step": 2568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097764030031522e-06,
+      "loss": 0.7445,
+      "step": 2569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09094772822258e-06,
+      "loss": 0.8262,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084131852270076e-06,
+      "loss": 0.7741,
+      "step": 2571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 0.7783,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07050139070608e-06,
+      "loss": 0.8047,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063686811479945e-06,
+      "loss": 0.7689,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056872670880946e-06,
+      "loss": 0.8161,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050058972101252e-06,
+      "loss": 0.7917,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043245718332821e-06,
+      "loss": 0.7936,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036432912767403e-06,
+      "loss": 0.7917,
+      "step": 2578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029620558596537e-06,
+      "loss": 0.8864,
+      "step": 2579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02280865901156e-06,
+      "loss": 0.8376,
+      "step": 2580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015997217203574e-06,
+      "loss": 0.8337,
+      "step": 2581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00918623636349e-06,
+      "loss": 0.8382,
+      "step": 2582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002375719681995e-06,
+      "loss": 0.7279,
+      "step": 2583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99556567034955e-06,
+      "loss": 0.8346,
+      "step": 2584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988756091556409e-06,
+      "loss": 0.776,
+      "step": 2585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981946986492604e-06,
+      "loss": 0.7917,
+      "step": 2586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975138358347931e-06,
+      "loss": 0.7695,
+      "step": 2587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968330210311985e-06,
+      "loss": 0.7474,
+      "step": 2588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961522545574122e-06,
+      "loss": 0.8132,
+      "step": 2589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.8385,
+      "step": 2590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94790867874893e-06,
+      "loss": 0.8643,
+      "step": 2591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941102483039188e-06,
+      "loss": 0.7891,
+      "step": 2592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934296783382681e-06,
+      "loss": 0.7923,
+      "step": 2593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92749158296762e-06,
+      "loss": 0.7891,
+      "step": 2594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920686884981991e-06,
+      "loss": 0.7767,
+      "step": 2595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913882692613522e-06,
+      "loss": 0.8011,
+      "step": 2596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907079009049728e-06,
+      "loss": 0.7985,
+      "step": 2597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900275837477879e-06,
+      "loss": 0.7275,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893473181084993e-06,
+      "loss": 0.8561,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886671043057863e-06,
+      "loss": 0.7858,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879869426583029e-06,
+      "loss": 0.7712,
+      "step": 2601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87306833484679e-06,
+      "loss": 0.8187,
+      "step": 2602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8662677710352e-06,
+      "loss": 0.7806,
+      "step": 2603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859467738334068e-06,
+      "loss": 0.8382,
+      "step": 2604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85266823992894e-06,
+      "loss": 0.8203,
+      "step": 2605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845869279005133e-06,
+      "loss": 0.7972,
+      "step": 2606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839070858747697e-06,
+      "loss": 0.7816,
+      "step": 2607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83227298234144e-06,
+      "loss": 0.7874,
+      "step": 2608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825475652970898e-06,
+      "loss": 0.8086,
+      "step": 2609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818678873820368e-06,
+      "loss": 0.7067,
+      "step": 2610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811882648073883e-06,
+      "loss": 0.8324,
+      "step": 2611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805086978915215e-06,
+      "loss": 0.7982,
+      "step": 2612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.798291869527873e-06,
+      "loss": 0.8148,
+      "step": 2613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791497323095116e-06,
+      "loss": 0.8128,
+      "step": 2614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784703342799923e-06,
+      "loss": 0.7904,
+      "step": 2615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777909931825019e-06,
+      "loss": 0.7663,
+      "step": 2616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771117093352861e-06,
+      "loss": 0.8311,
+      "step": 2617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76432483056563e-06,
+      "loss": 0.8317,
+      "step": 2618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757533146645244e-06,
+      "loss": 0.8587,
+      "step": 2619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750742044773354e-06,
+      "loss": 0.7656,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74395152813133e-06,
+      "loss": 0.7959,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737161599900267e-06,
+      "loss": 0.7712,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730372263260996e-06,
+      "loss": 0.833,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723583521394054e-06,
+      "loss": 0.821,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.716795377479715e-06,
+      "loss": 0.8122,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71000783469797e-06,
+      "loss": 0.7205,
+      "step": 2626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703220896228515e-06,
+      "loss": 0.8161,
+      "step": 2627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696434565250776e-06,
+      "loss": 0.777,
+      "step": 2628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689648844943897e-06,
+      "loss": 0.7526,
+      "step": 2629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682863738486722e-06,
+      "loss": 0.7793,
+      "step": 2630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676079249057818e-06,
+      "loss": 0.834,
+      "step": 2631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669295379835467e-06,
+      "loss": 0.7679,
+      "step": 2632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66251213399764e-06,
+      "loss": 0.7533,
+      "step": 2633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.655729514722044e-06,
+      "loss": 0.7936,
+      "step": 2634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648947525186073e-06,
+      "loss": 0.8031,
+      "step": 2635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.642166168566828e-06,
+      "loss": 0.8643,
+      "step": 2636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63538544804112e-06,
+      "loss": 0.8122,
+      "step": 2637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628605366785459e-06,
+      "loss": 0.7949,
+      "step": 2638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621825927976052e-06,
+      "loss": 0.8252,
+      "step": 2639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615047134788812e-06,
+      "loss": 0.8001,
+      "step": 2640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60826899039935e-06,
+      "loss": 0.686,
+      "step": 2641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601491497982956e-06,
+      "loss": 0.7656,
+      "step": 2642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594714660714638e-06,
+      "loss": 0.8096,
+      "step": 2643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58793848176909e-06,
+      "loss": 0.7484,
+      "step": 2644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581162964320682e-06,
+      "loss": 0.7887,
+      "step": 2645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574388111543493e-06,
+      "loss": 0.8193,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567613926611287e-06,
+      "loss": 0.8418,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560840412697507e-06,
+      "loss": 0.7653,
+      "step": 2648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55406757297529e-06,
+      "loss": 0.8031,
+      "step": 2649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547295410617453e-06,
+      "loss": 0.8473,
+      "step": 2650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540523928796494e-06,
+      "loss": 0.762,
+      "step": 2651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533753130684596e-06,
+      "loss": 0.7816,
+      "step": 2652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526983019453624e-06,
+      "loss": 0.8024,
+      "step": 2653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52021359827512e-06,
+      "loss": 0.8027,
+      "step": 2654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51344487032029e-06,
+      "loss": 0.7493,
+      "step": 2655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506676838760033e-06,
+      "loss": 0.8008,
+      "step": 2656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499909506764914e-06,
+      "loss": 0.7985,
+      "step": 2657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49314287750517e-06,
+      "loss": 0.7666,
+      "step": 2658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486376954150707e-06,
+      "loss": 0.7031,
+      "step": 2659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479611739871114e-06,
+      "loss": 0.7368,
+      "step": 2660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472847237835618e-06,
+      "loss": 0.847,
+      "step": 2661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466083451213145e-06,
+      "loss": 0.7803,
+      "step": 2662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459320383172274e-06,
+      "loss": 0.7751,
+      "step": 2663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452558036881234e-06,
+      "loss": 0.7913,
+      "step": 2664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445796415507932e-06,
+      "loss": 0.7871,
+      "step": 2665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439035522219935e-06,
+      "loss": 0.7962,
+      "step": 2666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432275360184458e-06,
+      "loss": 0.8337,
+      "step": 2667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425515932568383e-06,
+      "loss": 0.8438,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418757242538249e-06,
+      "loss": 0.736,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411999293260234e-06,
+      "loss": 0.8125,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405242087900192e-06,
+      "loss": 0.7686,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398485629623613e-06,
+      "loss": 0.7624,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391729921595639e-06,
+      "loss": 0.7643,
+      "step": 2673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384974966981063e-06,
+      "loss": 0.7858,
+      "step": 2674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378220768944328e-06,
+      "loss": 0.7643,
+      "step": 2675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371467330649512e-06,
+      "loss": 0.7684,
+      "step": 2676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36471465526035e-06,
+      "loss": 0.771,
+      "step": 2677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.357962745940215e-06,
+      "loss": 0.8353,
+      "step": 2678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351211605852112e-06,
+      "loss": 0.7839,
+      "step": 2679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3444612381587e-06,
+      "loss": 0.766,
+      "step": 2680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337711646022271e-06,
+      "loss": 0.8883,
+      "step": 2681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330962832604747e-06,
+      "loss": 0.8005,
+      "step": 2682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324214801067692e-06,
+      "loss": 0.7673,
+      "step": 2683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317467554572306e-06,
+      "loss": 0.8066,
+      "step": 2684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310721096279413e-06,
+      "loss": 0.8418,
+      "step": 2685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303975429349473e-06,
+      "loss": 0.8581,
+      "step": 2686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29723055694258e-06,
+      "loss": 0.8161,
+      "step": 2687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29048648221844e-06,
+      "loss": 0.777,
+      "step": 2688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283743208336403e-06,
+      "loss": 0.7962,
+      "step": 2689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27700073845544e-06,
+      "loss": 0.8031,
+      "step": 2690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270259075734133e-06,
+      "loss": 0.833,
+      "step": 2691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.8265,
+      "step": 2692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25677818440297e-06,
+      "loss": 0.8698,
+      "step": 2693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2500389621084e-06,
+      "loss": 0.8076,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243300559604054e-06,
+      "loss": 0.8206,
+      "step": 2695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236562980046627e-06,
+      "loss": 0.8132,
+      "step": 2696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22982622659241e-06,
+      "loss": 0.7845,
+      "step": 2697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223090302397313e-06,
+      "loss": 0.8385,
+      "step": 2698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216355210616874e-06,
+      "loss": 0.8001,
+      "step": 2699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209620954406216e-06,
+      "loss": 0.7489,
+      "step": 2700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202887536920084e-06,
+      "loss": 0.8083,
+      "step": 2701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19615496131283e-06,
+      "loss": 0.8252,
+      "step": 2702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189423230738413e-06,
+      "loss": 0.6955,
+      "step": 2703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182692348350386e-06,
+      "loss": 0.7902,
+      "step": 2704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175962317301914e-06,
+      "loss": 0.7676,
+      "step": 2705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169233140745768e-06,
+      "loss": 0.7617,
+      "step": 2706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162504821834296e-06,
+      "loss": 0.7845,
+      "step": 2707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155777363719472e-06,
+      "loss": 0.7891,
+      "step": 2708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.149050769552856e-06,
+      "loss": 0.762,
+      "step": 2709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142325042485592e-06,
+      "loss": 0.7705,
+      "step": 2710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135600185668433e-06,
+      "loss": 0.8125,
+      "step": 2711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128876202251719e-06,
+      "loss": 0.8337,
+      "step": 2712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122153095385378e-06,
+      "loss": 0.792,
+      "step": 2713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115430868218931e-06,
+      "loss": 0.7868,
+      "step": 2714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108709523901491e-06,
+      "loss": 0.791,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.101989065581742e-06,
+      "loss": 0.805,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095269496407972e-06,
+      "loss": 0.7861,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088550819528043e-06,
+      "loss": 0.7861,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081833038089394e-06,
+      "loss": 0.8193,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075116155239056e-06,
+      "loss": 0.818,
+      "step": 2720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068400174123632e-06,
+      "loss": 0.7972,
+      "step": 2721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0616850978893e-06,
+      "loss": 0.7487,
+      "step": 2722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.054970929681822e-06,
+      "loss": 0.8298,
+      "step": 2723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048257672646534e-06,
+      "loss": 0.7731,
+      "step": 2724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041545329928332e-06,
+      "loss": 0.7225,
+      "step": 2725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034833904671698e-06,
+      "loss": 0.8385,
+      "step": 2726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028123400020686e-06,
+      "loss": 0.7695,
+      "step": 2727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021413819118903e-06,
+      "loss": 0.776,
+      "step": 2728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014705165109535e-06,
+      "loss": 0.8066,
+      "step": 2729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007997441135335e-06,
+      "loss": 0.7933,
+      "step": 2730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001290650338614e-06,
+      "loss": 0.8311,
+      "step": 2731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994584795861248e-06,
+      "loss": 0.8271,
+      "step": 2732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98787988084468e-06,
+      "loss": 0.7394,
+      "step": 2733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9811759084299e-06,
+      "loss": 0.818,
+      "step": 2734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.974472881757466e-06,
+      "loss": 0.7839,
+      "step": 2735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967770803967498e-06,
+      "loss": 0.821,
+      "step": 2736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961069678199658e-06,
+      "loss": 0.8281,
+      "step": 2737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954369507593169e-06,
+      "loss": 0.7969,
+      "step": 2738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947670295286808e-06,
+      "loss": 0.7952,
+      "step": 2739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940972044418902e-06,
+      "loss": 0.7578,
+      "step": 2740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934274758127326e-06,
+      "loss": 0.8096,
+      "step": 2741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927578439549506e-06,
+      "loss": 0.8538,
+      "step": 2742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.7619,
+      "step": 2743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914188718082549e-06,
+      "loss": 0.7432,
+      "step": 2744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907495321465998e-06,
+      "loss": 0.8005,
+      "step": 2745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900802905108343e-06,
+      "loss": 0.7868,
+      "step": 2746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894111472144733e-06,
+      "loss": 0.8102,
+      "step": 2747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887421025709851e-06,
+      "loss": 0.7824,
+      "step": 2748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880731568937916e-06,
+      "loss": 0.8141,
+      "step": 2749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874043104962683e-06,
+      "loss": 0.7848,
+      "step": 2750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867355636917445e-06,
+      "loss": 0.847,
+      "step": 2751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860669167935028e-06,
+      "loss": 0.7731,
+      "step": 2752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853983701147782e-06,
+      "loss": 0.7637,
+      "step": 2753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847299239687597e-06,
+      "loss": 0.7451,
+      "step": 2754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840615786685899e-06,
+      "loss": 0.7607,
+      "step": 2755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833933345273616e-06,
+      "loss": 0.8258,
+      "step": 2756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827251918581225e-06,
+      "loss": 0.7812,
+      "step": 2757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820571509738724e-06,
+      "loss": 0.8008,
+      "step": 2758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813892121875623e-06,
+      "loss": 0.7331,
+      "step": 2759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807213758120965e-06,
+      "loss": 0.694,
+      "step": 2760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800536421603317e-06,
+      "loss": 0.8138,
+      "step": 2761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793860115450744e-06,
+      "loss": 0.7936,
+      "step": 2762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787184842790845e-06,
+      "loss": 0.8424,
+      "step": 2763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.780510606750742e-06,
+      "loss": 0.7887,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773837410457048e-06,
+      "loss": 0.7826,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767165257035907e-06,
+      "loss": 0.8343,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760494149612971e-06,
+      "loss": 0.7324,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753824091313398e-06,
+      "loss": 0.7861,
+      "step": 2768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747155085261857e-06,
+      "loss": 0.738,
+      "step": 2769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740487134582527e-06,
+      "loss": 0.8011,
+      "step": 2770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733820242399083e-06,
+      "loss": 0.8255,
+      "step": 2771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727154411834712e-06,
+      "loss": 0.8389,
+      "step": 2772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72048964601211e-06,
+      "loss": 0.738,
+      "step": 2773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713825948053457e-06,
+      "loss": 0.777,
+      "step": 2774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707163321080446e-06,
+      "loss": 0.7956,
+      "step": 2775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700501768214266e-06,
+      "loss": 0.7878,
+      "step": 2776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6938412925756e-06,
+      "loss": 0.7526,
+      "step": 2777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.687181897284625e-06,
+      "loss": 0.7308,
+      "step": 2778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.680523585461021e-06,
+      "loss": 0.8317,
+      "step": 2779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673866360223947e-06,
+      "loss": 0.7324,
+      "step": 2780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667210224692061e-06,
+      "loss": 0.8027,
+      "step": 2781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660555181983517e-06,
+      "loss": 0.8252,
+      "step": 2782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653901235215941e-06,
+      "loss": 0.7469,
+      "step": 2783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647248387506457e-06,
+      "loss": 0.7764,
+      "step": 2784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640596641971671e-06,
+      "loss": 0.7799,
+      "step": 2785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633946001727673e-06,
+      "loss": 0.8001,
+      "step": 2786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6272964698900356e-06,
+      "loss": 0.8154,
+      "step": 2787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620648049573815e-06,
+      "loss": 0.8392,
+      "step": 2788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614000743893537e-06,
+      "loss": 0.7471,
+      "step": 2789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607354555963211e-06,
+      "loss": 0.8112,
+      "step": 2790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.600709488896335e-06,
+      "loss": 0.7741,
+      "step": 2791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5940655458058575e-06,
+      "loss": 0.7725,
+      "step": 2792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587422729804217e-06,
+      "loss": 0.7249,
+      "step": 2793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 0.7692,
+      "step": 2794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574140491514552e-06,
+      "loss": 0.8213,
+      "step": 2795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5675010754487485e-06,
+      "loss": 0.8294,
+      "step": 2796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560862798916229e-06,
+      "loss": 0.7643,
+      "step": 2797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5542256650267746e-06,
+      "loss": 0.8122,
+      "step": 2798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547589676889624e-06,
+      "loss": 0.8285,
+      "step": 2799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.540954837613488e-06,
+      "loss": 0.7282,
+      "step": 2800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534321150306545e-06,
+      "loss": 0.763,
+      "step": 2801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527688618076413e-06,
+      "loss": 0.7077,
+      "step": 2802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521057244030188e-06,
+      "loss": 0.7171,
+      "step": 2803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514427031274416e-06,
+      "loss": 0.7645,
+      "step": 2804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507797982915097e-06,
+      "loss": 0.8031,
+      "step": 2805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501170102057691e-06,
+      "loss": 0.7946,
+      "step": 2806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494543391807112e-06,
+      "loss": 0.8128,
+      "step": 2807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487917855267712e-06,
+      "loss": 0.7757,
+      "step": 2808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481293495543307e-06,
+      "loss": 0.7913,
+      "step": 2809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474670315737165e-06,
+      "loss": 0.8161,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468048318951983e-06,
+      "loss": 0.7998,
+      "step": 2811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461427508289922e-06,
+      "loss": 0.8301,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454807886852578e-06,
+      "loss": 0.7526,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.448189457740991e-06,
+      "loss": 0.7272,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441572224055644e-06,
+      "loss": 0.8001,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434956188896464e-06,
+      "loss": 0.7835,
+      "step": 2816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428341355362803e-06,
+      "loss": 0.8278,
+      "step": 2817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421727726553463e-06,
+      "loss": 0.7585,
+      "step": 2818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415115305566685e-06,
+      "loss": 0.7495,
+      "step": 2819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408504095500124e-06,
+      "loss": 0.8229,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401894099450889e-06,
+      "loss": 0.8467,
+      "step": 2821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395285320515513e-06,
+      "loss": 0.7306,
+      "step": 2822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388677761789951e-06,
+      "loss": 0.7458,
+      "step": 2823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382071426369597e-06,
+      "loss": 0.8226,
+      "step": 2824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375466317349273e-06,
+      "loss": 0.8148,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368862437823211e-06,
+      "loss": 0.6706,
+      "step": 2826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36225979088508e-06,
+      "loss": 0.7806,
+      "step": 2827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355658379627981e-06,
+      "loss": 0.7272,
+      "step": 2828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349058207144408e-06,
+      "loss": 0.8044,
+      "step": 2829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342459276526301e-06,
+      "loss": 0.847,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3358615908650054e-06,
+      "loss": 0.7751,
+      "step": 2831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329265153251285e-06,
+      "loss": 0.7604,
+      "step": 2832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322669966775321e-06,
+      "loss": 0.7725,
+      "step": 2833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316076034526712e-06,
+      "loss": 0.7718,
+      "step": 2834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309483359594456e-06,
+      "loss": 0.8164,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302891945066975e-06,
+      "loss": 0.7314,
+      "step": 2836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296301794032097e-06,
+      "loss": 0.7443,
+      "step": 2837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289712909577056e-06,
+      "loss": 0.7686,
+      "step": 2838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.283125294788493e-06,
+      "loss": 0.7454,
+      "step": 2839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276538952752459e-06,
+      "loss": 0.7965,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2699538865544e-06,
+      "loss": 0.8366,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.7572,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256787594011032e-06,
+      "loss": 0.7702,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250206373833626e-06,
+      "loss": 0.8294,
+      "step": 2844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243626441830009e-06,
+      "loss": 0.7606,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2370478010826285e-06,
+      "loss": 0.7878,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230470454673335e-06,
+      "loss": 0.7591,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223894405683355e-06,
+      "loss": 0.7393,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21731965719332e-06,
+      "loss": 0.8477,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210746212283255e-06,
+      "loss": 0.8366,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204174074032562e-06,
+      "loss": 0.7812,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197603245520042e-06,
+      "loss": 0.8242,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191033729823882e-06,
+      "loss": 0.7562,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1844655300216405e-06,
+      "loss": 0.7926,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177898649190273e-06,
+      "loss": 0.7578,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171333090406123e-06,
+      "loss": 0.724,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164768856744893e-06,
+      "loss": 0.7747,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158205951281681e-06,
+      "loss": 0.7874,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.151644377090963e-06,
+      "loss": 0.8105,
+      "step": 2859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1450841372465806e-06,
+      "loss": 0.8236,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.138525234821759e-06,
+      "loss": 0.7682,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.7754,
+      "step": 2862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.125411454520563e-06,
+      "loss": 0.833,
+      "step": 2863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11885658278749e-06,
+      "loss": 0.7736,
+      "step": 2864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1123030607605985e-06,
+      "loss": 0.7799,
+      "step": 2865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105750891509951e-06,
+      "loss": 0.834,
+      "step": 2866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099200078104995e-06,
+      "loss": 0.805,
+      "step": 2867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.092650623614541e-06,
+      "loss": 0.7699,
+      "step": 2868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086102531106755e-06,
+      "loss": 0.7516,
+      "step": 2869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.079555803649169e-06,
+      "loss": 0.7865,
+      "step": 2870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0730104443086834e-06,
+      "loss": 0.755,
+      "step": 2871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066466456151541e-06,
+      "loss": 0.7747,
+      "step": 2872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059923842243355e-06,
+      "loss": 0.8376,
+      "step": 2873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053382605649094e-06,
+      "loss": 0.8424,
+      "step": 2874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0468427494330735e-06,
+      "loss": 0.7555,
+      "step": 2875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040304276658971e-06,
+      "loss": 0.7214,
+      "step": 2876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033767190389814e-06,
+      "loss": 0.8099,
+      "step": 2877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027231493687974e-06,
+      "loss": 0.8226,
+      "step": 2878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02069718961518e-06,
+      "loss": 0.7826,
+      "step": 2879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014164281232508e-06,
+      "loss": 0.7695,
+      "step": 2880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00763277160037e-06,
+      "loss": 0.7316,
+      "step": 2881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001102663778533e-06,
+      "loss": 0.7913,
+      "step": 2882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9945739608261065e-06,
+      "loss": 0.8099,
+      "step": 2883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988046665801536e-06,
+      "loss": 0.7923,
+      "step": 2884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981520781762614e-06,
+      "loss": 0.8242,
+      "step": 2885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97499631176647e-06,
+      "loss": 0.7323,
+      "step": 2886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968473258869566e-06,
+      "loss": 0.7122,
+      "step": 2887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961951626127707e-06,
+      "loss": 0.8245,
+      "step": 2888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955431416596034e-06,
+      "loss": 0.8267,
+      "step": 2889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948912633329008e-06,
+      "loss": 0.8187,
+      "step": 2890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942395279380438e-06,
+      "loss": 0.7904,
+      "step": 2891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.935879357803453e-06,
+      "loss": 0.7858,
+      "step": 2892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92936487165052e-06,
+      "loss": 0.8499,
+      "step": 2893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922851823973422e-06,
+      "loss": 0.721,
+      "step": 2894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916340217823278e-06,
+      "loss": 0.847,
+      "step": 2895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.7842,
+      "step": 2896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90332134230493e-06,
+      "loss": 0.8317,
+      "step": 2897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896814079035575e-06,
+      "loss": 0.7814,
+      "step": 2898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890308269490871e-06,
+      "loss": 0.82,
+      "step": 2899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883803916718534e-06,
+      "loss": 0.7563,
+      "step": 2900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87730102376561e-06,
+      "loss": 0.8128,
+      "step": 2901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870799593678459e-06,
+      "loss": 0.8115,
+      "step": 2902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8642996295027495e-06,
+      "loss": 0.797,
+      "step": 2903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8578011342834705e-06,
+      "loss": 0.8019,
+      "step": 2904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851304111064923e-06,
+      "loss": 0.7783,
+      "step": 2905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844808562890709e-06,
+      "loss": 0.7995,
+      "step": 2906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83831449280375e-06,
+      "loss": 0.806,
+      "step": 2907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831821903846274e-06,
+      "loss": 0.7829,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.825330799059804e-06,
+      "loss": 0.8626,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818841181485179e-06,
+      "loss": 0.8197,
+      "step": 2910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812353054162542e-06,
+      "loss": 0.7835,
+      "step": 2911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80586642013133e-06,
+      "loss": 0.8022,
+      "step": 2912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799381282430284e-06,
+      "loss": 0.7829,
+      "step": 2913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7928976440974504e-06,
+      "loss": 0.7832,
+      "step": 2914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78641550817016e-06,
+      "loss": 0.7554,
+      "step": 2915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77993487768505e-06,
+      "loss": 0.7718,
+      "step": 2916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.773455755678054e-06,
+      "loss": 0.7611,
+      "step": 2917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766978145184386e-06,
+      "loss": 0.7892,
+      "step": 2918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.760502049238564e-06,
+      "loss": 0.7868,
+      "step": 2919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7540274708743955e-06,
+      "loss": 0.7832,
+      "step": 2920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7475544131249705e-06,
+      "loss": 0.7617,
+      "step": 2921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741082879022671e-06,
+      "loss": 0.8141,
+      "step": 2922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734612871599169e-06,
+      "loss": 0.8118,
+      "step": 2923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728144393885411e-06,
+      "loss": 0.8073,
+      "step": 2924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721677448911635e-06,
+      "loss": 0.7905,
+      "step": 2925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715212039707364e-06,
+      "loss": 0.8073,
+      "step": 2926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708748169301389e-06,
+      "loss": 0.7822,
+      "step": 2927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7022858407217895e-06,
+      "loss": 0.8376,
+      "step": 2928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695825056995924e-06,
+      "loss": 0.7471,
+      "step": 2929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689365821150421e-06,
+      "loss": 0.8115,
+      "step": 2930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682908136211187e-06,
+      "loss": 0.7507,
+      "step": 2931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6764520052034054e-06,
+      "loss": 0.8346,
+      "step": 2932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6699974311515225e-06,
+      "loss": 0.7783,
+      "step": 2933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6635444170792665e-06,
+      "loss": 0.8291,
+      "step": 2934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6570929660096285e-06,
+      "loss": 0.7725,
+      "step": 2935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650643080964863e-06,
+      "loss": 0.7842,
+      "step": 2936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644194764966499e-06,
+      "loss": 0.7917,
+      "step": 2937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637748021035331e-06,
+      "loss": 0.7861,
+      "step": 2938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631302852191408e-06,
+      "loss": 0.7585,
+      "step": 2939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624859261454047e-06,
+      "loss": 0.7588,
+      "step": 2940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618417251841829e-06,
+      "loss": 0.805,
+      "step": 2941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.61197682637259e-06,
+      "loss": 0.7907,
+      "step": 2942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6055379880634214e-06,
+      "loss": 0.7904,
+      "step": 2943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599100739930677e-06,
+      "loss": 0.8118,
+      "step": 2944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592665084989965e-06,
+      "loss": 0.8203,
+      "step": 2945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586231026256139e-06,
+      "loss": 0.7336,
+      "step": 2946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.7539,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573367709464854e-06,
+      "loss": 0.7604,
+      "step": 2948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566938457433368e-06,
+      "loss": 0.7477,
+      "step": 2949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560510813660719e-06,
+      "loss": 0.7979,
+      "step": 2950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554084781158014e-06,
+      "loss": 0.7845,
+      "step": 2951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547660362935603e-06,
+      "loss": 0.8327,
+      "step": 2952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541237562003082e-06,
+      "loss": 0.7464,
+      "step": 2953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534816381369293e-06,
+      "loss": 0.848,
+      "step": 2954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5283968240423105e-06,
+      "loss": 0.8096,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521978893029452e-06,
+      "loss": 0.8057,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515562591337279e-06,
+      "loss": 0.7581,
+      "step": 2957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509147921971581e-06,
+      "loss": 0.7048,
+      "step": 2958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502734887937389e-06,
+      "loss": 0.721,
+      "step": 2959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496323492238966e-06,
+      "loss": 0.7686,
+      "step": 2960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489913737879805e-06,
+      "loss": 0.7653,
+      "step": 2961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483505627862632e-06,
+      "loss": 0.7567,
+      "step": 2962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47709916518941e-06,
+      "loss": 0.7754,
+      "step": 2963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4706943528613135e-06,
+      "loss": 0.7554,
+      "step": 2964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464291193878757e-06,
+      "loss": 0.8236,
+      "step": 2965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.45788969124138e-06,
+      "loss": 0.7842,
+      "step": 2966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451489847948039e-06,
+      "loss": 0.82,
+      "step": 2967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44509166699682e-06,
+      "loss": 0.8047,
+      "step": 2968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438695151385028e-06,
+      "loss": 0.8066,
+      "step": 2969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432300304109185e-06,
+      "loss": 0.7853,
+      "step": 2970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425907128165035e-06,
+      "loss": 0.7977,
+      "step": 2971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419515626547543e-06,
+      "loss": 0.8275,
+      "step": 2972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413125802250875e-06,
+      "loss": 0.7585,
+      "step": 2973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406737658268425e-06,
+      "loss": 0.7887,
+      "step": 2974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400351197592798e-06,
+      "loss": 0.7812,
+      "step": 2975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393966423215805e-06,
+      "loss": 0.7673,
+      "step": 2976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387583338128471e-06,
+      "loss": 0.8024,
+      "step": 2977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381201945321032e-06,
+      "loss": 0.7559,
+      "step": 2978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374822247782921e-06,
+      "loss": 0.7279,
+      "step": 2979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368444248502789e-06,
+      "loss": 0.7905,
+      "step": 2980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362067950468489e-06,
+      "loss": 0.793,
+      "step": 2981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3556933566670656e-06,
+      "loss": 0.7413,
+      "step": 2982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349320470084779e-06,
+      "loss": 0.7754,
+      "step": 2983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342949293707084e-06,
+      "loss": 0.7897,
+      "step": 2984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336579830518634e-06,
+      "loss": 0.7891,
+      "step": 2985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33021208350328e-06,
+      "loss": 0.7939,
+      "step": 2986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32384605564407e-06,
+      "loss": 0.8066,
+      "step": 2987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317481749923249e-06,
+      "loss": 0.8063,
+      "step": 2988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311119169322246e-06,
+      "loss": 0.7741,
+      "step": 2989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304758316821692e-06,
+      "loss": 0.7799,
+      "step": 2990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2983991954014075e-06,
+      "loss": 0.8363,
+      "step": 2991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292041808040393e-06,
+      "loss": 0.8477,
+      "step": 2992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285686157716846e-06,
+      "loss": 0.762,
+      "step": 2993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279332247408148e-06,
+      "loss": 0.7559,
+      "step": 2994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27298008009086e-06,
+      "loss": 0.8187,
+      "step": 2995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266629658740736e-06,
+      "loss": 0.8044,
+      "step": 2996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.260280986332707e-06,
+      "loss": 0.778,
+      "step": 2997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 0.791,
+      "step": 2998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247588900238549e-06,
+      "loss": 0.7757,
+      "step": 2999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2412454924981866e-06,
+      "loss": 0.7591,
+      "step": 3000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234903845591431e-06,
+      "loss": 0.7887,
+      "step": 3001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228563962489106e-06,
+      "loss": 0.7393,
+      "step": 3002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222225846161208e-06,
+      "loss": 0.7736,
+      "step": 3003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215889499576898e-06,
+      "loss": 0.7679,
+      "step": 3004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2095549257045185e-06,
+      "loss": 0.8027,
+      "step": 3005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.203222127511577e-06,
+      "loss": 0.8053,
+      "step": 3006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196891107964744e-06,
+      "loss": 0.8145,
+      "step": 3007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190561870029865e-06,
+      "loss": 0.7676,
+      "step": 3008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184234416671951e-06,
+      "loss": 0.8138,
+      "step": 3009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177908750855164e-06,
+      "loss": 0.8493,
+      "step": 3010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171584875542843e-06,
+      "loss": 0.819,
+      "step": 3011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165262793697486e-06,
+      "loss": 0.7969,
+      "step": 3012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158942508280743e-06,
+      "loss": 0.8516,
+      "step": 3013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152624022253429e-06,
+      "loss": 0.7793,
+      "step": 3014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146307338575519e-06,
+      "loss": 0.8076,
+      "step": 3015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139992460206132e-06,
+      "loss": 0.807,
+      "step": 3016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1336793901035526e-06,
+      "loss": 0.7739,
+      "step": 3017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127368131225217e-06,
+      "loss": 0.8008,
+      "step": 3018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121058686527703e-06,
+      "loss": 0.8053,
+      "step": 3019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1147510589667505e-06,
+      "loss": 0.7699,
+      "step": 3020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108445251497243e-06,
+      "loss": 0.8101,
+      "step": 3021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102141267073207e-06,
+      "loss": 0.8073,
+      "step": 3022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095839108647825e-06,
+      "loss": 0.7907,
+      "step": 3023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089538779173418e-06,
+      "loss": 0.7165,
+      "step": 3024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083240281601448e-06,
+      "loss": 0.8086,
+      "step": 3025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076943618882525e-06,
+      "loss": 0.7939,
+      "step": 3026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070648793966396e-06,
+      "loss": 0.8099,
+      "step": 3027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064355809801943e-06,
+      "loss": 0.7653,
+      "step": 3028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058064669337191e-06,
+      "loss": 0.7819,
+      "step": 3029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051775375519306e-06,
+      "loss": 0.8617,
+      "step": 3030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0454879312945755e-06,
+      "loss": 0.8171,
+      "step": 3031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.8034,
+      "step": 3032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0329186034054375e-06,
+      "loss": 0.7591,
+      "step": 3033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02663672562928e-06,
+      "loss": 0.7624,
+      "step": 3034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020356709222784e-06,
+      "loss": 0.7783,
+      "step": 3035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014078557127893e-06,
+      "loss": 0.7738,
+      "step": 3036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007802272285693e-06,
+      "loss": 0.7909,
+      "step": 3037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001527857636373e-06,
+      "loss": 0.7822,
+      "step": 3038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995255316119264e-06,
+      "loss": 0.8135,
+      "step": 3039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988984650672813e-06,
+      "loss": 0.804,
+      "step": 3040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.982715864234587e-06,
+      "loss": 0.8083,
+      "step": 3041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9764489597412744e-06,
+      "loss": 0.764,
+      "step": 3042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9701839401286844e-06,
+      "loss": 0.7767,
+      "step": 3043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963920808331736e-06,
+      "loss": 0.8092,
+      "step": 3044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95765956728447e-06,
+      "loss": 0.804,
+      "step": 3045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951400219920046e-06,
+      "loss": 0.7959,
+      "step": 3046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94514276917072e-06,
+      "loss": 0.8236,
+      "step": 3047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938887217967875e-06,
+      "loss": 0.8265,
+      "step": 3048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.7926,
+      "step": 3049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926381825922689e-06,
+      "loss": 0.7796,
+      "step": 3050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920131990938647e-06,
+      "loss": 0.8249,
+      "step": 3051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.913884067217686e-06,
+      "loss": 0.7565,
+      "step": 3052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907638057686718e-06,
+      "loss": 0.7695,
+      "step": 3053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901393965271762e-06,
+      "loss": 0.8346,
+      "step": 3054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895151792897942e-06,
+      "loss": 0.8161,
+      "step": 3055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888911543489472e-06,
+      "loss": 0.793,
+      "step": 3056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882673219969673e-06,
+      "loss": 0.7334,
+      "step": 3057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876436825260967e-06,
+      "loss": 0.7467,
+      "step": 3058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870202362284862e-06,
+      "loss": 0.8473,
+      "step": 3059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86396983396197e-06,
+      "loss": 0.7826,
+      "step": 3060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857739243211995e-06,
+      "loss": 0.736,
+      "step": 3061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851510592953729e-06,
+      "loss": 0.7158,
+      "step": 3062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845283886105058e-06,
+      "loss": 0.764,
+      "step": 3063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839059125582964e-06,
+      "loss": 0.7943,
+      "step": 3064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832836314303501e-06,
+      "loss": 0.7715,
+      "step": 3065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8266154551818225e-06,
+      "loss": 0.7959,
+      "step": 3066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82039655113217e-06,
+      "loss": 0.7721,
+      "step": 3067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814179605067856e-06,
+      "loss": 0.7373,
+      "step": 3068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807964619901283e-06,
+      "loss": 0.735,
+      "step": 3069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801751598543947e-06,
+      "loss": 0.7179,
+      "step": 3070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795540543906401e-06,
+      "loss": 0.8223,
+      "step": 3071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7893314588982905e-06,
+      "loss": 0.8587,
+      "step": 3072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783124346428342e-06,
+      "loss": 0.7435,
+      "step": 3073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776919209404342e-06,
+      "loss": 0.7484,
+      "step": 3074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770716050733165e-06,
+      "loss": 0.7217,
+      "step": 3075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764514873320761e-06,
+      "loss": 0.7129,
+      "step": 3076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758315680072137e-06,
+      "loss": 0.8424,
+      "step": 3077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752118473891379e-06,
+      "loss": 0.7777,
+      "step": 3078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745923257681656e-06,
+      "loss": 0.778,
+      "step": 3079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739730034345177e-06,
+      "loss": 0.7822,
+      "step": 3080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733538806783237e-06,
+      "loss": 0.7607,
+      "step": 3081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727349577896194e-06,
+      "loss": 0.7754,
+      "step": 3082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72116235058346e-06,
+      "loss": 0.8125,
+      "step": 3083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714977127743517e-06,
+      "loss": 0.7949,
+      "step": 3084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708793912273911e-06,
+      "loss": 0.7676,
+      "step": 3085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702612707071245e-06,
+      "loss": 0.7324,
+      "step": 3086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.696433515031169e-06,
+      "loss": 0.8229,
+      "step": 3087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6902563390484025e-06,
+      "loss": 0.7728,
+      "step": 3088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684081182016727e-06,
+      "loss": 0.7103,
+      "step": 3089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677908046828961e-06,
+      "loss": 0.7904,
+      "step": 3090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6717369363769834e-06,
+      "loss": 0.7972,
+      "step": 3091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66556785355173e-06,
+      "loss": 0.8366,
+      "step": 3092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659400801243176e-06,
+      "loss": 0.8138,
+      "step": 3093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653235782340351e-06,
+      "loss": 0.8294,
+      "step": 3094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647072799731339e-06,
+      "loss": 0.7467,
+      "step": 3095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640911856303255e-06,
+      "loss": 0.8112,
+      "step": 3096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634752954942264e-06,
+      "loss": 0.7407,
+      "step": 3097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628596098533592e-06,
+      "loss": 0.793,
+      "step": 3098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62244128996148e-06,
+      "loss": 0.7617,
+      "step": 3099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 0.7824,
+      "step": 3100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610137827859164e-06,
+      "loss": 0.793,
+      "step": 3101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603989180092661e-06,
+      "loss": 0.7842,
+      "step": 3102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597842591690128e-06,
+      "loss": 0.7839,
+      "step": 3103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.59169806553101e-06,
+      "loss": 0.7492,
+      "step": 3104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585555604493778e-06,
+      "loss": 0.8356,
+      "step": 3105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.579415211455941e-06,
+      "loss": 0.7646,
+      "step": 3106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57327688929405e-06,
+      "loss": 0.7809,
+      "step": 3107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567140640883666e-06,
+      "loss": 0.8091,
+      "step": 3108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561006469099391e-06,
+      "loss": 0.7832,
+      "step": 3109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554874376814856e-06,
+      "loss": 0.7656,
+      "step": 3110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548744366902705e-06,
+      "loss": 0.7712,
+      "step": 3111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.7562,
+      "step": 3112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.536490605681299e-06,
+      "loss": 0.8232,
+      "step": 3113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5303668601124615e-06,
+      "loss": 0.7909,
+      "step": 3114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524245208396848e-06,
+      "loss": 0.8239,
+      "step": 3115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518125653402232e-06,
+      "loss": 0.7441,
+      "step": 3116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512008197995379e-06,
+      "loss": 0.7764,
+      "step": 3117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.505892845042089e-06,
+      "loss": 0.8138,
+      "step": 3118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499779597407176e-06,
+      "loss": 0.7318,
+      "step": 3119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493668457954458e-06,
+      "loss": 0.7666,
+      "step": 3120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487559429546772e-06,
+      "loss": 0.7428,
+      "step": 3121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481452515045974e-06,
+      "loss": 0.7344,
+      "step": 3122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47534771731291e-06,
+      "loss": 0.8014,
+      "step": 3123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.469245039207452e-06,
+      "loss": 0.7827,
+      "step": 3124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463144483588469e-06,
+      "loss": 0.846,
+      "step": 3125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457046053313844e-06,
+      "loss": 0.7861,
+      "step": 3126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450949751240456e-06,
+      "loss": 0.7855,
+      "step": 3127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444855580224197e-06,
+      "loss": 0.7533,
+      "step": 3128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438763543119947e-06,
+      "loss": 0.7448,
+      "step": 3129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432673642781595e-06,
+      "loss": 0.7616,
+      "step": 3130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42658588206203e-06,
+      "loss": 0.8294,
+      "step": 3131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420500263813141e-06,
+      "loss": 0.7803,
+      "step": 3132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4144167908857994e-06,
+      "loss": 0.818,
+      "step": 3133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4083354661298816e-06,
+      "loss": 0.7562,
+      "step": 3134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402256292394268e-06,
+      "loss": 0.7861,
+      "step": 3135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39617927252681e-06,
+      "loss": 0.8197,
+      "step": 3136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.390104409374364e-06,
+      "loss": 0.7865,
+      "step": 3137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384031705782776e-06,
+      "loss": 0.793,
+      "step": 3138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3779611645968696e-06,
+      "loss": 0.7995,
+      "step": 3139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371892788660465e-06,
+      "loss": 0.7656,
+      "step": 3140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.365826580816371e-06,
+      "loss": 0.7728,
+      "step": 3141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3597625439063685e-06,
+      "loss": 0.8005,
+      "step": 3142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353700680771225e-06,
+      "loss": 0.7539,
+      "step": 3143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347640994250709e-06,
+      "loss": 0.7725,
+      "step": 3144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.341583487183537e-06,
+      "loss": 0.8216,
+      "step": 3145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335528162407428e-06,
+      "loss": 0.7949,
+      "step": 3146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329475022759074e-06,
+      "loss": 0.741,
+      "step": 3147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3234240710741335e-06,
+      "loss": 0.833,
+      "step": 3148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317375310187252e-06,
+      "loss": 0.7839,
+      "step": 3149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311328742932045e-06,
+      "loss": 0.7363,
+      "step": 3150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.305284372141095e-06,
+      "loss": 0.7819,
+      "step": 3151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299242200645959e-06,
+      "loss": 0.8145,
+      "step": 3152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293202231277174e-06,
+      "loss": 0.7441,
+      "step": 3153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28716446686423e-06,
+      "loss": 0.8164,
+      "step": 3154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281128910235586e-06,
+      "loss": 0.7507,
+      "step": 3155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275095564218683e-06,
+      "loss": 0.8141,
+      "step": 3156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269064431639901e-06,
+      "loss": 0.7591,
+      "step": 3157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263035515324601e-06,
+      "loss": 0.7467,
+      "step": 3158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257008818097108e-06,
+      "loss": 0.7884,
+      "step": 3159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2509843427806895e-06,
+      "loss": 0.7767,
+      "step": 3160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244962092197589e-06,
+      "loss": 0.7546,
+      "step": 3161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238942069169e-06,
+      "loss": 0.748,
+      "step": 3162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232924276515076e-06,
+      "loss": 0.7985,
+      "step": 3163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226908717054923e-06,
+      "loss": 0.8346,
+      "step": 3164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2208953936066085e-06,
+      "loss": 0.6673,
+      "step": 3165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.214884308987136e-06,
+      "loss": 0.8457,
+      "step": 3166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208875466012475e-06,
+      "loss": 0.7409,
+      "step": 3167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202868867497542e-06,
+      "loss": 0.791,
+      "step": 3168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196864516256193e-06,
+      "loss": 0.7806,
+      "step": 3169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1908624151012444e-06,
+      "loss": 0.7533,
+      "step": 3170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18486256684445e-06,
+      "loss": 0.8008,
+      "step": 3171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178864974296511e-06,
+      "loss": 0.7959,
+      "step": 3172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172869640267073e-06,
+      "loss": 0.7952,
+      "step": 3173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166876567564725e-06,
+      "loss": 0.8148,
+      "step": 3174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160885758996985e-06,
+      "loss": 0.7401,
+      "step": 3175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154897217370325e-06,
+      "loss": 0.7046,
+      "step": 3176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148910945490152e-06,
+      "loss": 0.7928,
+      "step": 3177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142926946160799e-06,
+      "loss": 0.7469,
+      "step": 3178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136945222185546e-06,
+      "loss": 0.7855,
+      "step": 3179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1309657763665984e-06,
+      "loss": 0.7767,
+      "step": 3180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124988611505114e-06,
+      "loss": 0.7751,
+      "step": 3181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119013730401152e-06,
+      "loss": 0.7663,
+      "step": 3182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113041135853724e-06,
+      "loss": 0.749,
+      "step": 3183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107070830660765e-06,
+      "loss": 0.7666,
+      "step": 3184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101102817619132e-06,
+      "loss": 0.779,
+      "step": 3185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0951370995246104e-06,
+      "loss": 0.8193,
+      "step": 3186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089173679171922e-06,
+      "loss": 0.779,
+      "step": 3187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08321255935469e-06,
+      "loss": 0.7699,
+      "step": 3188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07725374286548e-06,
+      "loss": 0.7503,
+      "step": 3189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0712972324957686e-06,
+      "loss": 0.7539,
+      "step": 3190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065343031035956e-06,
+      "loss": 0.8464,
+      "step": 3191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059391141275358e-06,
+      "loss": 0.7861,
+      "step": 3192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053441566002214e-06,
+      "loss": 0.8379,
+      "step": 3193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047494308003666e-06,
+      "loss": 0.7367,
+      "step": 3194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041549370065779e-06,
+      "loss": 0.749,
+      "step": 3195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0356067549735384e-06,
+      "loss": 0.766,
+      "step": 3196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029666465510825e-06,
+      "loss": 0.7686,
+      "step": 3197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023728504460441e-06,
+      "loss": 0.776,
+      "step": 3198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017792874604097e-06,
+      "loss": 0.7425,
+      "step": 3199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011859578722408e-06,
+      "loss": 0.7165,
+      "step": 3200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0059286195948996e-06,
+      "loss": 0.792,
+      "step": 3201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.7949,
+      "step": 3202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9940737227150435e-06,
+      "loss": 0.8148,
+      "step": 3203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9881497905162625e-06,
+      "loss": 0.7337,
+      "step": 3204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982228206178799e-06,
+      "loss": 0.8525,
+      "step": 3205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976308972476682e-06,
+      "loss": 0.8232,
+      "step": 3206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970392092182853e-06,
+      "loss": 0.7484,
+      "step": 3207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964477568069146e-06,
+      "loss": 0.7995,
+      "step": 3208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958565402906288e-06,
+      "loss": 0.8008,
+      "step": 3209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952655599463907e-06,
+      "loss": 0.8252,
+      "step": 3210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946748160510522e-06,
+      "loss": 0.7783,
+      "step": 3211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940843088813537e-06,
+      "loss": 0.7962,
+      "step": 3212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934940387139259e-06,
+      "loss": 0.7301,
+      "step": 3213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929040058252882e-06,
+      "loss": 0.7339,
+      "step": 3214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9231421049184766e-06,
+      "loss": 0.7887,
+      "step": 3215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917246529899017e-06,
+      "loss": 0.8197,
+      "step": 3216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911353335956353e-06,
+      "loss": 0.6961,
+      "step": 3217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9054625258512226e-06,
+      "loss": 0.7887,
+      "step": 3218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899574102343247e-06,
+      "loss": 0.7381,
+      "step": 3219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893688068190933e-06,
+      "loss": 0.8047,
+      "step": 3220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887804426151655e-06,
+      "loss": 0.7103,
+      "step": 3221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881923178981681e-06,
+      "loss": 0.777,
+      "step": 3222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876044329436154e-06,
+      "loss": 0.8288,
+      "step": 3223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870167880269085e-06,
+      "loss": 0.8066,
+      "step": 3224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86429383423337e-06,
+      "loss": 0.806,
+      "step": 3225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858422194080777e-06,
+      "loss": 0.722,
+      "step": 3226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852552962561946e-06,
+      "loss": 0.7274,
+      "step": 3227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846686142426389e-06,
+      "loss": 0.7345,
+      "step": 3228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8408217364224886e-06,
+      "loss": 0.7861,
+      "step": 3229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834959747297498e-06,
+      "loss": 0.7783,
+      "step": 3230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829100177797531e-06,
+      "loss": 0.7269,
+      "step": 3231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823243030667576e-06,
+      "loss": 0.7728,
+      "step": 3232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817388308651487e-06,
+      "loss": 0.7852,
+      "step": 3233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811536014491972e-06,
+      "loss": 0.7871,
+      "step": 3234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805686150930613e-06,
+      "loss": 0.7975,
+      "step": 3235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.799838720707847e-06,
+      "loss": 0.7845,
+      "step": 3236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7939937265629725e-06,
+      "loss": 0.7152,
+      "step": 3237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788151171234149e-06,
+      "loss": 0.7666,
+      "step": 3238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782311057458393e-06,
+      "loss": 0.7598,
+      "step": 3239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7764733879715706e-06,
+      "loss": 0.7651,
+      "step": 3240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.770638165508411e-06,
+      "loss": 0.7061,
+      "step": 3241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764805392802497e-06,
+      "loss": 0.806,
+      "step": 3242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758975072586255e-06,
+      "loss": 0.8008,
+      "step": 3243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753147207590972e-06,
+      "loss": 0.807,
+      "step": 3244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747321800546778e-06,
+      "loss": 0.8363,
+      "step": 3245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741498854182659e-06,
+      "loss": 0.7692,
+      "step": 3246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7356783712264405e-06,
+      "loss": 0.7637,
+      "step": 3247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729860354404805e-06,
+      "loss": 0.752,
+      "step": 3248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724044806443262e-06,
+      "loss": 0.7826,
+      "step": 3249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718231730066179e-06,
+      "loss": 0.7702,
+      "step": 3250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7124211279967656e-06,
+      "loss": 0.7666,
+      "step": 3251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.70661300295706e-06,
+      "loss": 0.7435,
+      "step": 3252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700807357667953e-06,
+      "loss": 0.734,
+      "step": 3253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695004194849166e-06,
+      "loss": 0.7503,
+      "step": 3254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689203517219263e-06,
+      "loss": 0.7223,
+      "step": 3255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683405327495638e-06,
+      "loss": 0.7965,
+      "step": 3256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677609628394529e-06,
+      "loss": 0.6797,
+      "step": 3257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671816422630993e-06,
+      "loss": 0.7578,
+      "step": 3258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666025712918929e-06,
+      "loss": 0.8574,
+      "step": 3259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66023750197107e-06,
+      "loss": 0.785,
+      "step": 3260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654451792498963e-06,
+      "loss": 0.7689,
+      "step": 3261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648668587212998e-06,
+      "loss": 0.7959,
+      "step": 3262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642887888822387e-06,
+      "loss": 0.7526,
+      "step": 3263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637109700035166e-06,
+      "loss": 0.7542,
+      "step": 3264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631334023558199e-06,
+      "loss": 0.818,
+      "step": 3265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625560862097174e-06,
+      "loss": 0.8366,
+      "step": 3266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619790218356589e-06,
+      "loss": 0.7874,
+      "step": 3267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614022095039776e-06,
+      "loss": 0.7607,
+      "step": 3268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608256494848886e-06,
+      "loss": 0.8271,
+      "step": 3269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6024934204848745e-06,
+      "loss": 0.7926,
+      "step": 3270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596732874647527e-06,
+      "loss": 0.7041,
+      "step": 3271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5909748600354395e-06,
+      "loss": 0.7516,
+      "step": 3272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5852193793460255e-06,
+      "loss": 0.7764,
+      "step": 3273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579466435275506e-06,
+      "loss": 0.7894,
+      "step": 3274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5737160305189185e-06,
+      "loss": 0.7428,
+      "step": 3275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567968167770113e-06,
+      "loss": 0.7812,
+      "step": 3276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562222849721735e-06,
+      "loss": 0.7588,
+      "step": 3277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556480079065253e-06,
+      "loss": 0.7396,
+      "step": 3278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55073985849094e-06,
+      "loss": 0.7926,
+      "step": 3279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545002190687865e-06,
+      "loss": 0.7464,
+      "step": 3280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.53926707834391e-06,
+      "loss": 0.8148,
+      "step": 3281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533534524145756e-06,
+      "loss": 0.806,
+      "step": 3282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5278045307788885e-06,
+      "loss": 0.7497,
+      "step": 3283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522077100927591e-06,
+      "loss": 0.7646,
+      "step": 3284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516352237274949e-06,
+      "loss": 0.7949,
+      "step": 3285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510629942502839e-06,
+      "loss": 0.7432,
+      "step": 3286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504910219291941e-06,
+      "loss": 0.7708,
+      "step": 3287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499193070321729e-06,
+      "loss": 0.7905,
+      "step": 3288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493478498270467e-06,
+      "loss": 0.7646,
+      "step": 3289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487766505815215e-06,
+      "loss": 0.8107,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482057095631826e-06,
+      "loss": 0.8047,
+      "step": 3291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476350270394942e-06,
+      "loss": 0.7572,
+      "step": 3292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470646032777992e-06,
+      "loss": 0.7633,
+      "step": 3293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4649443854532005e-06,
+      "loss": 0.8053,
+      "step": 3294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459245331091563e-06,
+      "loss": 0.7855,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453548872362875e-06,
+      "loss": 0.7682,
+      "step": 3296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447855011935714e-06,
+      "loss": 0.8132,
+      "step": 3297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442163752477429e-06,
+      "loss": 0.7956,
+      "step": 3298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436475096654163e-06,
+      "loss": 0.7581,
+      "step": 3299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430789047130836e-06,
+      "loss": 0.7926,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425105606571145e-06,
+      "loss": 0.7396,
+      "step": 3301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419424777637565e-06,
+      "loss": 0.7464,
+      "step": 3302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413746562991355e-06,
+      "loss": 0.7969,
+      "step": 3303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 0.8177,
+      "step": 3304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402397987199906e-06,
+      "loss": 0.7598,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396727631371049e-06,
+      "loss": 0.7925,
+      "step": 3306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391059900462305e-06,
+      "loss": 0.8021,
+      "step": 3307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385394797128791e-06,
+      "loss": 0.7738,
+      "step": 3308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379732324024395e-06,
+      "loss": 0.7712,
+      "step": 3309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374072483801769e-06,
+      "loss": 0.7331,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368415279112334e-06,
+      "loss": 0.751,
+      "step": 3311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362760712606278e-06,
+      "loss": 0.8073,
+      "step": 3312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357108786932542e-06,
+      "loss": 0.7731,
+      "step": 3313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351459504738844e-06,
+      "loss": 0.7699,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345812868671663e-06,
+      "loss": 0.7878,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340168881376222e-06,
+      "loss": 0.7539,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334527545496521e-06,
+      "loss": 0.8079,
+      "step": 3317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32888886367531e-06,
+      "loss": 0.7689,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323252838554099e-06,
+      "loss": 0.7555,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31761947277315e-06,
+      "loss": 0.8005,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311988768971484e-06,
+      "loss": 0.7751,
+      "step": 3321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306360729786867e-06,
+      "loss": 0.7985,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300735357855822e-06,
+      "loss": 0.7607,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295112655813622e-06,
+      "loss": 0.7767,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289492626294296e-06,
+      "loss": 0.79,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283875271930603e-06,
+      "loss": 0.8024,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278260595354067e-06,
+      "loss": 0.8141,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272648599194948e-06,
+      "loss": 0.7559,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267039286082255e-06,
+      "loss": 0.7295,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261432658643736e-06,
+      "loss": 0.8066,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2558287195058875e-06,
+      "loss": 0.7461,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2502274712939355e-06,
+      "loss": 0.7956,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244628916631853e-06,
+      "loss": 0.7852,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2390330581423564e-06,
+      "loss": 0.7917,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233439898446885e-06,
+      "loss": 0.793,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227849440165623e-06,
+      "loss": 0.7891,
+      "step": 3336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222261685917489e-06,
+      "loss": 0.7936,
+      "step": 3337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216676638320135e-06,
+      "loss": 0.7633,
+      "step": 3338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211094299989941e-06,
+      "loss": 0.8083,
+      "step": 3339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205514673542025e-06,
+      "loss": 0.7891,
+      "step": 3340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199937761590222e-06,
+      "loss": 0.7529,
+      "step": 3341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1943635667471095e-06,
+      "loss": 0.7845,
+      "step": 3342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188792091623985e-06,
+      "loss": 0.7956,
+      "step": 3343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18322333883087e-06,
+      "loss": 0.7611,
+      "step": 3344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177657310976515e-06,
+      "loss": 0.7201,
+      "step": 3345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1720940106683915e-06,
+      "loss": 0.7604,
+      "step": 3346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166533440512696e-06,
+      "loss": 0.7721,
+      "step": 3347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1609756031143445e-06,
+      "loss": 0.7868,
+      "step": 3348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155420501076975e-06,
+      "loss": 0.8086,
+      "step": 3349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149868137002934e-06,
+      "loss": 0.7956,
+      "step": 3350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144318513493298e-06,
+      "loss": 0.7965,
+      "step": 3351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138771633147856e-06,
+      "loss": 0.7741,
+      "step": 3352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133227498565107e-06,
+      "loss": 0.7819,
+      "step": 3353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127686112342268e-06,
+      "loss": 0.7848,
+      "step": 3354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.7467,
+      "step": 3355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.11661159535875e-06,
+      "loss": 0.7894,
+      "step": 3356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111078469786062e-06,
+      "loss": 0.7552,
+      "step": 3357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105548102949265e-06,
+      "loss": 0.7708,
+      "step": 3358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100020497439121e-06,
+      "loss": 0.7952,
+      "step": 3359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0944956558451055e-06,
+      "loss": 0.8109,
+      "step": 3360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088973580755401e-06,
+      "loss": 0.7806,
+      "step": 3361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083454274756881e-06,
+      "loss": 0.805,
+      "step": 3362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077937740435136e-06,
+      "loss": 0.7493,
+      "step": 3363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0724239803744524e-06,
+      "loss": 0.7484,
+      "step": 3364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066912997157816e-06,
+      "loss": 0.792,
+      "step": 3365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061404793366914e-06,
+      "loss": 0.7451,
+      "step": 3366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0558993715821335e-06,
+      "loss": 0.7301,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05039673438255e-06,
+      "loss": 0.7799,
+      "step": 3368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.04489688434594e-06,
+      "loss": 0.805,
+      "step": 3369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039399824048777e-06,
+      "loss": 0.7568,
+      "step": 3370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033905556066228e-06,
+      "loss": 0.7751,
+      "step": 3371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.7479,
+      "step": 3372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022925407339064e-06,
+      "loss": 0.7549,
+      "step": 3373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017439531738238e-06,
+      "loss": 0.7812,
+      "step": 3374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011956458739582e-06,
+      "loss": 0.8161,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00647619091171e-06,
+      "loss": 0.7692,
+      "step": 3376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000998730821922e-06,
+      "loss": 0.7393,
+      "step": 3377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9955240810361926e-06,
+      "loss": 0.7777,
+      "step": 3378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990052244119191e-06,
+      "loss": 0.7972,
+      "step": 3379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9845832226342696e-06,
+      "loss": 0.7721,
+      "step": 3380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979117019143447e-06,
+      "loss": 0.7708,
+      "step": 3381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 0.8249,
+      "step": 3382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968193076385628e-06,
+      "loss": 0.791,
+      "step": 3383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962735342236082e-06,
+      "loss": 0.7111,
+      "step": 3384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957280436315543e-06,
+      "loss": 0.8232,
+      "step": 3385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9518283611794286e-06,
+      "loss": 0.7331,
+      "step": 3386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946379119381822e-06,
+      "loss": 0.7145,
+      "step": 3387
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940932713475489e-06,
+      "loss": 0.7871,
+      "step": 3388
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9354891460118695e-06,
+      "loss": 0.8086,
+      "step": 3389
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930048419541057e-06,
+      "loss": 0.7894,
+      "step": 3390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924610536611833e-06,
+      "loss": 0.8105,
+      "step": 3391
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919175499771635e-06,
+      "loss": 0.7995,
+      "step": 3392
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913743311566573e-06,
+      "loss": 0.7497,
+      "step": 3393
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.908313974541422e-06,
+      "loss": 0.7633,
+      "step": 3394
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9028874912396205e-06,
+      "loss": 0.7676,
+      "step": 3395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897463864203266e-06,
+      "loss": 0.7507,
+      "step": 3396
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892043095973123e-06,
+      "loss": 0.7826,
+      "step": 3397
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886625189088617e-06,
+      "loss": 0.8291,
+      "step": 3398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881210146087829e-06,
+      "loss": 0.7679,
+      "step": 3399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875797969507502e-06,
+      "loss": 0.7393,
+      "step": 3400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870388661883034e-06,
+      "loss": 0.8187,
+      "step": 3401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864982225748481e-06,
+      "loss": 0.8073,
+      "step": 3402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859578663636554e-06,
+      "loss": 0.7878,
+      "step": 3403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854177978078617e-06,
+      "loss": 0.7243,
+      "step": 3404
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848780171604682e-06,
+      "loss": 0.8359,
+      "step": 3405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.8639,
+      "step": 3406
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.837993206022146e-06,
+      "loss": 0.7712,
+      "step": 3407
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832604051966825e-06,
+      "loss": 0.745,
+      "step": 3408
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827217787102072e-06,
+      "loss": 0.7523,
+      "step": 3409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821834413951148e-06,
+      "loss": 0.7826,
+      "step": 3410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81645393503596e-06,
+      "loss": 0.7964,
+      "step": 3411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8110763528770543e-06,
+      "loss": 0.7959,
+      "step": 3412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80570166999363e-06,
+      "loss": 0.7546,
+      "step": 3413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8003298889035146e-06,
+      "loss": 0.7975,
+      "step": 3414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7949610121231862e-06,
+      "loss": 0.7633,
+      "step": 3415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789595042167763e-06,
+      "loss": 0.7855,
+      "step": 3416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784231981550991e-06,
+      "loss": 0.7445,
+      "step": 3417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7788718327852625e-06,
+      "loss": 0.8301,
+      "step": 3418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.773514598381606e-06,
+      "loss": 0.7803,
+      "step": 3419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768160280849681e-06,
+      "loss": 0.7269,
+      "step": 3420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7628088826977815e-06,
+      "loss": 0.716,
+      "step": 3421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7574604064328336e-06,
+      "loss": 0.7686,
+      "step": 3422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7521148545604003e-06,
+      "loss": 0.7542,
+      "step": 3423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7467722295846596e-06,
+      "loss": 0.7467,
+      "step": 3424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7414325340084355e-06,
+      "loss": 0.7682,
+      "step": 3425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7360957703331734e-06,
+      "loss": 0.8044,
+      "step": 3426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730761941058938e-06,
+      "loss": 0.7995,
+      "step": 3427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725431048684428e-06,
+      "loss": 0.7738,
+      "step": 3428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720103095706965e-06,
+      "loss": 0.7923,
+      "step": 3429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7147780846224922e-06,
+      "loss": 0.7738,
+      "step": 3430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7094560179255743e-06,
+      "loss": 0.82,
+      "step": 3431
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704136898109403e-06,
+      "loss": 0.7524,
+      "step": 3432
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698820727665773e-06,
+      "loss": 0.7412,
+      "step": 3433
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.693507509085115e-06,
+      "loss": 0.7581,
+      "step": 3434
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688197244856471e-06,
+      "loss": 0.8073,
+      "step": 3435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6828899374674933e-06,
+      "loss": 0.7679,
+      "step": 3436
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6775855894044543e-06,
+      "loss": 0.7747,
+      "step": 3437
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6722842031522423e-06,
+      "loss": 0.7191,
+      "step": 3438
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666985781194352e-06,
+      "loss": 0.7598,
+      "step": 3439
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661690326012897e-06,
+      "loss": 0.7337,
+      "step": 3440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656397840088596e-06,
+      "loss": 0.7725,
+      "step": 3441
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651108325900773e-06,
+      "loss": 0.7803,
+      "step": 3442
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645821785927367e-06,
+      "loss": 0.7786,
+      "step": 3443
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.640538222644925e-06,
+      "loss": 0.7686,
+      "step": 3444
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.635257638528589e-06,
+      "loss": 0.7799,
+      "step": 3445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6299800360521133e-06,
+      "loss": 0.7754,
+      "step": 3446
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624705417687856e-06,
+      "loss": 0.798,
+      "step": 3447
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619433785906775e-06,
+      "loss": 0.7285,
+      "step": 3448
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6141651431784287e-06,
+      "loss": 0.793,
+      "step": 3449
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6088994919709795e-06,
+      "loss": 0.7596,
+      "step": 3450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60363683475118e-06,
+      "loss": 0.7337,
+      "step": 3451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5983771739843855e-06,
+      "loss": 0.7201,
+      "step": 3452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5931205121345514e-06,
+      "loss": 0.7936,
+      "step": 3453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.587866851664219e-06,
+      "loss": 0.7471,
+      "step": 3454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5826161950345295e-06,
+      "loss": 0.7474,
+      "step": 3455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5773685447052177e-06,
+      "loss": 0.7458,
+      "step": 3456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.7822,
+      "step": 3457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.566882272779614e-06,
+      "loss": 0.792,
+      "step": 3458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5616436560957447e-06,
+      "loss": 0.7985,
+      "step": 3459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556408055537087e-06,
+      "loss": 0.7197,
+      "step": 3460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551175473556323e-06,
+      "loss": 0.7334,
+      "step": 3461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5459459126047226e-06,
+      "loss": 0.7417,
+      "step": 3462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540719375132129e-06,
+      "loss": 0.7829,
+      "step": 3463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.535495863586981e-06,
+      "loss": 0.7438,
+      "step": 3464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5302753804162925e-06,
+      "loss": 0.7917,
+      "step": 3465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525057928065664e-06,
+      "loss": 0.7891,
+      "step": 3466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5198435089792726e-06,
+      "loss": 0.7969,
+      "step": 3467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514632125599876e-06,
+      "loss": 0.7396,
+      "step": 3468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5094237803688114e-06,
+      "loss": 0.7547,
+      "step": 3469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5042184757259844e-06,
+      "loss": 0.756,
+      "step": 3470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.499016214109886e-06,
+      "loss": 0.7988,
+      "step": 3471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493816997957582e-06,
+      "loss": 0.7288,
+      "step": 3472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488620829704699e-06,
+      "loss": 0.7637,
+      "step": 3473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483427711785449e-06,
+      "loss": 0.7334,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4782376466326108e-06,
+      "loss": 0.7943,
+      "step": 3475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4730506366775306e-06,
+      "loss": 0.743,
+      "step": 3476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4678666843501276e-06,
+      "loss": 0.7425,
+      "step": 3477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462685792078888e-06,
+      "loss": 0.7476,
+      "step": 3478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.457507962290858e-06,
+      "loss": 0.7217,
+      "step": 3479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452333197411657e-06,
+      "loss": 0.8011,
+      "step": 3480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.447161499865467e-06,
+      "loss": 0.743,
+      "step": 3481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4419928720750274e-06,
+      "loss": 0.7533,
+      "step": 3482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.436827316461646e-06,
+      "loss": 0.7503,
+      "step": 3483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4316648354451896e-06,
+      "loss": 0.765,
+      "step": 3484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4265054314440837e-06,
+      "loss": 0.7319,
+      "step": 3485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4213491068753135e-06,
+      "loss": 0.7853,
+      "step": 3486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416195864154426e-06,
+      "loss": 0.7529,
+      "step": 3487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4110457056955095e-06,
+      "loss": 0.7826,
+      "step": 3488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405898633911223e-06,
+      "loss": 0.8073,
+      "step": 3489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4007546512127764e-06,
+      "loss": 0.762,
+      "step": 3490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3956137600099248e-06,
+      "loss": 0.7646,
+      "step": 3491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3904759627109828e-06,
+      "loss": 0.7943,
+      "step": 3492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3853412617228123e-06,
+      "loss": 0.7435,
+      "step": 3493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380209659450827e-06,
+      "loss": 0.7223,
+      "step": 3494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3750811582989863e-06,
+      "loss": 0.7757,
+      "step": 3495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369955760669802e-06,
+      "loss": 0.7868,
+      "step": 3496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3648334689643214e-06,
+      "loss": 0.7638,
+      "step": 3497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359714285582146e-06,
+      "loss": 0.7456,
+      "step": 3498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3545982129214227e-06,
+      "loss": 0.8177,
+      "step": 3499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3494852533788313e-06,
+      "loss": 0.7949,
+      "step": 3500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3443754093496007e-06,
+      "loss": 0.779,
+      "step": 3501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.339268683227499e-06,
+      "loss": 0.7171,
+      "step": 3502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334165077404834e-06,
+      "loss": 0.7627,
+      "step": 3503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329064594272451e-06,
+      "loss": 0.6935,
+      "step": 3504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323967236219734e-06,
+      "loss": 0.7913,
+      "step": 3505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3188730056345974e-06,
+      "loss": 0.7998,
+      "step": 3506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3137819049034957e-06,
+      "loss": 0.7666,
+      "step": 3507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.7913,
+      "step": 3508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3036091025418857e-06,
+      "loss": 0.7588,
+      "step": 3509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2985274056769445e-06,
+      "loss": 0.7686,
+      "step": 3510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2934488481971806e-06,
+      "loss": 0.7796,
+      "step": 3511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288373432481703e-06,
+      "loss": 0.7871,
+      "step": 3512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283301160908151e-06,
+      "loss": 0.7734,
+      "step": 3513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2782320358526933e-06,
+      "loss": 0.791,
+      "step": 3514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273166059690024e-06,
+      "loss": 0.8216,
+      "step": 3515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2681032347933537e-06,
+      "loss": 0.742,
+      "step": 3516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2630435635344283e-06,
+      "loss": 0.7624,
+      "step": 3517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2579870482835117e-06,
+      "loss": 0.7594,
+      "step": 3518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2529336914093856e-06,
+      "loss": 0.8102,
+      "step": 3519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247883495279358e-06,
+      "loss": 0.7493,
+      "step": 3520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2428364622592522e-06,
+      "loss": 0.769,
+      "step": 3521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2377925947134137e-06,
+      "loss": 0.7816,
+      "step": 3522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232751895004701e-06,
+      "loss": 0.7559,
+      "step": 3523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2277143654944944e-06,
+      "loss": 0.7904,
+      "step": 3524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.222680008542678e-06,
+      "loss": 0.8008,
+      "step": 3525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.21764882650766e-06,
+      "loss": 0.7412,
+      "step": 3526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.212620821746362e-06,
+      "loss": 0.8001,
+      "step": 3527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2075959966142055e-06,
+      "loss": 0.7591,
+      "step": 3528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2025743534651343e-06,
+      "loss": 0.7274,
+      "step": 3529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197555894651596e-06,
+      "loss": 0.6925,
+      "step": 3530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1925406225245494e-06,
+      "loss": 0.7861,
+      "step": 3531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1875285394334575e-06,
+      "loss": 0.8112,
+      "step": 3532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1825196477262954e-06,
+      "loss": 0.7023,
+      "step": 3533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.17751394974953e-06,
+      "loss": 0.7988,
+      "step": 3534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1725114478481458e-06,
+      "loss": 0.7419,
+      "step": 3535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1675121443656266e-06,
+      "loss": 0.7581,
+      "step": 3536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1625160416439503e-06,
+      "loss": 0.7741,
+      "step": 3537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.157523142023604e-06,
+      "loss": 0.8011,
+      "step": 3538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152533447843571e-06,
+      "loss": 0.7575,
+      "step": 3539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1475469614413346e-06,
+      "loss": 0.7926,
+      "step": 3540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.142563685152873e-06,
+      "loss": 0.7378,
+      "step": 3541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.695,
+      "step": 3542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132606772253677e-06,
+      "loss": 0.8135,
+      "step": 3543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1276331403073733e-06,
+      "loss": 0.7347,
+      "step": 3544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122662727803717e-06,
+      "loss": 0.7194,
+      "step": 3545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.117695537071149e-06,
+      "loss": 0.7803,
+      "step": 3546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1127315704366144e-06,
+      "loss": 0.7386,
+      "step": 3547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107770830225543e-06,
+      "loss": 0.8206,
+      "step": 3548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1028133187618515e-06,
+      "loss": 0.7708,
+      "step": 3549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0978590383679476e-06,
+      "loss": 0.7679,
+      "step": 3550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0929079913647243e-06,
+      "loss": 0.7803,
+      "step": 3551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087960180071553e-06,
+      "loss": 0.7669,
+      "step": 3552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0830156068063e-06,
+      "loss": 0.8255,
+      "step": 3553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.078074273885312e-06,
+      "loss": 0.8558,
+      "step": 3554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.07313618362341e-06,
+      "loss": 0.7513,
+      "step": 3555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0682013383339026e-06,
+      "loss": 0.7949,
+      "step": 3556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063269740328579e-06,
+      "loss": 0.7796,
+      "step": 3557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0583413919177063e-06,
+      "loss": 0.7376,
+      "step": 3558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0534162954100264e-06,
+      "loss": 0.7992,
+      "step": 3559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.048494453112765e-06,
+      "loss": 0.8079,
+      "step": 3560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0435758673316098e-06,
+      "loss": 0.6912,
+      "step": 3561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0386605403707347e-06,
+      "loss": 0.6828,
+      "step": 3562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033748474532785e-06,
+      "loss": 0.8018,
+      "step": 3563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0288396721188786e-06,
+      "loss": 0.7897,
+      "step": 3564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0239341354285968e-06,
+      "loss": 0.7493,
+      "step": 3565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0190318667600003e-06,
+      "loss": 0.7461,
+      "step": 3566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014132868409617e-06,
+      "loss": 0.7565,
+      "step": 3567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0092371426724397e-06,
+      "loss": 0.7643,
+      "step": 3568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0043446918419294e-06,
+      "loss": 0.7533,
+      "step": 3569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999455518210018e-06,
+      "loss": 0.7757,
+      "step": 3570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9945696240670905e-06,
+      "loss": 0.7812,
+      "step": 3571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9896870117020073e-06,
+      "loss": 0.7676,
+      "step": 3572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9848076834020868e-06,
+      "loss": 0.7168,
+      "step": 3573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.979931641453104e-06,
+      "loss": 0.7249,
+      "step": 3574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975058888139304e-06,
+      "loss": 0.8346,
+      "step": 3575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.970189425743383e-06,
+      "loss": 0.7549,
+      "step": 3576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9653232565465017e-06,
+      "loss": 0.7367,
+      "step": 3577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9604603828282753e-06,
+      "loss": 0.7601,
+      "step": 3578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9556008068667776e-06,
+      "loss": 0.7217,
+      "step": 3579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95074453093853e-06,
+      "loss": 0.7559,
+      "step": 3580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9458915573185143e-06,
+      "loss": 0.7871,
+      "step": 3581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9410418882801682e-06,
+      "loss": 0.7646,
+      "step": 3582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936195526095372e-06,
+      "loss": 0.7422,
+      "step": 3583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9313524730344644e-06,
+      "loss": 0.7412,
+      "step": 3584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9265127313662325e-06,
+      "loss": 0.8076,
+      "step": 3585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.92167630335791e-06,
+      "loss": 0.7546,
+      "step": 3586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9168431912751805e-06,
+      "loss": 0.7757,
+      "step": 3587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9120133973821762e-06,
+      "loss": 0.8128,
+      "step": 3588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.907186923941466e-06,
+      "loss": 0.7943,
+      "step": 3589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.902363773214072e-06,
+      "loss": 0.7601,
+      "step": 3590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8975439474594604e-06,
+      "loss": 0.7585,
+      "step": 3591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8927274489355296e-06,
+      "loss": 0.7279,
+      "step": 3592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8879142798986293e-06,
+      "loss": 0.7816,
+      "step": 3593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883104442603547e-06,
+      "loss": 0.751,
+      "step": 3594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878297939303507e-06,
+      "loss": 0.7637,
+      "step": 3595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873494772250176e-06,
+      "loss": 0.8376,
+      "step": 3596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868694943693655e-06,
+      "loss": 0.752,
+      "step": 3597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8638984558824777e-06,
+      "loss": 0.7874,
+      "step": 3598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859105311063619e-06,
+      "loss": 0.762,
+      "step": 3599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8543155114824873e-06,
+      "loss": 0.7617,
+      "step": 3600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.849529059382915e-06,
+      "loss": 0.7607,
+      "step": 3601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844745957007178e-06,
+      "loss": 0.765,
+      "step": 3602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8399662065959766e-06,
+      "loss": 0.8236,
+      "step": 3603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8351898103884413e-06,
+      "loss": 0.7949,
+      "step": 3604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8304167706221344e-06,
+      "loss": 0.7894,
+      "step": 3605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8256470895330447e-06,
+      "loss": 0.7969,
+      "step": 3606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820880769355582e-06,
+      "loss": 0.7772,
+      "step": 3607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8161178123225873e-06,
+      "loss": 0.8587,
+      "step": 3608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811358220665328e-06,
+      "loss": 0.8001,
+      "step": 3609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 0.7526,
+      "step": 3610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.801849142395182e-06,
+      "loss": 0.7155,
+      "step": 3611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797099660236937e-06,
+      "loss": 0.7536,
+      "step": 3612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792353552363707e-06,
+      "loss": 0.8158,
+      "step": 3613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7876108209988616e-06,
+      "loss": 0.766,
+      "step": 3614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7828714683641923e-06,
+      "loss": 0.7607,
+      "step": 3615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778135496679908e-06,
+      "loss": 0.7855,
+      "step": 3616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.773402908164625e-06,
+      "loss": 0.7848,
+      "step": 3617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768673705035384e-06,
+      "loss": 0.7578,
+      "step": 3618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76394788950764e-06,
+      "loss": 0.7826,
+      "step": 3619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7592254637952533e-06,
+      "loss": 0.8464,
+      "step": 3620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754506430110501e-06,
+      "loss": 0.7699,
+      "step": 3621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749790790664074e-06,
+      "loss": 0.7956,
+      "step": 3622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.745078547665068e-06,
+      "loss": 0.8001,
+      "step": 3623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7403697033209907e-06,
+      "loss": 0.7297,
+      "step": 3624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7356642598377604e-06,
+      "loss": 0.776,
+      "step": 3625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7309622194196905e-06,
+      "loss": 0.7266,
+      "step": 3626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726263584269513e-06,
+      "loss": 0.7248,
+      "step": 3627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721568356588362e-06,
+      "loss": 0.8132,
+      "step": 3628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7168765385757677e-06,
+      "loss": 0.6979,
+      "step": 3629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712188132429671e-06,
+      "loss": 0.7617,
+      "step": 3630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7075031403464126e-06,
+      "loss": 0.7591,
+      "step": 3631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702821564520732e-06,
+      "loss": 0.7585,
+      "step": 3632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69814340714577e-06,
+      "loss": 0.8089,
+      "step": 3633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6934686704130698e-06,
+      "loss": 0.7861,
+      "step": 3634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68879735651256e-06,
+      "loss": 0.7533,
+      "step": 3635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684129467632579e-06,
+      "loss": 0.7103,
+      "step": 3636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.679465005959856e-06,
+      "loss": 0.7611,
+      "step": 3637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6748039736795086e-06,
+      "loss": 0.8122,
+      "step": 3638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6701463729750553e-06,
+      "loss": 0.7557,
+      "step": 3639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6654922060284074e-06,
+      "loss": 0.7513,
+      "step": 3640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.660841475019864e-06,
+      "loss": 0.7894,
+      "step": 3641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6561941821281145e-06,
+      "loss": 0.792,
+      "step": 3642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6515503295302446e-06,
+      "loss": 0.849,
+      "step": 3643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6469099194017144e-06,
+      "loss": 0.7679,
+      "step": 3644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642272953916384e-06,
+      "loss": 0.7539,
+      "step": 3645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.637639435246497e-06,
+      "loss": 0.7728,
+      "step": 3646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6330093655626777e-06,
+      "loss": 0.7819,
+      "step": 3647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6283827470339375e-06,
+      "loss": 0.7653,
+      "step": 3648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6237595818276727e-06,
+      "loss": 0.7425,
+      "step": 3649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6191398721096605e-06,
+      "loss": 0.8099,
+      "step": 3650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6145236200440595e-06,
+      "loss": 0.7708,
+      "step": 3651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 0.763,
+      "step": 3652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6053014975186265e-06,
+      "loss": 0.7809,
+      "step": 3653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6006956313790056e-06,
+      "loss": 0.8047,
+      "step": 3654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596093231532224e-06,
+      "loss": 0.7497,
+      "step": 3655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591494300134325e-06,
+      "loss": 0.7962,
+      "step": 3656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5868988393397376e-06,
+      "loss": 0.8109,
+      "step": 3657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5823068513012593e-06,
+      "loss": 0.7516,
+      "step": 3658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577718338170061e-06,
+      "loss": 0.7777,
+      "step": 3659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5731333020956883e-06,
+      "loss": 0.7643,
+      "step": 3660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.778,
+      "step": 3661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5639736697074525e-06,
+      "loss": 0.7381,
+      "step": 3662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559399077684526e-06,
+      "loss": 0.7878,
+      "step": 3663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5548279713003e-06,
+      "loss": 0.6499,
+      "step": 3664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55026035269617e-06,
+      "loss": 0.7471,
+      "step": 3665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545696224011884e-06,
+      "loss": 0.8065,
+      "step": 3666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5411355873855683e-06,
+      "loss": 0.7021,
+      "step": 3667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5365784449537033e-06,
+      "loss": 0.742,
+      "step": 3668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5320247988511402e-06,
+      "loss": 0.8187,
+      "step": 3669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527474651211089e-06,
+      "loss": 0.7721,
+      "step": 3670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.522928004165124e-06,
+      "loss": 0.7463,
+      "step": 3671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518384859843168e-06,
+      "loss": 0.7526,
+      "step": 3672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5138452203735175e-06,
+      "loss": 0.7656,
+      "step": 3673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509309087882823e-06,
+      "loss": 0.7451,
+      "step": 3674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504776464496084e-06,
+      "loss": 0.7995,
+      "step": 3675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5002473523366643e-06,
+      "loss": 0.7686,
+      "step": 3676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4957217535262824e-06,
+      "loss": 0.7236,
+      "step": 3677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4911996701850083e-06,
+      "loss": 0.7868,
+      "step": 3678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4866811044312667e-06,
+      "loss": 0.7207,
+      "step": 3679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.482166058381836e-06,
+      "loss": 0.7917,
+      "step": 3680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4776545341518387e-06,
+      "loss": 0.7484,
+      "step": 3681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4731465338547556e-06,
+      "loss": 0.7799,
+      "step": 3682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.468642059602415e-06,
+      "loss": 0.7757,
+      "step": 3683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4641411135049877e-06,
+      "loss": 0.7096,
+      "step": 3684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459643697670998e-06,
+      "loss": 0.7627,
+      "step": 3685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455149814207314e-06,
+      "loss": 0.762,
+      "step": 3686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4506594652191485e-06,
+      "loss": 0.7308,
+      "step": 3687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4461726528100615e-06,
+      "loss": 0.7414,
+      "step": 3688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.441689379081955e-06,
+      "loss": 0.8001,
+      "step": 3689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.437209646135068e-06,
+      "loss": 0.7979,
+      "step": 3690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4327334560679874e-06,
+      "loss": 0.7516,
+      "step": 3691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428260810977641e-06,
+      "loss": 0.8148,
+      "step": 3692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423791712959286e-06,
+      "loss": 0.7288,
+      "step": 3693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4193261641065304e-06,
+      "loss": 0.7741,
+      "step": 3694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4148641665113116e-06,
+      "loss": 0.708,
+      "step": 3695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4104057222639067e-06,
+      "loss": 0.7809,
+      "step": 3696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.405950833452928e-06,
+      "loss": 0.7868,
+      "step": 3697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401499502165324e-06,
+      "loss": 0.7601,
+      "step": 3698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3970517304863683e-06,
+      "loss": 0.7809,
+      "step": 3699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392607520499677e-06,
+      "loss": 0.7679,
+      "step": 3700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3881668742871943e-06,
+      "loss": 0.722,
+      "step": 3701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3837297939291893e-06,
+      "loss": 0.73,
+      "step": 3702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3792962815042686e-06,
+      "loss": 0.7943,
+      "step": 3703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3748663390893644e-06,
+      "loss": 0.75,
+      "step": 3704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3704399687597355e-06,
+      "loss": 0.7529,
+      "step": 3705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3660171725889703e-06,
+      "loss": 0.7858,
+      "step": 3706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3615979526489773e-06,
+      "loss": 0.7591,
+      "step": 3707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3571823110099988e-06,
+      "loss": 0.7461,
+      "step": 3708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352770249740588e-06,
+      "loss": 0.7262,
+      "step": 3709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3483617709076312e-06,
+      "loss": 0.7669,
+      "step": 3710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3439568765763354e-06,
+      "loss": 0.7389,
+      "step": 3711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.7539,
+      "step": 3712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.335157849671136e-06,
+      "loss": 0.7611,
+      "step": 3713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330763721219246e-06,
+      "loss": 0.7769,
+      "step": 3714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32637318551303e-06,
+      "loss": 0.8249,
+      "step": 3715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.32198624460929e-06,
+      "loss": 0.7487,
+      "step": 3716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317602900563143e-06,
+      "loss": 0.7747,
+      "step": 3717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3132231554280138e-06,
+      "loss": 0.6886,
+      "step": 3718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3088470112556483e-06,
+      "loss": 0.791,
+      "step": 3719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3044744700961065e-06,
+      "loss": 0.7109,
+      "step": 3720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300105533997753e-06,
+      "loss": 0.7484,
+      "step": 3721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2957402050072717e-06,
+      "loss": 0.778,
+      "step": 3722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2913784851696518e-06,
+      "loss": 0.7767,
+      "step": 3723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287020376528193e-06,
+      "loss": 0.7614,
+      "step": 3724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282665881124505e-06,
+      "loss": 0.7399,
+      "step": 3725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2783150009985054e-06,
+      "loss": 0.8044,
+      "step": 3726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2739677381884117e-06,
+      "loss": 0.7451,
+      "step": 3727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2696240947307525e-06,
+      "loss": 0.8252,
+      "step": 3728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265284072660362e-06,
+      "loss": 0.7738,
+      "step": 3729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260947674010372e-06,
+      "loss": 0.7334,
+      "step": 3730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256614900812222e-06,
+      "loss": 0.7507,
+      "step": 3731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.252285755095652e-06,
+      "loss": 0.8213,
+      "step": 3732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2479602388887013e-06,
+      "loss": 0.7744,
+      "step": 3733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2436383542177108e-06,
+      "loss": 0.7721,
+      "step": 3734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239320103107322e-06,
+      "loss": 0.8294,
+      "step": 3735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.235005487580466e-06,
+      "loss": 0.7432,
+      "step": 3736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2306945096583775e-06,
+      "loss": 0.7178,
+      "step": 3737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22638717136059e-06,
+      "loss": 0.7507,
+      "step": 3738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2220834747049214e-06,
+      "loss": 0.7391,
+      "step": 3739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2177834217074932e-06,
+      "loss": 0.779,
+      "step": 3740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2134870143827168e-06,
+      "loss": 0.7464,
+      "step": 3741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209194254743295e-06,
+      "loss": 0.7425,
+      "step": 3742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2049051448002245e-06,
+      "loss": 0.7829,
+      "step": 3743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2006196865627905e-06,
+      "loss": 0.8076,
+      "step": 3744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.196337882038563e-06,
+      "loss": 0.7734,
+      "step": 3745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192059733233408e-06,
+      "loss": 0.7793,
+      "step": 3746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1877852421514767e-06,
+      "loss": 0.805,
+      "step": 3747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183514410795202e-06,
+      "loss": 0.8031,
+      "step": 3748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1792472411653078e-06,
+      "loss": 0.8245,
+      "step": 3749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1749837352608005e-06,
+      "loss": 0.7445,
+      "step": 3750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170723895078972e-06,
+      "loss": 0.7266,
+      "step": 3751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166467722615394e-06,
+      "loss": 0.7676,
+      "step": 3752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1622152198639224e-06,
+      "loss": 0.7275,
+      "step": 3753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1579663888166956e-06,
+      "loss": 0.7855,
+      "step": 3754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153721231464123e-06,
+      "loss": 0.7546,
+      "step": 3755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1494797497949036e-06,
+      "loss": 0.7417,
+      "step": 3756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.145241945796014e-06,
+      "loss": 0.7358,
+      "step": 3757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1410078214526953e-06,
+      "loss": 0.7614,
+      "step": 3758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13677737874848e-06,
+      "loss": 0.7515,
+      "step": 3759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132550619665168e-06,
+      "loss": 0.8044,
+      "step": 3760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1283275461828344e-06,
+      "loss": 0.7738,
+      "step": 3761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.124108160279832e-06,
+      "loss": 0.7917,
+      "step": 3762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119892463932781e-06,
+      "loss": 0.763,
+      "step": 3763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1156804591165736e-06,
+      "loss": 0.8145,
+      "step": 3764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1114721478043733e-06,
+      "loss": 0.7884,
+      "step": 3765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1072675319676184e-06,
+      "loss": 0.7855,
+      "step": 3766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103066613576007e-06,
+      "loss": 0.7126,
+      "step": 3767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0988693945975102e-06,
+      "loss": 0.7223,
+      "step": 3768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0946758769983666e-06,
+      "loss": 0.7575,
+      "step": 3769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090486062743081e-06,
+      "loss": 0.805,
+      "step": 3770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0862999537944195e-06,
+      "loss": 0.8132,
+      "step": 3771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0821175521134208e-06,
+      "loss": 0.7848,
+      "step": 3772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077938859659373e-06,
+      "loss": 0.7321,
+      "step": 3773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0737638783898384e-06,
+      "loss": 0.7305,
+      "step": 3774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0695926102606402e-06,
+      "loss": 0.721,
+      "step": 3775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0654250572258528e-06,
+      "loss": 0.7259,
+      "step": 3776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06126122123782e-06,
+      "loss": 0.8141,
+      "step": 3777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05710110424714e-06,
+      "loss": 0.7523,
+      "step": 3778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052944708202669e-06,
+      "loss": 0.7471,
+      "step": 3779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048792035051521e-06,
+      "loss": 0.7673,
+      "step": 3780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04464308673907e-06,
+      "loss": 0.7629,
+      "step": 3781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0404978652089325e-06,
+      "loss": 0.7575,
+      "step": 3782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03635637240299e-06,
+      "loss": 0.7575,
+      "step": 3783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0322186102613793e-06,
+      "loss": 0.6842,
+      "step": 3784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.028084580722478e-06,
+      "loss": 0.7412,
+      "step": 3785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0239542857229245e-06,
+      "loss": 0.7913,
+      "step": 3786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019827727197605e-06,
+      "loss": 0.7321,
+      "step": 3787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0157049070796564e-06,
+      "loss": 0.7531,
+      "step": 3788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0115858273004607e-06,
+      "loss": 0.7368,
+      "step": 3789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0074704897896556e-06,
+      "loss": 0.8206,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0033588964751128e-06,
+      "loss": 0.8014,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999251049282962e-06,
+      "loss": 0.8034,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995146950137575e-06,
+      "loss": 0.776,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9910466009615604e-06,
+      "loss": 0.7414,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9869500036757805e-06,
+      "loss": 0.7995,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982857160199334e-06,
+      "loss": 0.7676,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9787680724495617e-06,
+      "loss": 0.7633,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9746827423420478e-06,
+      "loss": 0.7445,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970601171790616e-06,
+      "loss": 0.6844,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966523362707321e-06,
+      "loss": 0.8405,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962449317002465e-06,
+      "loss": 0.7734,
+      "step": 3801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9583790365845823e-06,
+      "loss": 0.7902,
+      "step": 3802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954312523360449e-06,
+      "loss": 0.7288,
+      "step": 3803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950249779235065e-06,
+      "loss": 0.7396,
+      "step": 3804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946190806111674e-06,
+      "loss": 0.7897,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9421356058917528e-06,
+      "loss": 0.7785,
+      "step": 3806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9380841804750063e-06,
+      "loss": 0.6859,
+      "step": 3807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9340365317593744e-06,
+      "loss": 0.7331,
+      "step": 3808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.929992661641029e-06,
+      "loss": 0.6953,
+      "step": 3809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9259525720143646e-06,
+      "loss": 0.7476,
+      "step": 3810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921916264772011e-06,
+      "loss": 0.8252,
+      "step": 3811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917883741804829e-06,
+      "loss": 0.7158,
+      "step": 3812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9138550050018967e-06,
+      "loss": 0.7983,
+      "step": 3813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.7002,
+      "step": 3814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9058088974362531e-06,
+      "loss": 0.7793,
+      "step": 3815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.901791530442838e-06,
+      "loss": 0.7467,
+      "step": 3816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8977779571522648e-06,
+      "loss": 0.7633,
+      "step": 3817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8937681794447404e-06,
+      "loss": 0.71,
+      "step": 3818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8897621991986914e-06,
+      "loss": 0.7718,
+      "step": 3819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8857600182907676e-06,
+      "loss": 0.819,
+      "step": 3820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8817616385958414e-06,
+      "loss": 0.8151,
+      "step": 3821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.877767061986997e-06,
+      "loss": 0.7992,
+      "step": 3822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.873776290335545e-06,
+      "loss": 0.7842,
+      "step": 3823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8697893255110088e-06,
+      "loss": 0.7231,
+      "step": 3824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8658061693811303e-06,
+      "loss": 0.7451,
+      "step": 3825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8618268238118674e-06,
+      "loss": 0.7337,
+      "step": 3826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857851290667394e-06,
+      "loss": 0.7676,
+      "step": 3827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.853879571810091e-06,
+      "loss": 0.716,
+      "step": 3828
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8499116691005614e-06,
+      "loss": 0.7461,
+      "step": 3829
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8459475843976193e-06,
+      "loss": 0.7614,
+      "step": 3830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8419873195582815e-06,
+      "loss": 0.7435,
+      "step": 3831
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8380308764377841e-06,
+      "loss": 0.6906,
+      "step": 3832
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.834078256889571e-06,
+      "loss": 0.7617,
+      "step": 3833
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8301294627652943e-06,
+      "loss": 0.7656,
+      "step": 3834
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8261844959148133e-06,
+      "loss": 0.7845,
+      "step": 3835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8222433581861986e-06,
+      "loss": 0.7581,
+      "step": 3836
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8183060514257167e-06,
+      "loss": 0.7845,
+      "step": 3837
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814372577477851e-06,
+      "loss": 0.7511,
+      "step": 3838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810442938185284e-06,
+      "loss": 0.7897,
+      "step": 3839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8065171353889e-06,
+      "loss": 0.7738,
+      "step": 3840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80259517092779e-06,
+      "loss": 0.765,
+      "step": 3841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7986770466392445e-06,
+      "loss": 0.7637,
+      "step": 3842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7947627643587574e-06,
+      "loss": 0.7819,
+      "step": 3843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7908523259200195e-06,
+      "loss": 0.7301,
+      "step": 3844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7869457331549257e-06,
+      "loss": 0.7913,
+      "step": 3845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7830429878935618e-06,
+      "loss": 0.8154,
+      "step": 3846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7791440919642178e-06,
+      "loss": 0.7633,
+      "step": 3847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7752490471933769e-06,
+      "loss": 0.8161,
+      "step": 3848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7713578554057243e-06,
+      "loss": 0.738,
+      "step": 3849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767470518424129e-06,
+      "loss": 0.7643,
+      "step": 3850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7635870380696629e-06,
+      "loss": 0.8535,
+      "step": 3851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.75970741616159e-06,
+      "loss": 0.7477,
+      "step": 3852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7558316545173648e-06,
+      "loss": 0.7334,
+      "step": 3853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7519597549526347e-06,
+      "loss": 0.7676,
+      "step": 3854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7480917192812397e-06,
+      "loss": 0.7314,
+      "step": 3855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7442275493152039e-06,
+      "loss": 0.723,
+      "step": 3856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7403672468647436e-06,
+      "loss": 0.7656,
+      "step": 3857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7365108137382692e-06,
+      "loss": 0.7544,
+      "step": 3858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7326582517423662e-06,
+      "loss": 0.7256,
+      "step": 3859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7288095626818169e-06,
+      "loss": 0.7189,
+      "step": 3860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7249647483595854e-06,
+      "loss": 0.8307,
+      "step": 3861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7211238105768213e-06,
+      "loss": 0.736,
+      "step": 3862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172867511328583e-06,
+      "loss": 0.7965,
+      "step": 3863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7134535718252142e-06,
+      "loss": 0.8014,
+      "step": 3864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709624274449584e-06,
+      "loss": 0.791,
+      "step": 3865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057988607998487e-06,
+      "loss": 0.8105,
+      "step": 3866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019773326680745e-06,
+      "loss": 0.75,
+      "step": 3867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6981596918444953e-06,
+      "loss": 0.7331,
+      "step": 3868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6943459401175322e-06,
+      "loss": 0.7663,
+      "step": 3869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6905360792737857e-06,
+      "loss": 0.7414,
+      "step": 3870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6867301110980283e-06,
+      "loss": 0.7367,
+      "step": 3871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6829280373732126e-06,
+      "loss": 0.7152,
+      "step": 3872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6791298598804685e-06,
+      "loss": 0.7523,
+      "step": 3873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6753355803990912e-06,
+      "loss": 0.7669,
+      "step": 3874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6715452007065602e-06,
+      "loss": 0.7306,
+      "step": 3875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6677587225785264e-06,
+      "loss": 0.7627,
+      "step": 3876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.663976147788806e-06,
+      "loss": 0.7624,
+      "step": 3877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6601974781093943e-06,
+      "loss": 0.7347,
+      "step": 3878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.656422715310454e-06,
+      "loss": 0.7832,
+      "step": 3879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6526518611603182e-06,
+      "loss": 0.7796,
+      "step": 3880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6488849174254884e-06,
+      "loss": 0.7689,
+      "step": 3881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.7669,
+      "step": 3882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6413627682585965e-06,
+      "loss": 0.7686,
+      "step": 3883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6376075663503732e-06,
+      "loss": 0.7575,
+      "step": 3884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6338562819051384e-06,
+      "loss": 0.7056,
+      "step": 3885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6301089166802232e-06,
+      "loss": 0.7493,
+      "step": 3886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626365472431125e-06,
+      "loss": 0.7087,
+      "step": 3887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6226259509115083e-06,
+      "loss": 0.7884,
+      "step": 3888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6188903538731947e-06,
+      "loss": 0.779,
+      "step": 3889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6151586830661704e-06,
+      "loss": 0.7839,
+      "step": 3890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611430940238583e-06,
+      "loss": 0.8076,
+      "step": 3891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607707127136734e-06,
+      "loss": 0.7383,
+      "step": 3892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6039872455050908e-06,
+      "loss": 0.7389,
+      "step": 3893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.600271297086279e-06,
+      "loss": 0.7425,
+      "step": 3894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596559283621074e-06,
+      "loss": 0.7424,
+      "step": 3895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5928512068484158e-06,
+      "loss": 0.7098,
+      "step": 3896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.589147068505398e-06,
+      "loss": 0.7549,
+      "step": 3897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5854468703272663e-06,
+      "loss": 0.8034,
+      "step": 3898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5817506140474248e-06,
+      "loss": 0.7474,
+      "step": 3899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5780583013974294e-06,
+      "loss": 0.7155,
+      "step": 3900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5743699341069895e-06,
+      "loss": 0.7852,
+      "step": 3901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.57068551390396e-06,
+      "loss": 0.7578,
+      "step": 3902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5670050425143557e-06,
+      "loss": 0.6751,
+      "step": 3903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5633285216623384e-06,
+      "loss": 0.6732,
+      "step": 3904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5596559530702138e-06,
+      "loss": 0.7516,
+      "step": 3905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5559873384584445e-06,
+      "loss": 0.7858,
+      "step": 3906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5523226795456349e-06,
+      "loss": 0.6908,
+      "step": 3907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.548661978048539e-06,
+      "loss": 0.7336,
+      "step": 3908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5450052356820577e-06,
+      "loss": 0.7533,
+      "step": 3909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5413524541592372e-06,
+      "loss": 0.7689,
+      "step": 3910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5377036351912612e-06,
+      "loss": 0.7653,
+      "step": 3911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5340587804874662e-06,
+      "loss": 0.8255,
+      "step": 3912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5304178917553302e-06,
+      "loss": 0.7313,
+      "step": 3913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5267809707004665e-06,
+      "loss": 0.7454,
+      "step": 3914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5231480190266367e-06,
+      "loss": 0.7095,
+      "step": 3915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 0.8239,
+      "step": 3916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.515894030627817e-06,
+      "loss": 0.734,
+      "step": 3917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5122729973010454e-06,
+      "loss": 0.7414,
+      "step": 3918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508655940151743e-06,
+      "loss": 0.6974,
+      "step": 3919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5050428608743606e-06,
+      "loss": 0.7461,
+      "step": 3920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5014337611614903e-06,
+      "loss": 0.7695,
+      "step": 3921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 0.7399,
+      "step": 3922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4942275071903268e-06,
+      "loss": 0.7591,
+      "step": 3923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906303563078871e-06,
+      "loss": 0.7303,
+      "step": 3924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4870371917416703e-06,
+      "loss": 0.7305,
+      "step": 3925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4834480151749365e-06,
+      "loss": 0.7861,
+      "step": 3926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4798628282890793e-06,
+      "loss": 0.776,
+      "step": 3927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4762816327636242e-06,
+      "loss": 0.742,
+      "step": 3928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727044302762218e-06,
+      "loss": 0.7386,
+      "step": 3929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4691312225026554e-06,
+      "loss": 0.7559,
+      "step": 3930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4655620111168422e-06,
+      "loss": 0.8122,
+      "step": 3931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4619967977908157e-06,
+      "loss": 0.6839,
+      "step": 3932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4584355841947452e-06,
+      "loss": 0.7653,
+      "step": 3933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454878371996924e-06,
+      "loss": 0.8118,
+      "step": 3934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4513251628637714e-06,
+      "loss": 0.7565,
+      "step": 3935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4477759584598294e-06,
+      "loss": 0.7445,
+      "step": 3936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.444230760447769e-06,
+      "loss": 0.7555,
+      "step": 3937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.440689570488376e-06,
+      "loss": 0.792,
+      "step": 3938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.437152390240566e-06,
+      "loss": 0.7435,
+      "step": 3939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4336192213613742e-06,
+      "loss": 0.7627,
+      "step": 3940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4300900655059524e-06,
+      "loss": 0.6842,
+      "step": 3941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4265649243275782e-06,
+      "loss": 0.7484,
+      "step": 3942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4230437994776458e-06,
+      "loss": 0.7044,
+      "step": 3943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4195266926056694e-06,
+      "loss": 0.7629,
+      "step": 3944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4160136053592788e-06,
+      "loss": 0.7568,
+      "step": 3945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4125045393842219e-06,
+      "loss": 0.7451,
+      "step": 3946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4089994963243658e-06,
+      "loss": 0.693,
+      "step": 3947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405498477821685e-06,
+      "loss": 0.7541,
+      "step": 3948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4020014855162755e-06,
+      "loss": 0.7526,
+      "step": 3949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3985085210463479e-06,
+      "loss": 0.7549,
+      "step": 3950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3950195860482185e-06,
+      "loss": 0.7933,
+      "step": 3951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3915346821563235e-06,
+      "loss": 0.8434,
+      "step": 3952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.388053811003207e-06,
+      "loss": 0.7412,
+      "step": 3953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.384576974219526e-06,
+      "loss": 0.7923,
+      "step": 3954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3811041734340447e-06,
+      "loss": 0.7511,
+      "step": 3955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3776354102736423e-06,
+      "loss": 0.848,
+      "step": 3956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3741706863632976e-06,
+      "loss": 0.7471,
+      "step": 3957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3707100033261035e-06,
+      "loss": 0.7848,
+      "step": 3958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3672533627832607e-06,
+      "loss": 0.7865,
+      "step": 3959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3638007663540697e-06,
+      "loss": 0.8226,
+      "step": 3960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3603522156559446e-06,
+      "loss": 0.7168,
+      "step": 3961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3569077123043973e-06,
+      "loss": 0.7702,
+      "step": 3962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3534672579130492e-06,
+      "loss": 0.751,
+      "step": 3963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3500308540936203e-06,
+      "loss": 0.7889,
+      "step": 3964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3465985024559392e-06,
+      "loss": 0.7562,
+      "step": 3965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3431702046079276e-06,
+      "loss": 0.7702,
+      "step": 3966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.7894,
+      "step": 3967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336325776703128e-06,
+      "loss": 0.7425,
+      "step": 3968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3329096498526917e-06,
+      "loss": 0.7321,
+      "step": 3969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3294975832046353e-06,
+      "loss": 0.7998,
+      "step": 3970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.326089578357379e-06,
+      "loss": 0.7201,
+      "step": 3971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.322685636907447e-06,
+      "loss": 0.79,
+      "step": 3972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3192857604494535e-06,
+      "loss": 0.7546,
+      "step": 3973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3158899505761147e-06,
+      "loss": 0.7982,
+      "step": 3974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3124982088782345e-06,
+      "loss": 0.7799,
+      "step": 3975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3091105369447166e-06,
+      "loss": 0.7731,
+      "step": 3976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305726936362559e-06,
+      "loss": 0.7617,
+      "step": 3977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3023474087168453e-06,
+      "loss": 0.7301,
+      "step": 3978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2989719555907587e-06,
+      "loss": 0.7985,
+      "step": 3979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2956005785655689e-06,
+      "loss": 0.7412,
+      "step": 3980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2922332792206404e-06,
+      "loss": 0.7939,
+      "step": 3981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2888700591334225e-06,
+      "loss": 0.7796,
+      "step": 3982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2855109198794603e-06,
+      "loss": 0.7858,
+      "step": 3983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.282155863032377e-06,
+      "loss": 0.7773,
+      "step": 3984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2788048901638917e-06,
+      "loss": 0.7529,
+      "step": 3985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2754580028438102e-06,
+      "loss": 0.834,
+      "step": 3986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2721152026400174e-06,
+      "loss": 0.8018,
+      "step": 3987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2687764911184908e-06,
+      "loss": 0.7461,
+      "step": 3988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2654418698432913e-06,
+      "loss": 0.7393,
+      "step": 3989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2621113403765561e-06,
+      "loss": 0.7816,
+      "step": 3990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2587849042785182e-06,
+      "loss": 0.7747,
+      "step": 3991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2554625631074846e-06,
+      "loss": 0.7555,
+      "step": 3992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2521443184198468e-06,
+      "loss": 0.7731,
+      "step": 3993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2488301717700735e-06,
+      "loss": 0.7297,
+      "step": 3994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2455201247107173e-06,
+      "loss": 0.7568,
+      "step": 3995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2422141787924136e-06,
+      "loss": 0.75,
+      "step": 3996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2389123355638655e-06,
+      "loss": 0.7995,
+      "step": 3997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2356145965718646e-06,
+      "loss": 0.7816,
+      "step": 3998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2323209633612765e-06,
+      "loss": 0.7171,
+      "step": 3999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2290314374750423e-06,
+      "loss": 0.7598,
+      "step": 4000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2257460204541793e-06,
+      "loss": 0.7161,
+      "step": 4001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2224647138377854e-06,
+      "loss": 0.7249,
+      "step": 4002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2191875191630209e-06,
+      "loss": 0.7591,
+      "step": 4003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.21591443796513e-06,
+      "loss": 0.8115,
+      "step": 4004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2126454717774294e-06,
+      "loss": 0.7441,
+      "step": 4005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209380622131301e-06,
+      "loss": 0.7194,
+      "step": 4006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2061198905562043e-06,
+      "loss": 0.7122,
+      "step": 4007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.202863278579669e-06,
+      "loss": 0.752,
+      "step": 4008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1996107877272944e-06,
+      "loss": 0.7601,
+      "step": 4009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1963624195227464e-06,
+      "loss": 0.6973,
+      "step": 4010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1931181754877673e-06,
+      "loss": 0.8034,
+      "step": 4011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1898780571421554e-06,
+      "loss": 0.7663,
+      "step": 4012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.186642066003786e-06,
+      "loss": 0.804,
+      "step": 4013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183410203588601e-06,
+      "loss": 0.7513,
+      "step": 4014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1801824714105992e-06,
+      "loss": 0.818,
+      "step": 4015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1769588709818535e-06,
+      "loss": 0.7137,
+      "step": 4016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1737394038124994e-06,
+      "loss": 0.7943,
+      "step": 4017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 0.7368,
+      "step": 4018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1673128752828123e-06,
+      "loss": 0.7272,
+      "step": 4019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1641058169330688e-06,
+      "loss": 0.7871,
+      "step": 4020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1609028978638804e-06,
+      "loss": 0.7982,
+      "step": 4021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1577041195756954e-06,
+      "loss": 0.7482,
+      "step": 4022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1545094835670224e-06,
+      "loss": 0.7451,
+      "step": 4023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1513189913344214e-06,
+      "loss": 0.717,
+      "step": 4024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481326443725193e-06,
+      "loss": 0.7461,
+      "step": 4025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1449504441740022e-06,
+      "loss": 0.7747,
+      "step": 4026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1417723922296008e-06,
+      "loss": 0.7796,
+      "step": 4027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.138598490028121e-06,
+      "loss": 0.7039,
+      "step": 4028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1354287390564134e-06,
+      "loss": 0.7507,
+      "step": 4029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132263140799381e-06,
+      "loss": 0.7839,
+      "step": 4030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291016967399904e-06,
+      "loss": 0.7907,
+      "step": 4031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1259444083592585e-06,
+      "loss": 0.7645,
+      "step": 4032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1227912771362525e-06,
+      "loss": 0.6826,
+      "step": 4033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1196423045480942e-06,
+      "loss": 0.8011,
+      "step": 4034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164974920699611e-06,
+      "loss": 0.7425,
+      "step": 4035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1133568411750729e-06,
+      "loss": 0.7718,
+      "step": 4036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1102203533347089e-06,
+      "loss": 0.737,
+      "step": 4037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1070880300181963e-06,
+      "loss": 0.7816,
+      "step": 4038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1039598726929046e-06,
+      "loss": 0.7044,
+      "step": 4039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1008358828242595e-06,
+      "loss": 0.8158,
+      "step": 4040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097716061875731e-06,
+      "loss": 0.7669,
+      "step": 4041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0946004113088381e-06,
+      "loss": 0.8223,
+      "step": 4042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0914889325831412e-06,
+      "loss": 0.7277,
+      "step": 4043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088381627156253e-06,
+      "loss": 0.7171,
+      "step": 4044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0852784964838247e-06,
+      "loss": 0.7627,
+      "step": 4045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0821795420195592e-06,
+      "loss": 0.776,
+      "step": 4046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.079084765215196e-06,
+      "loss": 0.763,
+      "step": 4047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0759941675205221e-06,
+      "loss": 0.7275,
+      "step": 4048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.072907750383363e-06,
+      "loss": 0.7666,
+      "step": 4049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0698255152495895e-06,
+      "loss": 0.8395,
+      "step": 4050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0667474635631126e-06,
+      "loss": 0.8102,
+      "step": 4051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.7969,
+      "step": 4052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0606039162978798e-06,
+      "loss": 0.8236,
+      "step": 4053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0575384235971463e-06,
+      "loss": 0.805,
+      "step": 4054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0544771200997394e-06,
+      "loss": 0.737,
+      "step": 4055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0514200072397695e-06,
+      "loss": 0.778,
+      "step": 4056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0483670864493777e-06,
+      "loss": 0.7799,
+      "step": 4057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045318359158737e-06,
+      "loss": 0.6673,
+      "step": 4058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042273826796062e-06,
+      "loss": 0.7492,
+      "step": 4059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0392334907876022e-06,
+      "loss": 0.7064,
+      "step": 4060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0361973525576374e-06,
+      "loss": 0.7508,
+      "step": 4061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033165413528483e-06,
+      "loss": 0.7295,
+      "step": 4062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0301376751204895e-06,
+      "loss": 0.7897,
+      "step": 4063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0271141387520322e-06,
+      "loss": 0.8294,
+      "step": 4064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0240948058395284e-06,
+      "loss": 0.7201,
+      "step": 4065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0210796777974196e-06,
+      "loss": 0.7051,
+      "step": 4066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0180687560381764e-06,
+      "loss": 0.7507,
+      "step": 4067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0150620419723022e-06,
+      "loss": 0.7933,
+      "step": 4068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012059537008332e-06,
+      "loss": 0.7419,
+      "step": 4069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.009061242552818e-06,
+      "loss": 0.8353,
+      "step": 4070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0060671600103523e-06,
+      "loss": 0.8128,
+      "step": 4071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0030772907835484e-06,
+      "loss": 0.7513,
+      "step": 4072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0000916362730429e-06,
+      "loss": 0.7142,
+      "step": 4073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.971101978775056e-07,
+      "loss": 0.834,
+      "step": 4074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94132976993627e-07,
+      "loss": 0.7269,
+      "step": 4075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.911599750161183e-07,
+      "loss": 0.7432,
+      "step": 4076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.881911933377197e-07,
+      "loss": 0.7832,
+      "step": 4077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.852266333491955e-07,
+      "loss": 0.752,
+      "step": 4078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.822662964393249e-07,
+      "loss": 0.7669,
+      "step": 4079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.793101839949148e-07,
+      "loss": 0.7881,
+      "step": 4080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.763582974007934e-07,
+      "loss": 0.6781,
+      "step": 4081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.734106380398022e-07,
+      "loss": 0.8161,
+      "step": 4082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.704672072928134e-07,
+      "loss": 0.7223,
+      "step": 4083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.675280065387117e-07,
+      "loss": 0.7585,
+      "step": 4084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.645930371543977e-07,
+      "loss": 0.7521,
+      "step": 4085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.616623005147952e-07,
+      "loss": 0.6961,
+      "step": 4086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.587357979928414e-07,
+      "loss": 0.7257,
+      "step": 4087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558135309594941e-07,
+      "loss": 0.75,
+      "step": 4088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.528955007837215e-07,
+      "loss": 0.7497,
+      "step": 4089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499817088325103e-07,
+      "loss": 0.7702,
+      "step": 4090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.470721564708651e-07,
+      "loss": 0.7461,
+      "step": 4091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.441668450617924e-07,
+      "loss": 0.7562,
+      "step": 4092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.412657759663279e-07,
+      "loss": 0.7334,
+      "step": 4093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.383689505435112e-07,
+      "loss": 0.7373,
+      "step": 4094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.35476370150391e-07,
+      "loss": 0.8044,
+      "step": 4095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.325880361420336e-07,
+      "loss": 0.6958,
+      "step": 4096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.297039498715155e-07,
+      "loss": 0.7295,
+      "step": 4097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.26824112689918e-07,
+      "loss": 0.7995,
+      "step": 4098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.239485259463366e-07,
+      "loss": 0.7826,
+      "step": 4099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.210771909878769e-07,
+      "loss": 0.8398,
+      "step": 4100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.18210109159644e-07,
+      "loss": 0.8014,
+      "step": 4101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.153472818047627e-07,
+      "loss": 0.7404,
+      "step": 4102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.124887102643576e-07,
+      "loss": 0.7786,
+      "step": 4103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.0963439587756e-07,
+      "loss": 0.7581,
+      "step": 4104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.067843399815068e-07,
+      "loss": 0.7835,
+      "step": 4105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.039385439113435e-07,
+      "loss": 0.6784,
+      "step": 4106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.010970090002135e-07,
+      "loss": 0.776,
+      "step": 4107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.982597365792711e-07,
+      "loss": 0.7516,
+      "step": 4108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.954267279776707e-07,
+      "loss": 0.7998,
+      "step": 4109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.925979845225641e-07,
+      "loss": 0.7523,
+      "step": 4110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.897735075391156e-07,
+      "loss": 0.7275,
+      "step": 4111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.869532983504859e-07,
+      "loss": 0.696,
+      "step": 4112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.841373582778323e-07,
+      "loss": 0.7104,
+      "step": 4113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.813256886403165e-07,
+      "loss": 0.7464,
+      "step": 4114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.785182907551004e-07,
+      "loss": 0.7988,
+      "step": 4115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.757151659373408e-07,
+      "loss": 0.7679,
+      "step": 4116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.729163155001975e-07,
+      "loss": 0.7633,
+      "step": 4117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.701217407548245e-07,
+      "loss": 0.7607,
+      "step": 4118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.673314430103707e-07,
+      "loss": 0.7298,
+      "step": 4119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.8083,
+      "step": 4120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617636837508259e-07,
+      "loss": 0.8014,
+      "step": 4121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.58986224844014e-07,
+      "loss": 0.7848,
+      "step": 4122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.562130481546904e-07,
+      "loss": 0.7256,
+      "step": 4123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.534441549819849e-07,
+      "loss": 0.7611,
+      "step": 4124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.506795466230156e-07,
+      "loss": 0.7699,
+      "step": 4125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.479192243728962e-07,
+      "loss": 0.7842,
+      "step": 4126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.45163189524737e-07,
+      "loss": 0.7438,
+      "step": 4127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.424114433696296e-07,
+      "loss": 0.7975,
+      "step": 4128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.39663987196665e-07,
+      "loss": 0.7249,
+      "step": 4129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.369208222929248e-07,
+      "loss": 0.7402,
+      "step": 4130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.341819499434745e-07,
+      "loss": 0.7487,
+      "step": 4131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.31447371431372e-07,
+      "loss": 0.7611,
+      "step": 4132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.28717088037665e-07,
+      "loss": 0.7656,
+      "step": 4133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259911010413846e-07,
+      "loss": 0.7456,
+      "step": 4134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.232694117195539e-07,
+      "loss": 0.6494,
+      "step": 4135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.205520213471808e-07,
+      "loss": 0.7288,
+      "step": 4136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.178389311972612e-07,
+      "loss": 0.7194,
+      "step": 4137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151301425407699e-07,
+      "loss": 0.82,
+      "step": 4138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124256566466749e-07,
+      "loss": 0.7725,
+      "step": 4139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.097254747819272e-07,
+      "loss": 0.778,
+      "step": 4140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.07029598211454e-07,
+      "loss": 0.7943,
+      "step": 4141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.043380281981739e-07,
+      "loss": 0.7487,
+      "step": 4142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.016507660029871e-07,
+      "loss": 0.7324,
+      "step": 4143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.9896781288477e-07,
+      "loss": 0.7611,
+      "step": 4144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.962891701003849e-07,
+      "loss": 0.7653,
+      "step": 4145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.936148389046772e-07,
+      "loss": 0.7448,
+      "step": 4146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.909448205504633e-07,
+      "loss": 0.751,
+      "step": 4147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.882791162885505e-07,
+      "loss": 0.7868,
+      "step": 4148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.856177273677212e-07,
+      "loss": 0.7432,
+      "step": 4149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.829606550347313e-07,
+      "loss": 0.7943,
+      "step": 4150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.803079005343173e-07,
+      "loss": 0.7557,
+      "step": 4151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.776594651091995e-07,
+      "loss": 0.7992,
+      "step": 4152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.750153500000623e-07,
+      "loss": 0.7285,
+      "step": 4153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.723755564455771e-07,
+      "loss": 0.7181,
+      "step": 4154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.697400856823866e-07,
+      "loss": 0.8125,
+      "step": 4155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.671089389451059e-07,
+      "loss": 0.7318,
+      "step": 4156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.644821174663308e-07,
+      "loss": 0.8307,
+      "step": 4157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618596224766283e-07,
+      "loss": 0.8066,
+      "step": 4158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.592414552045324e-07,
+      "loss": 0.7393,
+      "step": 4159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.566276168765585e-07,
+      "loss": 0.8473,
+      "step": 4160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.540181087171916e-07,
+      "loss": 0.7552,
+      "step": 4161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.514129319488839e-07,
+      "loss": 0.7471,
+      "step": 4162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.488120877920624e-07,
+      "loss": 0.7819,
+      "step": 4163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.462155774651258e-07,
+      "loss": 0.7637,
+      "step": 4164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.43623402184438e-07,
+      "loss": 0.7884,
+      "step": 4165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.41035563164334e-07,
+      "loss": 0.7972,
+      "step": 4166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.384520616171232e-07,
+      "loss": 0.7326,
+      "step": 4167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.358728987530728e-07,
+      "loss": 0.6922,
+      "step": 4168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.332980757804242e-07,
+      "loss": 0.7132,
+      "step": 4169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.30727593905386e-07,
+      "loss": 0.7197,
+      "step": 4170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.281614543321269e-07,
+      "loss": 0.7402,
+      "step": 4171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.255996582627878e-07,
+      "loss": 0.7949,
+      "step": 4172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230422068974763e-07,
+      "loss": 0.7445,
+      "step": 4173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.204891014342552e-07,
+      "loss": 0.7734,
+      "step": 4174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179403430691589e-07,
+      "loss": 0.7253,
+      "step": 4175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.15395932996188e-07,
+      "loss": 0.7471,
+      "step": 4176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.128558724072976e-07,
+      "loss": 0.7288,
+      "step": 4177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103201624924105e-07,
+      "loss": 0.7754,
+      "step": 4178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.077888044394121e-07,
+      "loss": 0.7324,
+      "step": 4179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.052617994341449e-07,
+      "loss": 0.8262,
+      "step": 4180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.027391486604162e-07,
+      "loss": 0.7233,
+      "step": 4181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.002208532999933e-07,
+      "loss": 0.6925,
+      "step": 4182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.977069145325987e-07,
+      "loss": 0.7874,
+      "step": 4183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.951973335359164e-07,
+      "loss": 0.7262,
+      "step": 4184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926921114855945e-07,
+      "loss": 0.7375,
+      "step": 4185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.901912495552332e-07,
+      "loss": 0.8009,
+      "step": 4186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.876947489163877e-07,
+      "loss": 0.7591,
+      "step": 4187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.852026107385756e-07,
+      "loss": 0.7699,
+      "step": 4188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.827148361892721e-07,
+      "loss": 0.765,
+      "step": 4189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.802314264338994e-07,
+      "loss": 0.7744,
+      "step": 4190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.777523826358434e-07,
+      "loss": 0.7829,
+      "step": 4191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 0.8083,
+      "step": 4192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.728073975549865e-07,
+      "loss": 0.7747,
+      "step": 4193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.703414585887224e-07,
+      "loss": 0.7643,
+      "step": 4194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.678798902128513e-07,
+      "loss": 0.7581,
+      "step": 4195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.654226935805197e-07,
+      "loss": 0.7612,
+      "step": 4196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629698698428333e-07,
+      "loss": 0.7848,
+      "step": 4197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.605214201488485e-07,
+      "loss": 0.7643,
+      "step": 4198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.580773456455669e-07,
+      "loss": 0.7865,
+      "step": 4199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.556376474779469e-07,
+      "loss": 0.7474,
+      "step": 4200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.532023267888954e-07,
+      "loss": 0.7355,
+      "step": 4201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.507713847192643e-07,
+      "loss": 0.7409,
+      "step": 4202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.483448224078581e-07,
+      "loss": 0.7969,
+      "step": 4203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.459226409914332e-07,
+      "loss": 0.7881,
+      "step": 4204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.435048416046863e-07,
+      "loss": 0.7438,
+      "step": 4205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.410914253802636e-07,
+      "loss": 0.7354,
+      "step": 4206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386823934487619e-07,
+      "loss": 0.6839,
+      "step": 4207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.362777469387182e-07,
+      "loss": 0.7796,
+      "step": 4208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.338774869766184e-07,
+      "loss": 0.7594,
+      "step": 4209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.314816146868951e-07,
+      "loss": 0.7611,
+      "step": 4210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.290901311919217e-07,
+      "loss": 0.7477,
+      "step": 4211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.267030376120154e-07,
+      "loss": 0.7712,
+      "step": 4212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.243203350654436e-07,
+      "loss": 0.8356,
+      "step": 4213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.219420246684094e-07,
+      "loss": 0.737,
+      "step": 4214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.195681075350591e-07,
+      "loss": 0.7466,
+      "step": 4215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.171985847774864e-07,
+      "loss": 0.7428,
+      "step": 4216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.148334575057191e-07,
+      "loss": 0.8115,
+      "step": 4217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.124727268277309e-07,
+      "loss": 0.7376,
+      "step": 4218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.101163938494359e-07,
+      "loss": 0.7933,
+      "step": 4219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.077644596746834e-07,
+      "loss": 0.7917,
+      "step": 4220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054169254052655e-07,
+      "loss": 0.7721,
+      "step": 4221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.772,
+      "step": 4222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.007350609793016e-07,
+      "loss": 0.7389,
+      "step": 4223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.984007330160291e-07,
+      "loss": 0.7523,
+      "step": 4224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.960708093446432e-07,
+      "loss": 0.752,
+      "step": 4225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.937452910566221e-07,
+      "loss": 0.722,
+      "step": 4226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.914241792413855e-07,
+      "loss": 0.7952,
+      "step": 4227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891074749862857e-07,
+      "loss": 0.7049,
+      "step": 4228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.867951793766091e-07,
+      "loss": 0.7887,
+      "step": 4229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844872934955781e-07,
+      "loss": 0.7738,
+      "step": 4230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8218381842435e-07,
+      "loss": 0.7728,
+      "step": 4231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.798847552420184e-07,
+      "loss": 0.8031,
+      "step": 4232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775901050256027e-07,
+      "loss": 0.7865,
+      "step": 4233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.752998688500611e-07,
+      "loss": 0.7542,
+      "step": 4234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.730140477882828e-07,
+      "loss": 0.8037,
+      "step": 4235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.707326429110871e-07,
+      "loss": 0.7581,
+      "step": 4236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.684556552872256e-07,
+      "loss": 0.7419,
+      "step": 4237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.661830859833817e-07,
+      "loss": 0.7633,
+      "step": 4238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63914936064165e-07,
+      "loss": 0.7474,
+      "step": 4239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.616512065921187e-07,
+      "loss": 0.7484,
+      "step": 4240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.593918986277169e-07,
+      "loss": 0.7396,
+      "step": 4241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.571370132293552e-07,
+      "loss": 0.7754,
+      "step": 4242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.548865514533641e-07,
+      "loss": 0.779,
+      "step": 4243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.526405143539992e-07,
+      "loss": 0.8164,
+      "step": 4244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.503989029834434e-07,
+      "loss": 0.6763,
+      "step": 4245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.481617183918053e-07,
+      "loss": 0.703,
+      "step": 4246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.459289616271224e-07,
+      "loss": 0.7692,
+      "step": 4247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437006337353556e-07,
+      "loss": 0.7383,
+      "step": 4248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.414767357603901e-07,
+      "loss": 0.723,
+      "step": 4249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.392572687440423e-07,
+      "loss": 0.8102,
+      "step": 4250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.370422337260428e-07,
+      "loss": 0.7549,
+      "step": 4251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.348316317440549e-07,
+      "loss": 0.7666,
+      "step": 4252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.326254638336614e-07,
+      "loss": 0.778,
+      "step": 4253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.304237310283655e-07,
+      "loss": 0.765,
+      "step": 4254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282264343595977e-07,
+      "loss": 0.7616,
+      "step": 4255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.26033574856708e-07,
+      "loss": 0.7402,
+      "step": 4256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.238451535469658e-07,
+      "loss": 0.7695,
+      "step": 4257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.216611714555631e-07,
+      "loss": 0.7933,
+      "step": 4258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.194816296056171e-07,
+      "loss": 0.7357,
+      "step": 4259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.173065290181544e-07,
+      "loss": 0.7692,
+      "step": 4260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.151358707121301e-07,
+      "loss": 0.766,
+      "step": 4261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.129696557044173e-07,
+      "loss": 0.7627,
+      "step": 4262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.108078850097997e-07,
+      "loss": 0.7523,
+      "step": 4263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086505596409885e-07,
+      "loss": 0.765,
+      "step": 4264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.064976806086086e-07,
+      "loss": 0.8005,
+      "step": 4265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.043492489211999e-07,
+      "loss": 0.821,
+      "step": 4266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022052655852228e-07,
+      "loss": 0.7552,
+      "step": 4267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.000657316050505e-07,
+      "loss": 0.7679,
+      "step": 4268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.979306479829759e-07,
+      "loss": 0.8298,
+      "step": 4269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958000157192023e-07,
+      "loss": 0.6673,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.936738358118509e-07,
+      "loss": 0.7611,
+      "step": 4271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.915521092569553e-07,
+      "loss": 0.7471,
+      "step": 4272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.8125,
+      "step": 4273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.873220201782414e-07,
+      "loss": 0.8122,
+      "step": 4274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852136596360569e-07,
+      "loss": 0.7982,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.831097564095999e-07,
+      "loss": 0.7988,
+      "step": 4276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.810103114844688e-07,
+      "loss": 0.792,
+      "step": 4277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.789153258441737e-07,
+      "loss": 0.7497,
+      "step": 4278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.768248004701359e-07,
+      "loss": 0.7415,
+      "step": 4279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.747387363416878e-07,
+      "loss": 0.7477,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7265713443607086e-07,
+      "loss": 0.7568,
+      "step": 4281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7057999572843516e-07,
+      "loss": 0.7113,
+      "step": 4282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.685073211918412e-07,
+      "loss": 0.7806,
+      "step": 4283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6643911179726107e-07,
+      "loss": 0.7773,
+      "step": 4284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.643753685135688e-07,
+      "loss": 0.8392,
+      "step": 4285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.623160923075498e-07,
+      "loss": 0.8184,
+      "step": 4286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.602612841439014e-07,
+      "loss": 0.8216,
+      "step": 4287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.582109449852168e-07,
+      "loss": 0.7327,
+      "step": 4288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5616507579200663e-07,
+      "loss": 0.7607,
+      "step": 4289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5412367752268094e-07,
+      "loss": 0.7881,
+      "step": 4290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.520867511335569e-07,
+      "loss": 0.7148,
+      "step": 4291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.50054297578858e-07,
+      "loss": 0.7536,
+      "step": 4292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.480263178107114e-07,
+      "loss": 0.7692,
+      "step": 4293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4600281277914715e-07,
+      "loss": 0.7607,
+      "step": 4294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4398378343210035e-07,
+      "loss": 0.7246,
+      "step": 4295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4196923071541334e-07,
+      "loss": 0.8158,
+      "step": 4296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.399591555728233e-07,
+      "loss": 0.7144,
+      "step": 4297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3795355894597494e-07,
+      "loss": 0.8114,
+      "step": 4298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3595244177441654e-07,
+      "loss": 0.7503,
+      "step": 4299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3395580499559276e-07,
+      "loss": 0.765,
+      "step": 4300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3196364954485205e-07,
+      "loss": 0.8678,
+      "step": 4301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2997597635544563e-07,
+      "loss": 0.7467,
+      "step": 4302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2799278635851983e-07,
+      "loss": 0.7454,
+      "step": 4303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2601408048312585e-07,
+      "loss": 0.7604,
+      "step": 4304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.240398596562123e-07,
+      "loss": 0.7402,
+      "step": 4305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2207012480262486e-07,
+      "loss": 0.7284,
+      "step": 4306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2010487684511105e-07,
+      "loss": 0.7168,
+      "step": 4307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.181441167043154e-07,
+      "loss": 0.7575,
+      "step": 4308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.161878452987778e-07,
+      "loss": 0.7842,
+      "step": 4309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1423606354493716e-07,
+      "loss": 0.7578,
+      "step": 4310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1228877235713115e-07,
+      "loss": 0.7734,
+      "step": 4311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103459726475889e-07,
+      "loss": 0.7126,
+      "step": 4312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.084076653264402e-07,
+      "loss": 0.7699,
+      "step": 4313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0647385130170637e-07,
+      "loss": 0.7526,
+      "step": 4314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0454453147930727e-07,
+      "loss": 0.7342,
+      "step": 4315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.026197067630555e-07,
+      "loss": 0.7311,
+      "step": 4316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0069937805466084e-07,
+      "loss": 0.7471,
+      "step": 4317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9878354625371927e-07,
+      "loss": 0.7611,
+      "step": 4318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.968722122577284e-07,
+      "loss": 0.7466,
+      "step": 4319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949653769620765e-07,
+      "loss": 0.756,
+      "step": 4320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9306304126004004e-07,
+      "loss": 0.7988,
+      "step": 4321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9116520604279285e-07,
+      "loss": 0.7721,
+      "step": 4322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8927187219939823e-07,
+      "loss": 0.7721,
+      "step": 4323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 0.7598,
+      "step": 4324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8549871217987924e-07,
+      "loss": 0.8509,
+      "step": 4325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.836188877713354e-07,
+      "loss": 0.7686,
+      "step": 4326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.817435682718096e-07,
+      "loss": 0.8145,
+      "step": 4327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7987275455981467e-07,
+      "loss": 0.7266,
+      "step": 4328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.780064475117573e-07,
+      "loss": 0.6945,
+      "step": 4329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.761446480019315e-07,
+      "loss": 0.7451,
+      "step": 4330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.742873569025185e-07,
+      "loss": 0.7881,
+      "step": 4331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7243457508358784e-07,
+      "loss": 0.7617,
+      "step": 4332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7058630341310074e-07,
+      "loss": 0.7721,
+      "step": 4333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6874254275689913e-07,
+      "loss": 0.7601,
+      "step": 4334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.669032939787154e-07,
+      "loss": 0.7965,
+      "step": 4335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.650685579401692e-07,
+      "loss": 0.7235,
+      "step": 4336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.632383355007629e-07,
+      "loss": 0.748,
+      "step": 4337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6141262751788643e-07,
+      "loss": 0.6865,
+      "step": 4338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.595914348468166e-07,
+      "loss": 0.7946,
+      "step": 4339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5777475834070985e-07,
+      "loss": 0.7585,
+      "step": 4340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.55962598850611e-07,
+      "loss": 0.7578,
+      "step": 4341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.541549572254488e-07,
+      "loss": 0.7207,
+      "step": 4342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.523518343120347e-07,
+      "loss": 0.7816,
+      "step": 4343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505532309550619e-07,
+      "loss": 0.7155,
+      "step": 4344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4875914799710975e-07,
+      "loss": 0.7178,
+      "step": 4345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4696958627863596e-07,
+      "loss": 0.7694,
+      "step": 4346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451845466379833e-07,
+      "loss": 0.7386,
+      "step": 4347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.43404029911375e-07,
+      "loss": 0.6842,
+      "step": 4348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4162803693291393e-07,
+      "loss": 0.7344,
+      "step": 4349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.398565685345878e-07,
+      "loss": 0.7285,
+      "step": 4350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380896255462596e-07,
+      "loss": 0.7611,
+      "step": 4351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3632720879567594e-07,
+      "loss": 0.8285,
+      "step": 4352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3456931910846313e-07,
+      "loss": 0.7865,
+      "step": 4353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328159573081258e-07,
+      "loss": 0.7715,
+      "step": 4354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.310671242160457e-07,
+      "loss": 0.7236,
+      "step": 4355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2932282065148533e-07,
+      "loss": 0.7476,
+      "step": 4356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2758304743158554e-07,
+      "loss": 0.7363,
+      "step": 4357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2584780537136206e-07,
+      "loss": 0.7949,
+      "step": 4358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2411709528371026e-07,
+      "loss": 0.7624,
+      "step": 4359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223909179794027e-07,
+      "loss": 0.7796,
+      "step": 4360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2066927426708806e-07,
+      "loss": 0.7181,
+      "step": 4361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1895216495329116e-07,
+      "loss": 0.7939,
+      "step": 4362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1723959084241195e-07,
+      "loss": 0.7489,
+      "step": 4363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.155315527367264e-07,
+      "loss": 0.8389,
+      "step": 4364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.138280514363845e-07,
+      "loss": 0.7624,
+      "step": 4365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1212908773941344e-07,
+      "loss": 0.7214,
+      "step": 4366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1043466244171204e-07,
+      "loss": 0.7575,
+      "step": 4367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.087447763370544e-07,
+      "loss": 0.7734,
+      "step": 4368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.070594302170871e-07,
+      "loss": 0.7786,
+      "step": 4369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.053786248713331e-07,
+      "loss": 0.7031,
+      "step": 4370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0370236108718366e-07,
+      "loss": 0.751,
+      "step": 4371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.020306396499062e-07,
+      "loss": 0.79,
+      "step": 4372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.003634613426365e-07,
+      "loss": 0.7533,
+      "step": 4373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9870082694638644e-07,
+      "loss": 0.766,
+      "step": 4374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.970427372400353e-07,
+      "loss": 0.7695,
+      "step": 4375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.953891930003372e-07,
+      "loss": 0.7542,
+      "step": 4376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9374019500191255e-07,
+      "loss": 0.7799,
+      "step": 4377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.920957440172556e-07,
+      "loss": 0.7998,
+      "step": 4378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9045584081672793e-07,
+      "loss": 0.7051,
+      "step": 4379
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.888204861685628e-07,
+      "loss": 0.7438,
+      "step": 4380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.871896808388608e-07,
+      "loss": 0.7279,
+      "step": 4381
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8556342559159513e-07,
+      "loss": 0.7497,
+      "step": 4382
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8394172118860087e-07,
+      "loss": 0.7663,
+      "step": 4383
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.82324568389587e-07,
+      "loss": 0.7098,
+      "step": 4384
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.807119679521275e-07,
+      "loss": 0.7689,
+      "step": 4385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.791039206316637e-07,
+      "loss": 0.7184,
+      "step": 4386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7750042718150514e-07,
+      "loss": 0.776,
+      "step": 4387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.759014883528288e-07,
+      "loss": 0.8034,
+      "step": 4388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7430710489467437e-07,
+      "loss": 0.8112,
+      "step": 4389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727172775539522e-07,
+      "loss": 0.7624,
+      "step": 4390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.711320070754353e-07,
+      "loss": 0.7956,
+      "step": 4391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.7614,
+      "step": 4392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6797513967343645e-07,
+      "loss": 0.7093,
+      "step": 4393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6640354422882706e-07,
+      "loss": 0.6991,
+      "step": 4394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.648365086041671e-07,
+      "loss": 0.7546,
+      "step": 4395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6327403353355264e-07,
+      "loss": 0.7643,
+      "step": 4396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.617161197489426e-07,
+      "loss": 0.7057,
+      "step": 4397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601627679801633e-07,
+      "loss": 0.8112,
+      "step": 4398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5861397895489914e-07,
+      "loss": 0.7393,
+      "step": 4399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.570697533986999e-07,
+      "loss": 0.7194,
+      "step": 4400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.555300920349746e-07,
+      "loss": 0.7051,
+      "step": 4401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.539949955849985e-07,
+      "loss": 0.7725,
+      "step": 4402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.524644647679042e-07,
+      "loss": 0.7461,
+      "step": 4403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5093850030068813e-07,
+      "loss": 0.7936,
+      "step": 4404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.494171028982062e-07,
+      "loss": 0.7572,
+      "step": 4405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.47900273273175e-07,
+      "loss": 0.7643,
+      "step": 4406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.463880121361717e-07,
+      "loss": 0.7373,
+      "step": 4407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.44880320195634e-07,
+      "loss": 0.7363,
+      "step": 4408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.433771981578581e-07,
+      "loss": 0.806,
+      "step": 4409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4187864672699843e-07,
+      "loss": 0.7254,
+      "step": 4410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.403846666050691e-07,
+      "loss": 0.7676,
+      "step": 4411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3889525849194573e-07,
+      "loss": 0.7051,
+      "step": 4412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.374104230853569e-07,
+      "loss": 0.7938,
+      "step": 4413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3593016108089172e-07,
+      "loss": 0.8027,
+      "step": 4414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3445447317199775e-07,
+      "loss": 0.8203,
+      "step": 4415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3298336004997756e-07,
+      "loss": 0.6829,
+      "step": 4416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.315168224039932e-07,
+      "loss": 0.7542,
+      "step": 4417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3005486092106066e-07,
+      "loss": 0.7383,
+      "step": 4418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2859747628605322e-07,
+      "loss": 0.7835,
+      "step": 4419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271446691817014e-07,
+      "loss": 0.7454,
+      "step": 4420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2569644028858862e-07,
+      "loss": 0.7487,
+      "step": 4421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2425279028515658e-07,
+      "loss": 0.7598,
+      "step": 4422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2281371984769762e-07,
+      "loss": 0.7786,
+      "step": 4423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2137922965036473e-07,
+      "loss": 0.7606,
+      "step": 4424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1994932036516038e-07,
+      "loss": 0.7627,
+      "step": 4425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.7998,
+      "step": 4426
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1710324720842556e-07,
+      "loss": 0.7083,
+      "step": 4427
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1568708467017197e-07,
+      "loss": 0.7445,
+      "step": 4428
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1427550571060051e-07,
+      "loss": 0.7357,
+      "step": 4429
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1286851099098339e-07,
+      "loss": 0.7205,
+      "step": 4430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.114661011704433e-07,
+      "loss": 0.7464,
+      "step": 4431
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1006827690595478e-07,
+      "loss": 0.6776,
+      "step": 4432
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.086750388523473e-07,
+      "loss": 0.7542,
+      "step": 4433
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0728638766229993e-07,
+      "loss": 0.682,
+      "step": 4434
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0590232398634114e-07,
+      "loss": 0.7539,
+      "step": 4435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0452284847285343e-07,
+      "loss": 0.7713,
+      "step": 4436
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0314796176806984e-07,
+      "loss": 0.7712,
+      "step": 4437
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0177766451607073e-07,
+      "loss": 0.7764,
+      "step": 4438
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0041195735878816e-07,
+      "loss": 0.7236,
+      "step": 4439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.990508409360048e-07,
+      "loss": 0.7546,
+      "step": 4440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.976943158853517e-07,
+      "loss": 0.7529,
+      "step": 4441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9634238284230945e-07,
+      "loss": 0.7629,
+      "step": 4442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9499504244020694e-07,
+      "loss": 0.7331,
+      "step": 4443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9365229531022267e-07,
+      "loss": 0.7728,
+      "step": 4444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9231414208138123e-07,
+      "loss": 0.7653,
+      "step": 4445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.909805833805589e-07,
+      "loss": 0.7777,
+      "step": 4446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8965161983247494e-07,
+      "loss": 0.7085,
+      "step": 4447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8832725205969904e-07,
+      "loss": 0.8008,
+      "step": 4448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8700748068264828e-07,
+      "loss": 0.8333,
+      "step": 4449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8569230631958258e-07,
+      "loss": 0.7227,
+      "step": 4450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.843817295866146e-07,
+      "loss": 0.7445,
+      "step": 4451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.830757510976966e-07,
+      "loss": 0.7546,
+      "step": 4452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.817743714646314e-07,
+      "loss": 0.7855,
+      "step": 4453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8047759129706577e-07,
+      "loss": 0.7176,
+      "step": 4454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7918541120249377e-07,
+      "loss": 0.7598,
+      "step": 4455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7789783178624898e-07,
+      "loss": 0.8034,
+      "step": 4456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7661485365151553e-07,
+      "loss": 0.7673,
+      "step": 4457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7533647739932046e-07,
+      "loss": 0.791,
+      "step": 4458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.740627036285325e-07,
+      "loss": 0.7677,
+      "step": 4459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7279353293586765e-07,
+      "loss": 0.7785,
+      "step": 4460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7152896591588363e-07,
+      "loss": 0.7601,
+      "step": 4461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 0.7676,
+      "step": 4462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6901364526140884e-07,
+      "loss": 0.7523,
+      "step": 4463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6776289280524992e-07,
+      "loss": 0.7168,
+      "step": 4464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6651674637843562e-07,
+      "loss": 0.779,
+      "step": 4465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6527520656473784e-07,
+      "loss": 0.686,
+      "step": 4466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6403827394577244e-07,
+      "loss": 0.7686,
+      "step": 4467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6280594910099257e-07,
+      "loss": 0.7673,
+      "step": 4468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6157823260769755e-07,
+      "loss": 0.6955,
+      "step": 4469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035512504102624e-07,
+      "loss": 0.8122,
+      "step": 4470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5913662697395583e-07,
+      "loss": 0.7559,
+      "step": 4471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5792273897730858e-07,
+      "loss": 0.7738,
+      "step": 4472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.567134616197441e-07,
+      "loss": 0.7484,
+      "step": 4473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5550879546776364e-07,
+      "loss": 0.7803,
+      "step": 4474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.543087410857058e-07,
+      "loss": 0.7826,
+      "step": 4475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.531132990357509e-07,
+      "loss": 0.8245,
+      "step": 4476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.7428,
+      "step": 4477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5073625417006855e-07,
+      "loss": 0.736,
+      "step": 4478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4955465246789592e-07,
+      "loss": 0.7575,
+      "step": 4479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4837766532493469e-07,
+      "loss": 0.7327,
+      "step": 4480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.472052932925616e-07,
+      "loss": 0.7409,
+      "step": 4481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4603753691998735e-07,
+      "loss": 0.7445,
+      "step": 4482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4487439675426096e-07,
+      "loss": 0.7676,
+      "step": 4483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4371587334026992e-07,
+      "loss": 0.7713,
+      "step": 4484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.425619672207379e-07,
+      "loss": 0.7422,
+      "step": 4485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.414126789362269e-07,
+      "loss": 0.7406,
+      "step": 4486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4026800902513293e-07,
+      "loss": 0.8177,
+      "step": 4487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.391279580236926e-07,
+      "loss": 0.8057,
+      "step": 4488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3799252646597428e-07,
+      "loss": 0.7552,
+      "step": 4489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.368617148838869e-07,
+      "loss": 0.7809,
+      "step": 4490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3573552380717002e-07,
+      "loss": 0.724,
+      "step": 4491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3461395376340502e-07,
+      "loss": 0.7288,
+      "step": 4492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3349700527800157e-07,
+      "loss": 0.7663,
+      "step": 4493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.323846788742078e-07,
+      "loss": 0.7907,
+      "step": 4494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3127697507310911e-07,
+      "loss": 0.7515,
+      "step": 4495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3017389439361928e-07,
+      "loss": 0.6823,
+      "step": 4496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2907543735249163e-07,
+      "loss": 0.7705,
+      "step": 4497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2798160446431006e-07,
+      "loss": 0.7367,
+      "step": 4498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2689239624149473e-07,
+      "loss": 0.7383,
+      "step": 4499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2580781319429858e-07,
+      "loss": 0.8096,
+      "step": 4500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.247278558308085e-07,
+      "loss": 0.7713,
+      "step": 4501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2365252465694088e-07,
+      "loss": 0.7653,
+      "step": 4502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2258182017644839e-07,
+      "loss": 0.7345,
+      "step": 4503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.215157428909175e-07,
+      "loss": 0.7738,
+      "step": 4504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.20454293299761e-07,
+      "loss": 0.7731,
+      "step": 4505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.193974719002311e-07,
+      "loss": 0.734,
+      "step": 4506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1834527918740624e-07,
+      "loss": 0.7249,
+      "step": 4507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.172977156541999e-07,
+      "loss": 0.7679,
+      "step": 4508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1625478179135508e-07,
+      "loss": 0.7402,
+      "step": 4509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1521647808744873e-07,
+      "loss": 0.7419,
+      "step": 4510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1418280502888401e-07,
+      "loss": 0.7686,
+      "step": 4511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.13153763099898e-07,
+      "loss": 0.7673,
+      "step": 4512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1212935278255954e-07,
+      "loss": 0.7038,
+      "step": 4513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1110957455676252e-07,
+      "loss": 0.7663,
+      "step": 4514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1009442890023814e-07,
+      "loss": 0.7393,
+      "step": 4515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0908391628854042e-07,
+      "loss": 0.7539,
+      "step": 4516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0807803719505849e-07,
+      "loss": 0.7477,
+      "step": 4517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0707679209100652e-07,
+      "loss": 0.7979,
+      "step": 4518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0608018144543042e-07,
+      "loss": 0.7012,
+      "step": 4519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0508820572520672e-07,
+      "loss": 0.7663,
+      "step": 4520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0410086539503372e-07,
+      "loss": 0.7181,
+      "step": 4521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0311816091744698e-07,
+      "loss": 0.7832,
+      "step": 4522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.021400927528049e-07,
+      "loss": 0.764,
+      "step": 4523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.011666613592932e-07,
+      "loss": 0.8193,
+      "step": 4524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0019786719293046e-07,
+      "loss": 0.7956,
+      "step": 4525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.923371070755805e-08,
+      "loss": 0.7682,
+      "step": 4526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.827419235484803e-08,
+      "loss": 0.7663,
+      "step": 4527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 0.7471,
+      "step": 4528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.636907184323085e-08,
+      "loss": 0.7599,
+      "step": 4529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.54234705767998e-08,
+      "loss": 0.8135,
+      "step": 4530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.44825092279833e-08,
+      "loss": 0.7926,
+      "step": 4531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.354618823758654e-08,
+      "loss": 0.8122,
+      "step": 4532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.261450804423867e-08,
+      "loss": 0.8366,
+      "step": 4533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.168746908439718e-08,
+      "loss": 0.7529,
+      "step": 4534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.076507179234473e-08,
+      "loss": 0.7839,
+      "step": 4535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.984731660019008e-08,
+      "loss": 0.7412,
+      "step": 4536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.89342039378649e-08,
+      "loss": 0.7389,
+      "step": 4537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.802573423313032e-08,
+      "loss": 0.7751,
+      "step": 4538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.712190791156705e-08,
+      "loss": 0.7513,
+      "step": 4539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.622272539658416e-08,
+      "loss": 0.7559,
+      "step": 4540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.532818710941581e-08,
+      "loss": 0.7702,
+      "step": 4541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.443829346911792e-08,
+      "loss": 0.721,
+      "step": 4542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.355304489257254e-08,
+      "loss": 0.7371,
+      "step": 4543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.267244179448464e-08,
+      "loss": 0.7357,
+      "step": 4544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.179648458738309e-08,
+      "loss": 0.8034,
+      "step": 4545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.092517368162078e-08,
+      "loss": 0.766,
+      "step": 4546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.005850948537453e-08,
+      "loss": 0.7907,
+      "step": 4547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.919649240464177e-08,
+      "loss": 0.7324,
+      "step": 4548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.833912284324618e-08,
+      "loss": 0.7383,
+      "step": 4549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.748640120283201e-08,
+      "loss": 0.8359,
+      "step": 4550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.663832788286752e-08,
+      "loss": 0.7555,
+      "step": 4551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.579490328064265e-08,
+      "loss": 0.7861,
+      "step": 4552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.495612779126915e-08,
+      "loss": 0.7699,
+      "step": 4553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.412200180768269e-08,
+      "loss": 0.7614,
+      "step": 4554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.329252572063961e-08,
+      "loss": 0.7464,
+      "step": 4555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.246769991871683e-08,
+      "loss": 0.7363,
+      "step": 4556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.164752478831305e-08,
+      "loss": 0.7425,
+      "step": 4557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.083200071365204e-08,
+      "loss": 0.7751,
+      "step": 4558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.002112807677375e-08,
+      "loss": 0.7539,
+      "step": 4559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.921490725754098e-08,
+      "loss": 0.7668,
+      "step": 4560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.841333863363831e-08,
+      "loss": 0.752,
+      "step": 4561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.735,
+      "step": 4562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.682415947166122e-08,
+      "loss": 0.7565,
+      "step": 4563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.603654967805684e-08,
+      "loss": 0.7699,
+      "step": 4564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.525359356872263e-08,
+      "loss": 0.7744,
+      "step": 4565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.447529151044296e-08,
+      "loss": 0.7923,
+      "step": 4566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.370164386782285e-08,
+      "loss": 0.7293,
+      "step": 4567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.293265100328682e-08,
+      "loss": 0.7546,
+      "step": 4568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.216831327708006e-08,
+      "loss": 0.7822,
+      "step": 4569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.140863104726391e-08,
+      "loss": 0.6908,
+      "step": 4570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.065360466972148e-08,
+      "loss": 0.7995,
+      "step": 4571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.990323449815316e-08,
+      "loss": 0.7114,
+      "step": 4572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9157520884079996e-08,
+      "loss": 0.7201,
+      "step": 4573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8416464176840323e-08,
+      "loss": 0.7248,
+      "step": 4574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.768006472359089e-08,
+      "loss": 0.7718,
+      "step": 4575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.694832286930685e-08,
+      "loss": 0.7121,
+      "step": 4576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6221238956780664e-08,
+      "loss": 0.6992,
+      "step": 4577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5498813326624325e-08,
+      "loss": 0.7308,
+      "step": 4578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.7777,
+      "step": 4579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.406793826495449e-08,
+      "loss": 0.7979,
+      "step": 4580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.335948950375036e-08,
+      "loss": 0.7542,
+      "step": 4581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.265570036553813e-08,
+      "loss": 0.7202,
+      "step": 4582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.195657118001407e-08,
+      "loss": 0.7718,
+      "step": 4583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.126210227469508e-08,
+      "loss": 0.7816,
+      "step": 4584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0572293974912034e-08,
+      "loss": 0.7663,
+      "step": 4585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.988714660381422e-08,
+      "loss": 0.7135,
+      "step": 4586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.920666048236933e-08,
+      "loss": 0.6899,
+      "step": 4587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.85308359293557e-08,
+      "loss": 0.7565,
+      "step": 4588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7859673261374527e-08,
+      "loss": 0.8512,
+      "step": 4589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7193172792837635e-08,
+      "loss": 0.7536,
+      "step": 4590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6531334835976383e-08,
+      "loss": 0.7816,
+      "step": 4591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.58741597008372e-08,
+      "loss": 0.7591,
+      "step": 4592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.522164769528048e-08,
+      "loss": 0.7214,
+      "step": 4593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.457379912498394e-08,
+      "loss": 0.7731,
+      "step": 4594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3930614293440364e-08,
+      "loss": 0.7533,
+      "step": 4595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.329209350195651e-08,
+      "loss": 0.7331,
+      "step": 4596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2658237049655325e-08,
+      "loss": 0.8005,
+      "step": 4597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.202904523347484e-08,
+      "loss": 0.7406,
+      "step": 4598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1404518348167054e-08,
+      "loss": 0.7725,
+      "step": 4599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.078465668629905e-08,
+      "loss": 0.7861,
+      "step": 4600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.016946053825299e-08,
+      "loss": 0.7568,
+      "step": 4601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.955893019222501e-08,
+      "loss": 0.8324,
+      "step": 4602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.895306593422521e-08,
+      "loss": 0.7279,
+      "step": 4603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.835186804807656e-08,
+      "loss": 0.6882,
+      "step": 4604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.775533681541932e-08,
+      "loss": 0.7021,
+      "step": 4605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.716347251570551e-08,
+      "loss": 0.7461,
+      "step": 4606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6576275426200014e-08,
+      "loss": 0.8096,
+      "step": 4607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.599374582198278e-08,
+      "loss": 0.7438,
+      "step": 4608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.541588397594664e-08,
+      "loss": 0.7412,
+      "step": 4609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.484269015879838e-08,
+      "loss": 0.7428,
+      "step": 4610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.427416463905764e-08,
+      "loss": 0.7705,
+      "step": 4611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.371030768305583e-08,
+      "loss": 0.7067,
+      "step": 4612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.315111955493944e-08,
+      "loss": 0.7868,
+      "step": 4613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.259660051666669e-08,
+      "loss": 0.6955,
+      "step": 4614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.204675082800979e-08,
+      "loss": 0.7445,
+      "step": 4615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.150157074655047e-08,
+      "loss": 0.7806,
+      "step": 4616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0961060527685546e-08,
+      "loss": 0.6995,
+      "step": 4617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.042522042462359e-08,
+      "loss": 0.7298,
+      "step": 4618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9894050688386024e-08,
+      "loss": 0.8125,
+      "step": 4619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.93675515678038e-08,
+      "loss": 0.7367,
+      "step": 4620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8845723309525175e-08,
+      "loss": 0.79,
+      "step": 4621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8328566158002392e-08,
+      "loss": 0.7622,
+      "step": 4622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7816080355508312e-08,
+      "loss": 0.7943,
+      "step": 4623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7308266142119788e-08,
+      "loss": 0.7594,
+      "step": 4624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6805123755729857e-08,
+      "loss": 0.785,
+      "step": 4625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6306653432041086e-08,
+      "loss": 0.7451,
+      "step": 4626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5812855404568903e-08,
+      "loss": 0.7497,
+      "step": 4627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5323729904637162e-08,
+      "loss": 0.7646,
+      "step": 4628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.483927716138479e-08,
+      "loss": 0.7884,
+      "step": 4629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 0.7803,
+      "step": 4630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.388439085051486e-08,
+      "loss": 0.7959,
+      "step": 4631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3413957730226144e-08,
+      "loss": 0.7485,
+      "step": 4632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.294819826127115e-08,
+      "loss": 0.714,
+      "step": 4633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2487112661840894e-08,
+      "loss": 0.7624,
+      "step": 4634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.203070114793593e-08,
+      "loss": 0.7484,
+      "step": 4635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1578963933367448e-08,
+      "loss": 0.7686,
+      "step": 4636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.113190122975839e-08,
+      "loss": 0.7873,
+      "step": 4637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0689513246540115e-08,
+      "loss": 0.7689,
+      "step": 4638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025180019095352e-08,
+      "loss": 0.737,
+      "step": 4639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9818762268052348e-08,
+      "loss": 0.7858,
+      "step": 4640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.939039968069656e-08,
+      "loss": 0.7933,
+      "step": 4641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.896671262955896e-08,
+      "loss": 0.7031,
+      "step": 4642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.854770131312078e-08,
+      "loss": 0.7539,
+      "step": 4643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8133365927672785e-08,
+      "loss": 0.7653,
+      "step": 4644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7723706667314155e-08,
+      "loss": 0.7773,
+      "step": 4645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.731872372395693e-08,
+      "loss": 0.7256,
+      "step": 4646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6918417287318245e-08,
+      "loss": 0.7438,
+      "step": 4647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6522787544926977e-08,
+      "loss": 0.7503,
+      "step": 4648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613183468212154e-08,
+      "loss": 0.7731,
+      "step": 4649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574555888204765e-08,
+      "loss": 0.7331,
+      "step": 4650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5363960325660565e-08,
+      "loss": 0.8158,
+      "step": 4651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.498703919172506e-08,
+      "loss": 0.7646,
+      "step": 4652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.461479565681434e-08,
+      "loss": 0.7471,
+      "step": 4653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4247229895311132e-08,
+      "loss": 0.776,
+      "step": 4654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3884342079405478e-08,
+      "loss": 0.7884,
+      "step": 4655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3526132379095835e-08,
+      "loss": 0.7747,
+      "step": 4656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3172600962190196e-08,
+      "loss": 0.8447,
+      "step": 4657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.282374799430497e-08,
+      "loss": 0.7454,
+      "step": 4658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2479573638863873e-08,
+      "loss": 0.7176,
+      "step": 4659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2140078057101269e-08,
+      "loss": 0.7402,
+      "step": 4660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1805261408055491e-08,
+      "loss": 0.7699,
+      "step": 4661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.147512384857663e-08,
+      "loss": 0.8324,
+      "step": 4662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1149665533320974e-08,
+      "loss": 0.749,
+      "step": 4663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0828886614754342e-08,
+      "loss": 0.7451,
+      "step": 4664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0512787243149858e-08,
+      "loss": 0.7796,
+      "step": 4665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0201367566585741e-08,
+      "loss": 0.7126,
+      "step": 4666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.894627730953066e-09,
+      "loss": 0.762,
+      "step": 4667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.59256787994578e-09,
+      "loss": 0.7148,
+      "step": 4668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.29518815506847e-09,
+      "loss": 0.7803,
+      "step": 4669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.002488695631918e-09,
+      "loss": 0.734,
+      "step": 4670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.71446963875422e-09,
+      "loss": 0.7799,
+      "step": 4671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.431131119361891e-09,
+      "loss": 0.7344,
+      "step": 4672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.152473270187644e-09,
+      "loss": 0.7617,
+      "step": 4673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.878496221773724e-09,
+      "loss": 0.736,
+      "step": 4674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.609200102465241e-09,
+      "loss": 0.7126,
+      "step": 4675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.344585038420171e-09,
+      "loss": 0.7269,
+      "step": 4676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.084651153599353e-09,
+      "loss": 0.7332,
+      "step": 4677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.82939856977094e-09,
+      "loss": 0.7731,
+      "step": 4678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5788274065126115e-09,
+      "loss": 0.7874,
+      "step": 4679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3329377812060276e-09,
+      "loss": 0.708,
+      "step": 4680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.091729809042379e-09,
+      "loss": 0.6859,
+      "step": 4681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.855203603017945e-09,
+      "loss": 0.723,
+      "step": 4682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.623359273936313e-09,
+      "loss": 0.7803,
+      "step": 4683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.396196930407272e-09,
+      "loss": 0.7744,
+      "step": 4684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.173716678849028e-09,
+      "loss": 0.7464,
+      "step": 4685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.955918623483769e-09,
+      "loss": 0.7773,
+      "step": 4686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.74280286634099e-09,
+      "loss": 0.766,
+      "step": 4687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.534369507259717e-09,
+      "loss": 0.764,
+      "step": 4688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3306186438807354e-09,
+      "loss": 0.7507,
+      "step": 4689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1315503716554685e-09,
+      "loss": 0.7542,
+      "step": 4690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.93716478383821e-09,
+      "loss": 0.7702,
+      "step": 4691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.747461971492783e-09,
+      "loss": 0.7682,
+      "step": 4692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.562442023485879e-09,
+      "loss": 0.7637,
+      "step": 4693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.382105026494831e-09,
+      "loss": 0.7793,
+      "step": 4694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2064510649976176e-09,
+      "loss": 0.8385,
+      "step": 4695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0354802212839705e-09,
+      "loss": 0.7451,
+      "step": 4696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8691925754453785e-09,
+      "loss": 0.7839,
+      "step": 4697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7075882053828605e-09,
+      "loss": 0.7493,
+      "step": 4698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5506671868014145e-09,
+      "loss": 0.7448,
+      "step": 4699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.398429593212237e-09,
+      "loss": 0.7819,
+      "step": 4700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.250875495932725e-09,
+      "loss": 0.7855,
+      "step": 4701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.108004964086474e-09,
+      "loss": 0.7878,
+      "step": 4702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9698180646032793e-09,
+      "loss": 0.7751,
+      "step": 4703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.836314862219135e-09,
+      "loss": 0.7533,
+      "step": 4704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7074954194729044e-09,
+      "loss": 0.8112,
+      "step": 4705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.58335979671409e-09,
+      "loss": 0.7319,
+      "step": 4706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4639080520939541e-09,
+      "loss": 0.8086,
+      "step": 4707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3491402415710674e-09,
+      "loss": 0.8691,
+      "step": 4708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2390564189113108e-09,
+      "loss": 0.7624,
+      "step": 4709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1336566356834334e-09,
+      "loss": 0.7913,
+      "step": 4710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0329409412634938e-09,
+      "loss": 0.7982,
+      "step": 4711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.369093828326403e-10,
+      "loss": 0.8369,
+      "step": 4712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.455620053793301e-10,
+      "loss": 0.7254,
+      "step": 4713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.588988516937789e-10,
+      "loss": 0.8089,
+      "step": 4714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.769199623779532e-10,
+      "loss": 0.7493,
+      "step": 4715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.996253758322468e-10,
+      "loss": 0.7493,
+      "step": 4716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.270151282688041e-10,
+      "loss": 0.7988,
+      "step": 4717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.590892537015279e-10,
+      "loss": 0.7181,
+      "step": 4718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.958477839516306e-10,
+      "loss": 0.7721,
+      "step": 4719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3729074864541354e-10,
+      "loss": 0.7222,
+      "step": 4720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.834181752142673e-10,
+      "loss": 0.7881,
+      "step": 4721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3423008889467134e-10,
+      "loss": 0.7503,
+      "step": 4722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.897265127315251e-10,
+      "loss": 0.7139,
+      "step": 4723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.49907467570376e-10,
+      "loss": 0.7731,
+      "step": 4724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1477297206741179e-10,
+      "loss": 0.7386,
+      "step": 4725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.432304268057856e-11,
+      "loss": 0.7767,
+      "step": 4726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.855769367402176e-11,
+      "loss": 0.7695,
+      "step": 4727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.747693711919631e-11,
+      "loss": 0.7633,
+      "step": 4728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.108078289042581e-11,
+      "loss": 0.7425,
+      "step": 4729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.36923866934336e-12,
+      "loss": 0.7318,
+      "step": 4730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3423099415609272e-12,
+      "loss": 0.7617,
+      "step": 4731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.7528,
+      "step": 4732
+    },
+    {
+      "epoch": 1.0,
+      "step": 4732,
+      "total_flos": 1.5125912174198784e+16,
+      "train_loss": 0.8187214962804308,
+      "train_runtime": 158109.7293,
+      "train_samples_per_second": 3.95,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "max_steps": 4732,
+  "num_train_epochs": 1,
+  "total_flos": 1.5125912174198784e+16,
+  "trial_name": null,
+  "trial_params": null
+}