diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.117242291914781,
+  "global_step": 2000,
+  "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": 1.3588,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 1.3532,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.225352112676057e-07,
+      "loss": 1.4188,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 1.3776,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.042253521126762e-07,
+      "loss": 1.327,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 1.3892,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.859154929577465e-07,
+      "loss": 1.2351,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 1.1658,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.267605633802817e-06,
+      "loss": 1.1165,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 1.1155,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5492957746478873e-06,
+      "loss": 0.985,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 0.8844,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8309859154929579e-06,
+      "loss": 0.8122,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 0.7783,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1126760563380285e-06,
+      "loss": 0.6728,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 0.5941,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3943661971830984e-06,
+      "loss": 0.5069,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 0.5503,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.676056338028169e-06,
+      "loss": 0.5109,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 0.5798,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9577464788732396e-06,
+      "loss": 0.542,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 0.4672,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2394366197183104e-06,
+      "loss": 0.4059,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 0.4133,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5211267605633804e-06,
+      "loss": 0.4773,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 0.4849,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8028169014084508e-06,
+      "loss": 0.4071,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 0.3821,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0845070422535216e-06,
+      "loss": 0.3784,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 0.4295,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3661971830985915e-06,
+      "loss": 0.3354,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 0.3738,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.647887323943662e-06,
+      "loss": 0.3997,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 0.3898,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.929577464788733e-06,
+      "loss": 0.4637,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 0.4109,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.211267605633803e-06,
+      "loss": 0.3599,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 0.327,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.492957746478874e-06,
+      "loss": 0.3182,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 0.3268,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.774647887323944e-06,
+      "loss": 0.3395,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 0.3895,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.056338028169015e-06,
+      "loss": 0.3311,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 0.3925,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3380281690140855e-06,
+      "loss": 0.3229,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 0.3355,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.619718309859155e-06,
+      "loss": 0.3685,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 0.3004,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.901408450704225e-06,
+      "loss": 0.3244,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 0.3356,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.183098591549297e-06,
+      "loss": 0.317,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 0.3049,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.464788732394367e-06,
+      "loss": 0.3493,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 0.4302,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.746478873239436e-06,
+      "loss": 0.3039,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 0.2895,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.028169014084509e-06,
+      "loss": 0.3047,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 0.2937,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.309859154929578e-06,
+      "loss": 0.345,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 0.2762,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.591549295774648e-06,
+      "loss": 0.2838,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 0.3404,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.87323943661972e-06,
+      "loss": 0.3144,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 0.3802,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.15492957746479e-06,
+      "loss": 0.2606,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 0.2838,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.43661971830986e-06,
+      "loss": 0.3846,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 0.3866,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.71830985915493e-06,
+      "loss": 0.2888,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 0.301,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.3582,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 0.329,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0281690140845072e-05,
+      "loss": 0.242,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 0.2772,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.056338028169014e-05,
+      "loss": 0.2344,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 0.2539,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0845070422535213e-05,
+      "loss": 0.4211,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 0.168,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1126760563380284e-05,
+      "loss": 0.2134,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 0.2802,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1408450704225353e-05,
+      "loss": 0.2083,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 0.2594,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1690140845070424e-05,
+      "loss": 0.2853,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 0.3088,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1971830985915493e-05,
+      "loss": 0.3551,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 0.3158,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2253521126760564e-05,
+      "loss": 0.3,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 0.2901,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2535211267605636e-05,
+      "loss": 0.3123,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 0.3006,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2816901408450705e-05,
+      "loss": 0.2535,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 0.2758,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3098591549295776e-05,
+      "loss": 0.2801,
+      "step": 93
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 0.3236,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3380281690140845e-05,
+      "loss": 0.3269,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 0.2689,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3661971830985916e-05,
+      "loss": 0.3679,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 0.2487,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3943661971830987e-05,
+      "loss": 0.2409,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 0.2366,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4225352112676058e-05,
+      "loss": 0.312,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 0.1876,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4507042253521129e-05,
+      "loss": 0.2703,
+      "step": 103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 0.2234,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.47887323943662e-05,
+      "loss": 0.2932,
+      "step": 105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 0.2811,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5070422535211269e-05,
+      "loss": 0.3166,
+      "step": 107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 0.1605,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.535211267605634e-05,
+      "loss": 0.2785,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 0.3103,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.563380281690141e-05,
+      "loss": 0.3095,
+      "step": 111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 0.3486,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.591549295774648e-05,
+      "loss": 0.2373,
+      "step": 113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 0.2885,
+      "step": 114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.619718309859155e-05,
+      "loss": 0.2938,
+      "step": 115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 0.3567,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6478873239436623e-05,
+      "loss": 0.2582,
+      "step": 117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 0.2583,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.676056338028169e-05,
+      "loss": 0.2482,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 0.1909,
+      "step": 120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.704225352112676e-05,
+      "loss": 0.2548,
+      "step": 121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 0.2672,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.732394366197183e-05,
+      "loss": 0.3282,
+      "step": 123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 0.264,
+      "step": 124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7605633802816902e-05,
+      "loss": 0.2659,
+      "step": 125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 0.2986,
+      "step": 126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7887323943661975e-05,
+      "loss": 0.2417,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 0.263,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8169014084507044e-05,
+      "loss": 0.2653,
+      "step": 129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 0.1904,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8450704225352113e-05,
+      "loss": 0.3,
+      "step": 131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 0.2522,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8732394366197186e-05,
+      "loss": 0.2263,
+      "step": 133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 0.2753,
+      "step": 134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9014084507042255e-05,
+      "loss": 0.1879,
+      "step": 135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 0.2746,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295774647887327e-05,
+      "loss": 0.2915,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 0.327,
+      "step": 138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9577464788732396e-05,
+      "loss": 0.2386,
+      "step": 139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 0.3134,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9859154929577465e-05,
+      "loss": 0.2452,
+      "step": 141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.2057,
+      "step": 142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999764539449e-05,
+      "loss": 0.2475,
+      "step": 143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999058157907e-05,
+      "loss": 0.2109,
+      "step": 144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999978808557063e-05,
+      "loss": 0.2774,
+      "step": 145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999962326334016e-05,
+      "loss": 0.3203,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999994113491769e-05,
+      "loss": 0.2016,
+      "step": 147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999915234318064e-05,
+      "loss": 0.2281,
+      "step": 148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999884624547334e-05,
+      "loss": 0.3156,
+      "step": 149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999849305619918e-05,
+      "loss": 0.2569,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999809277552445e-05,
+      "loss": 0.2951,
+      "step": 151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999764540363774e-05,
+      "loss": 0.1925,
+      "step": 152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999715094074957e-05,
+      "loss": 0.2392,
+      "step": 153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999966093870929e-05,
+      "loss": 0.1908,
+      "step": 154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999602074292277e-05,
+      "loss": 0.1818,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999538500851633e-05,
+      "loss": 0.2781,
+      "step": 156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99994702184173e-05,
+      "loss": 0.183,
+      "step": 157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999939722702143e-05,
+      "loss": 0.1785,
+      "step": 158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999319526698397e-05,
+      "loss": 0.2867,
+      "step": 159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99992371174848e-05,
+      "loss": 0.2966,
+      "step": 160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999149999419433e-05,
+      "loss": 0.2355,
+      "step": 161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999905817254333e-05,
+      "loss": 0.2922,
+      "step": 162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998961636899736e-05,
+      "loss": 0.179,
+      "step": 163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998860392534105e-05,
+      "loss": 0.2152,
+      "step": 164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999875443949412e-05,
+      "loss": 0.4178,
+      "step": 165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998643777829674e-05,
+      "loss": 0.2585,
+      "step": 166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998528407592883e-05,
+      "loss": 0.2625,
+      "step": 167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999840832883808e-05,
+      "loss": 0.3226,
+      "step": 168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998283541621803e-05,
+      "loss": 0.1698,
+      "step": 169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998154046002822e-05,
+      "loss": 0.333,
+      "step": 170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999801984204212e-05,
+      "loss": 0.228,
+      "step": 171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997880929802895e-05,
+      "loss": 0.2478,
+      "step": 172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999773730935057e-05,
+      "loss": 0.231,
+      "step": 173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999758898075277e-05,
+      "loss": 0.2161,
+      "step": 174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997435944079347e-05,
+      "loss": 0.254,
+      "step": 175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997278199402373e-05,
+      "loss": 0.3393,
+      "step": 176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997115746796132e-05,
+      "loss": 0.2811,
+      "step": 177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996948586337127e-05,
+      "loss": 0.2451,
+      "step": 178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996776718104076e-05,
+      "loss": 0.266,
+      "step": 179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996600142177917e-05,
+      "loss": 0.2087,
+      "step": 180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996418858641805e-05,
+      "loss": 0.2447,
+      "step": 181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996232867581104e-05,
+      "loss": 0.2866,
+      "step": 182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996042169083404e-05,
+      "loss": 0.2374,
+      "step": 183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9995846763238514e-05,
+      "loss": 0.2446,
+      "step": 184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995646650138446e-05,
+      "loss": 0.2867,
+      "step": 185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995441829877444e-05,
+      "loss": 0.3097,
+      "step": 186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995232302551965e-05,
+      "loss": 0.2481,
+      "step": 187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995018068260667e-05,
+      "loss": 0.259,
+      "step": 188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999479912710445e-05,
+      "loss": 0.203,
+      "step": 189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999457547918641e-05,
+      "loss": 0.2391,
+      "step": 190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994347124611873e-05,
+      "loss": 0.3012,
+      "step": 191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994114063488376e-05,
+      "loss": 0.2428,
+      "step": 192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993876295925667e-05,
+      "loss": 0.2919,
+      "step": 193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993633822035722e-05,
+      "loss": 0.2051,
+      "step": 194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993386641932724e-05,
+      "loss": 0.2055,
+      "step": 195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993134755733075e-05,
+      "loss": 0.3715,
+      "step": 196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999287816355539e-05,
+      "loss": 0.2142,
+      "step": 197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992616865520515e-05,
+      "loss": 0.2245,
+      "step": 198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999235086175149e-05,
+      "loss": 0.2536,
+      "step": 199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992080152373587e-05,
+      "loss": 0.1905,
+      "step": 200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991804737514282e-05,
+      "loss": 0.261,
+      "step": 201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991524617303282e-05,
+      "loss": 0.1881,
+      "step": 202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991239791872497e-05,
+      "loss": 0.3032,
+      "step": 203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999095026135606e-05,
+      "loss": 0.2575,
+      "step": 204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990656025890315e-05,
+      "loss": 0.2001,
+      "step": 205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990357085613828e-05,
+      "loss": 0.3263,
+      "step": 206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990053440667368e-05,
+      "loss": 0.1952,
+      "step": 207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998974509119393e-05,
+      "loss": 0.2277,
+      "step": 208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989432037338733e-05,
+      "loss": 0.2162,
+      "step": 209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989114279249185e-05,
+      "loss": 0.2282,
+      "step": 210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988791817074933e-05,
+      "loss": 0.2923,
+      "step": 211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9988464650967834e-05,
+      "loss": 0.2478,
+      "step": 212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998813278108195e-05,
+      "loss": 0.2539,
+      "step": 213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987796207573573e-05,
+      "loss": 0.2271,
+      "step": 214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987454930601196e-05,
+      "loss": 0.2197,
+      "step": 215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998710895032554e-05,
+      "loss": 0.2409,
+      "step": 216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998675826690953e-05,
+      "loss": 0.2292,
+      "step": 217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998640288051831e-05,
+      "loss": 0.2381,
+      "step": 218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998604279131924e-05,
+      "loss": 0.2874,
+      "step": 219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985677999481898e-05,
+      "loss": 0.2396,
+      "step": 220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998530850517806e-05,
+      "loss": 0.2662,
+      "step": 221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998493430858174e-05,
+      "loss": 0.3043,
+      "step": 222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984555409869152e-05,
+      "loss": 0.2461,
+      "step": 223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998417180921873e-05,
+      "loss": 0.278,
+      "step": 224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983783506811115e-05,
+      "loss": 0.2596,
+      "step": 225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983390502829168e-05,
+      "loss": 0.275,
+      "step": 226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982992797457962e-05,
+      "loss": 0.3269,
+      "step": 227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982590390884787e-05,
+      "loss": 0.2692,
+      "step": 228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982183283299144e-05,
+      "loss": 0.2947,
+      "step": 229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981771474892752e-05,
+      "loss": 0.2649,
+      "step": 230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981354965859532e-05,
+      "loss": 0.2776,
+      "step": 231
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980933756395635e-05,
+      "loss": 0.2096,
+      "step": 232
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998050784669941e-05,
+      "loss": 0.1722,
+      "step": 233
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980077236971435e-05,
+      "loss": 0.2546,
+      "step": 234
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979641927414487e-05,
+      "loss": 0.3051,
+      "step": 235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979201918233567e-05,
+      "loss": 0.2236,
+      "step": 236
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978757209635882e-05,
+      "loss": 0.2749,
+      "step": 237
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978307801830855e-05,
+      "loss": 0.1621,
+      "step": 238
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997785369503012e-05,
+      "loss": 0.3455,
+      "step": 239
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977394889447526e-05,
+      "loss": 0.3188,
+      "step": 240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976931385299133e-05,
+      "loss": 0.2431,
+      "step": 241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997646318280322e-05,
+      "loss": 0.1676,
+      "step": 242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975990282180274e-05,
+      "loss": 0.2863,
+      "step": 243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975512683652985e-05,
+      "loss": 0.2066,
+      "step": 244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975030387446273e-05,
+      "loss": 0.2088,
+      "step": 245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997454339378726e-05,
+      "loss": 0.2908,
+      "step": 246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997405170290528e-05,
+      "loss": 0.257,
+      "step": 247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973555315031873e-05,
+      "loss": 0.2499,
+      "step": 248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973054230400813e-05,
+      "loss": 0.2086,
+      "step": 249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997254844924806e-05,
+      "loss": 0.2502,
+      "step": 250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972037971811802e-05,
+      "loss": 0.2836,
+      "step": 251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971522798332436e-05,
+      "loss": 0.1836,
+      "step": 252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997100292905256e-05,
+      "loss": 0.2634,
+      "step": 253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970478364217e-05,
+      "loss": 0.2305,
+      "step": 254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996994910407278e-05,
+      "loss": 0.2191,
+      "step": 255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996941514886914e-05,
+      "loss": 0.1912,
+      "step": 256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996887649885753e-05,
+      "loss": 0.2236,
+      "step": 257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9968333154291615e-05,
+      "loss": 0.2034,
+      "step": 258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967785115427268e-05,
+      "loss": 0.2636,
+      "step": 259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9967232382522564e-05,
+      "loss": 0.2273,
+      "step": 260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966674955837808e-05,
+      "loss": 0.2908,
+      "step": 261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966112835635493e-05,
+      "loss": 0.3204,
+      "step": 262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965546022180343e-05,
+      "loss": 0.2236,
+      "step": 263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964974515739275e-05,
+      "loss": 0.2227,
+      "step": 264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996439831658143e-05,
+      "loss": 0.1786,
+      "step": 265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963817424978144e-05,
+      "loss": 0.2504,
+      "step": 266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963231841202976e-05,
+      "loss": 0.2678,
+      "step": 267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962641565531694e-05,
+      "loss": 0.2869,
+      "step": 268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962046598242263e-05,
+      "loss": 0.2725,
+      "step": 269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996144693961487e-05,
+      "loss": 0.2796,
+      "step": 270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960842589931907e-05,
+      "loss": 0.2201,
+      "step": 271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960233549477973e-05,
+      "loss": 0.2732,
+      "step": 272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995961981853988e-05,
+      "loss": 0.1685,
+      "step": 273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995900139740665e-05,
+      "loss": 0.2394,
+      "step": 274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99583782863695e-05,
+      "loss": 0.1615,
+      "step": 275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995775048572188e-05,
+      "loss": 0.2536,
+      "step": 276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957117995759423e-05,
+      "loss": 0.183,
+      "step": 277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995648081677999e-05,
+      "loss": 0.217,
+      "step": 278
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955838949083633e-05,
+      "loss": 0.2385,
+      "step": 279
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955192392972628e-05,
+      "loss": 0.3389,
+      "step": 280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954541148751453e-05,
+      "loss": 0.2051,
+      "step": 281
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953885216726788e-05,
+      "loss": 0.2042,
+      "step": 282
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953224597207528e-05,
+      "loss": 0.2598,
+      "step": 283
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9952559290504773e-05,
+      "loss": 0.2316,
+      "step": 284
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951889296931826e-05,
+      "loss": 0.1929,
+      "step": 285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951214616804203e-05,
+      "loss": 0.2735,
+      "step": 286
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995053525043963e-05,
+      "loss": 0.2331,
+      "step": 287
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.994985119815803e-05,
+      "loss": 0.2275,
+      "step": 288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994916246028154e-05,
+      "loss": 0.2466,
+      "step": 289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948469037134496e-05,
+      "loss": 0.1711,
+      "step": 290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947770929043453e-05,
+      "loss": 0.2124,
+      "step": 291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994706813633716e-05,
+      "loss": 0.1516,
+      "step": 292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946360659346577e-05,
+      "loss": 0.2247,
+      "step": 293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945648498404875e-05,
+      "loss": 0.2372,
+      "step": 294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994493165384742e-05,
+      "loss": 0.2404,
+      "step": 295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994421012601179e-05,
+      "loss": 0.1487,
+      "step": 296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943483915237768e-05,
+      "loss": 0.2602,
+      "step": 297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994275302186734e-05,
+      "loss": 0.2173,
+      "step": 298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942017446244705e-05,
+      "loss": 0.2384,
+      "step": 299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994127718871626e-05,
+      "loss": 0.2545,
+      "step": 300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940532249630605e-05,
+      "loss": 0.2273,
+      "step": 301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939782629338543e-05,
+      "loss": 0.2348,
+      "step": 302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939028328193097e-05,
+      "loss": 0.2655,
+      "step": 303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938269346549473e-05,
+      "loss": 0.2908,
+      "step": 304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937505684765098e-05,
+      "loss": 0.2092,
+      "step": 305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936737343199596e-05,
+      "loss": 0.237,
+      "step": 306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9935964322214794e-05,
+      "loss": 0.2529,
+      "step": 307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935186622174723e-05,
+      "loss": 0.2462,
+      "step": 308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993440424344562e-05,
+      "loss": 0.1769,
+      "step": 309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933617186395917e-05,
+      "loss": 0.2325,
+      "step": 310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932825451396267e-05,
+      "loss": 0.2857,
+      "step": 311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9932029038819507e-05,
+      "loss": 0.209,
+      "step": 312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993122794904069e-05,
+      "loss": 0.1895,
+      "step": 313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930422182437058e-05,
+      "loss": 0.1818,
+      "step": 314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992961173938807e-05,
+      "loss": 0.2161,
+      "step": 315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928796620275377e-05,
+      "loss": 0.2288,
+      "step": 316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992797682548284e-05,
+      "loss": 0.2542,
+      "step": 317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927152355396515e-05,
+      "loss": 0.1581,
+      "step": 318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926323210404663e-05,
+      "loss": 0.1897,
+      "step": 319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925489390897744e-05,
+      "loss": 0.2069,
+      "step": 320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924650897268427e-05,
+      "loss": 0.2136,
+      "step": 321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923807729911567e-05,
+      "loss": 0.2184,
+      "step": 322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922959889224238e-05,
+      "loss": 0.2418,
+      "step": 323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.99221073756057e-05,
+      "loss": 0.2375,
+      "step": 324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921250189457426e-05,
+      "loss": 0.2802,
+      "step": 325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920388331183076e-05,
+      "loss": 0.2067,
+      "step": 326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991952180118852e-05,
+      "loss": 0.1856,
+      "step": 327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918650599881828e-05,
+      "loss": 0.2391,
+      "step": 328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917774727673265e-05,
+      "loss": 0.2068,
+      "step": 329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916894184975298e-05,
+      "loss": 0.1902,
+      "step": 330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916008972202586e-05,
+      "loss": 0.2211,
+      "step": 331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915119089772008e-05,
+      "loss": 0.2544,
+      "step": 332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991422453810262e-05,
+      "loss": 0.2947,
+      "step": 333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913325317615684e-05,
+      "loss": 0.2437,
+      "step": 334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991242142873467e-05,
+      "loss": 0.2702,
+      "step": 335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991151287188523e-05,
+      "loss": 0.2231,
+      "step": 336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910599647495227e-05,
+      "loss": 0.2319,
+      "step": 337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990968175599471e-05,
+      "loss": 0.2659,
+      "step": 338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908759197815945e-05,
+      "loss": 0.1548,
+      "step": 339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907831973393377e-05,
+      "loss": 0.2584,
+      "step": 340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906900083163664e-05,
+      "loss": 0.2779,
+      "step": 341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905963527565636e-05,
+      "loss": 0.2607,
+      "step": 342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905022307040353e-05,
+      "loss": 0.2353,
+      "step": 343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9904076422031048e-05,
+      "loss": 0.2368,
+      "step": 344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990312587298316e-05,
+      "loss": 0.2235,
+      "step": 345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902170660344323e-05,
+      "loss": 0.2003,
+      "step": 346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901210784564365e-05,
+      "loss": 0.199,
+      "step": 347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9900246246095315e-05,
+      "loss": 0.3057,
+      "step": 348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989927704539139e-05,
+      "loss": 0.2234,
+      "step": 349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989830318290901e-05,
+      "loss": 0.2569,
+      "step": 350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897324659106784e-05,
+      "loss": 0.2409,
+      "step": 351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896341474445526e-05,
+      "loss": 0.2396,
+      "step": 352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895353629388232e-05,
+      "loss": 0.1947,
+      "step": 353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894361124400104e-05,
+      "loss": 0.2105,
+      "step": 354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989336395994853e-05,
+      "loss": 0.2569,
+      "step": 355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892362136503102e-05,
+      "loss": 0.1851,
+      "step": 356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989135565453559e-05,
+      "loss": 0.2149,
+      "step": 357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890344514519974e-05,
+      "loss": 0.2738,
+      "step": 358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988932871693242e-05,
+      "loss": 0.126,
+      "step": 359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.1821,
+      "step": 360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988728315095713e-05,
+      "loss": 0.2242,
+      "step": 361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886253383532698e-05,
+      "loss": 0.2833,
+      "step": 362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885218960462924e-05,
+      "loss": 0.3323,
+      "step": 363
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884179882234946e-05,
+      "loss": 0.1555,
+      "step": 364
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883136149338083e-05,
+      "loss": 0.1766,
+      "step": 365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882087762263857e-05,
+      "loss": 0.1735,
+      "step": 366
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881034721505966e-05,
+      "loss": 0.2928,
+      "step": 367
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879977027560317e-05,
+      "loss": 0.2703,
+      "step": 368
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878914680925e-05,
+      "loss": 0.2238,
+      "step": 369
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877847682100294e-05,
+      "loss": 0.2453,
+      "step": 370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987677603158867e-05,
+      "loss": 0.2667,
+      "step": 371
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875699729894797e-05,
+      "loss": 0.1867,
+      "step": 372
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987461877752552e-05,
+      "loss": 0.2003,
+      "step": 373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9873533174989884e-05,
+      "loss": 0.1935,
+      "step": 374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872442922799127e-05,
+      "loss": 0.2179,
+      "step": 375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871348021466673e-05,
+      "loss": 0.2794,
+      "step": 376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9870248471508123e-05,
+      "loss": 0.2739,
+      "step": 377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986914427344129e-05,
+      "loss": 0.225,
+      "step": 378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986803542778616e-05,
+      "loss": 0.3317,
+      "step": 379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866921935064907e-05,
+      "loss": 0.2943,
+      "step": 380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865803795801905e-05,
+      "loss": 0.2157,
+      "step": 381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986468101052371e-05,
+      "loss": 0.1766,
+      "step": 382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863553579759062e-05,
+      "loss": 0.1984,
+      "step": 383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9862421504038887e-05,
+      "loss": 0.1836,
+      "step": 384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986128478389631e-05,
+      "loss": 0.2109,
+      "step": 385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986014341986664e-05,
+      "loss": 0.1852,
+      "step": 386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858997412487362e-05,
+      "loss": 0.2356,
+      "step": 387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9857846762298157e-05,
+      "loss": 0.209,
+      "step": 388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985669146984089e-05,
+      "loss": 0.2598,
+      "step": 389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855531535659612e-05,
+      "loss": 0.2871,
+      "step": 390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854366960300566e-05,
+      "loss": 0.2776,
+      "step": 391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985319774431217e-05,
+      "loss": 0.2588,
+      "step": 392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852023888245036e-05,
+      "loss": 0.3537,
+      "step": 393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985084539265195e-05,
+      "loss": 0.2696,
+      "step": 394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98496622580879e-05,
+      "loss": 0.1604,
+      "step": 395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848474485110043e-05,
+      "loss": 0.1738,
+      "step": 396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847282074277728e-05,
+      "loss": 0.2319,
+      "step": 397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984608502615249e-05,
+      "loss": 0.1838,
+      "step": 398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844883341298035e-05,
+      "loss": 0.1356,
+      "step": 399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984367702028027e-05,
+      "loss": 0.2351,
+      "step": 400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842466063667278e-05,
+      "loss": 0.2965,
+      "step": 401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9841250472029316e-05,
+      "loss": 0.2258,
+      "step": 402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984003024593884e-05,
+      "loss": 0.1767,
+      "step": 403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838805385970472e-05,
+      "loss": 0.2682,
+      "step": 404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837575892701033e-05,
+      "loss": 0.2179,
+      "step": 405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983634176670951e-05,
+      "loss": 0.2103,
+      "step": 406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9835103008577087e-05,
+      "loss": 0.2381,
+      "step": 407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983385961888711e-05,
+      "loss": 0.1793,
+      "step": 408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832611598225127e-05,
+      "loss": 0.2104,
+      "step": 409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831358947178855e-05,
+      "loss": 0.3119,
+      "step": 410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9830101666338195e-05,
+      "loss": 0.1814,
+      "step": 411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9828839756295223e-05,
+      "loss": 0.3456,
+      "step": 412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98275732176442e-05,
+      "loss": 0.1911,
+      "step": 413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982630205098157e-05,
+      "loss": 0.2159,
+      "step": 414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982502625690595e-05,
+      "loss": 0.1494,
+      "step": 415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823745836018132e-05,
+      "loss": 0.2542,
+      "step": 416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822460788921103e-05,
+      "loss": 0.1781,
+      "step": 417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982117111622001e-05,
+      "loss": 0.2328,
+      "step": 418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98198768185222e-05,
+      "loss": 0.2007,
+      "step": 419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981857789643717e-05,
+      "loss": 0.2472,
+      "step": 420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981727435057662e-05,
+      "loss": 0.2372,
+      "step": 421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815966181554412e-05,
+      "loss": 0.2445,
+      "step": 422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814653389986596e-05,
+      "loss": 0.315,
+      "step": 423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9813335976491387e-05,
+      "loss": 0.2185,
+      "step": 424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9812013941689182e-05,
+      "loss": 0.2276,
+      "step": 425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9810687286202564e-05,
+      "loss": 0.2731,
+      "step": 426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9809356010656274e-05,
+      "loss": 0.3174,
+      "step": 427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980802011567724e-05,
+      "loss": 0.1412,
+      "step": 428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980667960189457e-05,
+      "loss": 0.135,
+      "step": 429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980533446993953e-05,
+      "loss": 0.2627,
+      "step": 430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9803984720445576e-05,
+      "loss": 0.158,
+      "step": 431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802630354048335e-05,
+      "loss": 0.2426,
+      "step": 432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.98012713713856e-05,
+      "loss": 0.3373,
+      "step": 433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9799907773097355e-05,
+      "loss": 0.2839,
+      "step": 434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979853955982574e-05,
+      "loss": 0.2453,
+      "step": 435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797166732215078e-05,
+      "loss": 0.2471,
+      "step": 436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979578929091186e-05,
+      "loss": 0.207,
+      "step": 437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9794407236564753e-05,
+      "loss": 0.2541,
+      "step": 438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.97930205698246e-05,
+      "loss": 0.1886,
+      "step": 439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791629291344404e-05,
+      "loss": 0.1798,
+      "step": 440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790233401779354e-05,
+      "loss": 0.2795,
+      "step": 441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97888329017868e-05,
+      "loss": 0.2148,
+      "step": 442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978742779202627e-05,
+      "loss": 0.1864,
+      "step": 443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786018073159458e-05,
+      "loss": 0.2931,
+      "step": 444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978460374585023e-05,
+      "loss": 0.27,
+      "step": 445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978318481076462e-05,
+      "loss": 0.2334,
+      "step": 446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978176126857084e-05,
+      "loss": 0.1693,
+      "step": 447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780333119939264e-05,
+      "loss": 0.1427,
+      "step": 448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778900365542437e-05,
+      "loss": 0.2572,
+      "step": 449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977746300605507e-05,
+      "loss": 0.2146,
+      "step": 450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776021042154054e-05,
+      "loss": 0.3199,
+      "step": 451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774574474518432e-05,
+      "loss": 0.291,
+      "step": 452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977312330382943e-05,
+      "loss": 0.1539,
+      "step": 453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771667530770427e-05,
+      "loss": 0.2962,
+      "step": 454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770207156026987e-05,
+      "loss": 0.2151,
+      "step": 455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976874218028682e-05,
+      "loss": 0.2908,
+      "step": 456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767272604239823e-05,
+      "loss": 0.1857,
+      "step": 457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9765798428578045e-05,
+      "loss": 0.2946,
+      "step": 458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764319653995712e-05,
+      "loss": 0.2778,
+      "step": 459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762836281189207e-05,
+      "loss": 0.2532,
+      "step": 460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976134831085708e-05,
+      "loss": 0.2434,
+      "step": 461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975985574370005e-05,
+      "loss": 0.1976,
+      "step": 462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758358580420998e-05,
+      "loss": 0.2348,
+      "step": 463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975685682172497e-05,
+      "loss": 0.1494,
+      "step": 464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975535046831917e-05,
+      "loss": 0.3462,
+      "step": 465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753839520912984e-05,
+      "loss": 0.1888,
+      "step": 466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752323980217935e-05,
+      "loss": 0.1836,
+      "step": 467
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.975080384694773e-05,
+      "loss": 0.1967,
+      "step": 468
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.2195,
+      "step": 469
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747749805547474e-05,
+      "loss": 0.2479,
+      "step": 470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974621589885563e-05,
+      "loss": 0.1494,
+      "step": 471
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744677402465053e-05,
+      "loss": 0.2717,
+      "step": 472
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743134317100255e-05,
+      "loss": 0.2749,
+      "step": 473
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974158664348791e-05,
+      "loss": 0.2905,
+      "step": 474
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740034382356846e-05,
+      "loss": 0.162,
+      "step": 475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9738477534438055e-05,
+      "loss": 0.142,
+      "step": 476
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9736916100464692e-05,
+      "loss": 0.2746,
+      "step": 477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973535008117207e-05,
+      "loss": 0.1995,
+      "step": 478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733779477297655e-05,
+      "loss": 0.3014,
+      "step": 479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732204289581088e-05,
+      "loss": 0.2261,
+      "step": 480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973062451876415e-05,
+      "loss": 0.2244,
+      "step": 481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9729040165590784e-05,
+      "loss": 0.2273,
+      "step": 482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727451230807106e-05,
+      "loss": 0.1866,
+      "step": 483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9725857715161375e-05,
+      "loss": 0.1842,
+      "step": 484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972425961940401e-05,
+      "loss": 0.2946,
+      "step": 485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972265694428759e-05,
+      "loss": 0.2007,
+      "step": 486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721049690566843e-05,
+      "loss": 0.1497,
+      "step": 487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9719437858998667e-05,
+      "loss": 0.2807,
+      "step": 488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9717821450342102e-05,
+      "loss": 0.2672,
+      "step": 489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716200465358352e-05,
+      "loss": 0.2934,
+      "step": 490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9714574904810767e-05,
+      "loss": 0.1915,
+      "step": 491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712944769464864e-05,
+      "loss": 0.195,
+      "step": 492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711310060088307e-05,
+      "loss": 0.2371,
+      "step": 493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9709670777450912e-05,
+      "loss": 0.1941,
+      "step": 494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708026922324655e-05,
+      "loss": 0.2578,
+      "step": 495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706378495483664e-05,
+      "loss": 0.2871,
+      "step": 496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704725497704213e-05,
+      "loss": 0.2182,
+      "step": 497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703067929764736e-05,
+      "loss": 0.2002,
+      "step": 498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9701405792445815e-05,
+      "loss": 0.2202,
+      "step": 499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.969973908653019e-05,
+      "loss": 0.1896,
+      "step": 500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698067812802743e-05,
+      "loss": 0.2118,
+      "step": 501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696391972050516e-05,
+      "loss": 0.2486,
+      "step": 502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9694711565062696e-05,
+      "loss": 0.1477,
+      "step": 503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693026592630614e-05,
+      "loss": 0.2328,
+      "step": 504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691337055547777e-05,
+      "loss": 0.3002,
+      "step": 505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9689642954609808e-05,
+      "loss": 0.293,
+      "step": 506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687944290614502e-05,
+      "loss": 0.2379,
+      "step": 507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686241064361792e-05,
+      "loss": 0.1872,
+      "step": 508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968453327665377e-05,
+      "loss": 0.2975,
+      "step": 509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9682820928294664e-05,
+      "loss": 0.2274,
+      "step": 510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681104020090854e-05,
+      "loss": 0.1789,
+      "step": 511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9679382552850867e-05,
+      "loss": 0.1762,
+      "step": 512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677656527385386e-05,
+      "loss": 0.275,
+      "step": 513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675925944507226e-05,
+      "loss": 0.3315,
+      "step": 514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967419080503136e-05,
+      "loss": 0.2151,
+      "step": 515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967245110977489e-05,
+      "loss": 0.2113,
+      "step": 516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670706859557093e-05,
+      "loss": 0.2714,
+      "step": 517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9668958055199365e-05,
+      "loss": 0.3302,
+      "step": 518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667204697525247e-05,
+      "loss": 0.1901,
+      "step": 519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9665446787360444e-05,
+      "loss": 0.2217,
+      "step": 520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966368432553279e-05,
+      "loss": 0.1616,
+      "step": 521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661917312872258e-05,
+      "loss": 0.2059,
+      "step": 522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966014575021098e-05,
+      "loss": 0.2378,
+      "step": 523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.965836963838322e-05,
+      "loss": 0.1845,
+      "step": 524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9656588978225386e-05,
+      "loss": 0.2151,
+      "step": 525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965480377057603e-05,
+      "loss": 0.1991,
+      "step": 526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9653014016275838e-05,
+      "loss": 0.2233,
+      "step": 527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965121971616765e-05,
+      "loss": 0.1851,
+      "step": 528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964942087109644e-05,
+      "loss": 0.2454,
+      "step": 529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647617481909314e-05,
+      "loss": 0.1709,
+      "step": 530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645809549455533e-05,
+      "loss": 0.2835,
+      "step": 531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964399707458649e-05,
+      "loss": 0.2774,
+      "step": 532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.964218005815572e-05,
+      "loss": 0.2023,
+      "step": 533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9640358501018885e-05,
+      "loss": 0.3693,
+      "step": 534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9638532404033807e-05,
+      "loss": 0.2202,
+      "step": 535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963670176806042e-05,
+      "loss": 0.1481,
+      "step": 536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634866593960822e-05,
+      "loss": 0.2199,
+      "step": 537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633026882599228e-05,
+      "loss": 0.1547,
+      "step": 538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9631182634842e-05,
+      "loss": 0.1701,
+      "step": 539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962933385155763e-05,
+      "loss": 0.2319,
+      "step": 540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962748053361675e-05,
+      "loss": 0.1936,
+      "step": 541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962562268189213e-05,
+      "loss": 0.31,
+      "step": 542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9623760297258668e-05,
+      "loss": 0.2202,
+      "step": 543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621893380593398e-05,
+      "loss": 0.218,
+      "step": 544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9620021932775496e-05,
+      "loss": 0.1992,
+      "step": 545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618145954686265e-05,
+      "loss": 0.1995,
+      "step": 546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961626544720914e-05,
+      "loss": 0.1633,
+      "step": 547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614380411229693e-05,
+      "loss": 0.1717,
+      "step": 548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612490847635626e-05,
+      "loss": 0.2307,
+      "step": 549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961059675731678e-05,
+      "loss": 0.2534,
+      "step": 550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608698141165114e-05,
+      "loss": 0.3233,
+      "step": 551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9606795000074732e-05,
+      "loss": 0.286,
+      "step": 552
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604887334941863e-05,
+      "loss": 0.2279,
+      "step": 553
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9602975146664863e-05,
+      "loss": 0.193,
+      "step": 554
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601058436144223e-05,
+      "loss": 0.2023,
+      "step": 555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599137204282566e-05,
+      "loss": 0.1558,
+      "step": 556
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597211451984637e-05,
+      "loss": 0.2873,
+      "step": 557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595281180157315e-05,
+      "loss": 0.2274,
+      "step": 558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593346389709603e-05,
+      "loss": 0.2212,
+      "step": 559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.959140708155264e-05,
+      "loss": 0.2337,
+      "step": 560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9589463256599685e-05,
+      "loss": 0.1472,
+      "step": 561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9587514915766124e-05,
+      "loss": 0.1986,
+      "step": 562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585562059969476e-05,
+      "loss": 0.2846,
+      "step": 563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583604690129375e-05,
+      "loss": 0.163,
+      "step": 564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581642807167593e-05,
+      "loss": 0.2724,
+      "step": 565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957967641200802e-05,
+      "loss": 0.2417,
+      "step": 566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9577705505576674e-05,
+      "loss": 0.2193,
+      "step": 567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9575730088801696e-05,
+      "loss": 0.1995,
+      "step": 568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573750162613353e-05,
+      "loss": 0.2505,
+      "step": 569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571765727944034e-05,
+      "loss": 0.199,
+      "step": 570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569776785728248e-05,
+      "loss": 0.2762,
+      "step": 571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956778333690263e-05,
+      "loss": 0.218,
+      "step": 572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565785382405944e-05,
+      "loss": 0.239,
+      "step": 573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9563782923179063e-05,
+      "loss": 0.2525,
+      "step": 574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561775960164986e-05,
+      "loss": 0.2552,
+      "step": 575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9559764494308838e-05,
+      "loss": 0.2068,
+      "step": 576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9557748526557858e-05,
+      "loss": 0.266,
+      "step": 577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.224,
+      "step": 578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553703089170974e-05,
+      "loss": 0.2532,
+      "step": 579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955167362144015e-05,
+      "loss": 0.2153,
+      "step": 580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549639655624657e-05,
+      "loss": 0.2759,
+      "step": 581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547601192682335e-05,
+      "loss": 0.2503,
+      "step": 582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545558233573136e-05,
+      "loss": 0.2258,
+      "step": 583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543510779259135e-05,
+      "loss": 0.2252,
+      "step": 584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954145883070452e-05,
+      "loss": 0.1888,
+      "step": 585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539402388875598e-05,
+      "loss": 0.2282,
+      "step": 586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9537341454740794e-05,
+      "loss": 0.182,
+      "step": 587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535276029270636e-05,
+      "loss": 0.2568,
+      "step": 588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533206113437787e-05,
+      "loss": 0.2605,
+      "step": 589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9531131708217005e-05,
+      "loss": 0.2006,
+      "step": 590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952905281458518e-05,
+      "loss": 0.2883,
+      "step": 591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526969433521298e-05,
+      "loss": 0.2844,
+      "step": 592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952488156600647e-05,
+      "loss": 0.2232,
+      "step": 593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952278921302392e-05,
+      "loss": 0.2985,
+      "step": 594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520692375558982e-05,
+      "loss": 0.1948,
+      "step": 595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518591054599095e-05,
+      "loss": 0.2319,
+      "step": 596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516485251133825e-05,
+      "loss": 0.2649,
+      "step": 597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9514374966154826e-05,
+      "loss": 0.2548,
+      "step": 598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9512260200655885e-05,
+      "loss": 0.2471,
+      "step": 599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.951014095563289e-05,
+      "loss": 0.2135,
+      "step": 600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950801723208383e-05,
+      "loss": 0.2273,
+      "step": 601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950588903100882e-05,
+      "loss": 0.1957,
+      "step": 602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950375635341007e-05,
+      "loss": 0.2191,
+      "step": 603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950161920029191e-05,
+      "loss": 0.2395,
+      "step": 604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949947757266076e-05,
+      "loss": 0.2123,
+      "step": 605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497331471525165e-05,
+      "loss": 0.241,
+      "step": 606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9495180897895766e-05,
+      "loss": 0.2101,
+      "step": 607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9493025852785314e-05,
+      "loss": 0.256,
+      "step": 608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949086633720867e-05,
+      "loss": 0.1585,
+      "step": 609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948870235218279e-05,
+      "loss": 0.2294,
+      "step": 610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9486533898726735e-05,
+      "loss": 0.2536,
+      "step": 611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948436097786169e-05,
+      "loss": 0.3007,
+      "step": 612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9482183590610915e-05,
+      "loss": 0.1515,
+      "step": 613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480001737999796e-05,
+      "loss": 0.2865,
+      "step": 614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947781542105581e-05,
+      "loss": 0.2164,
+      "step": 615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475624640808542e-05,
+      "loss": 0.2196,
+      "step": 616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947342939828968e-05,
+      "loss": 0.1324,
+      "step": 617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9471229694533003e-05,
+      "loss": 0.2269,
+      "step": 618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9469025530574395e-05,
+      "loss": 0.2626,
+      "step": 619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9466816907451857e-05,
+      "loss": 0.2994,
+      "step": 620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9464603826205466e-05,
+      "loss": 0.2491,
+      "step": 621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946238628787741e-05,
+      "loss": 0.2233,
+      "step": 622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460164293511975e-05,
+      "loss": 0.1763,
+      "step": 623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9457937844155546e-05,
+      "loss": 0.2151,
+      "step": 624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455706940856602e-05,
+      "loss": 0.2018,
+      "step": 625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945347158466573e-05,
+      "loss": 0.1834,
+      "step": 626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451231776635598e-05,
+      "loss": 0.3342,
+      "step": 627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9448987517820982e-05,
+      "loss": 0.2636,
+      "step": 628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446738809278753e-05,
+      "loss": 0.299,
+      "step": 629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944448565206787e-05,
+      "loss": 0.1176,
+      "step": 630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94422280472494e-05,
+      "loss": 0.2168,
+      "step": 631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943996599588649e-05,
+      "loss": 0.2264,
+      "step": 632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437699499044388e-05,
+      "loss": 0.2238,
+      "step": 633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943542855779044e-05,
+      "loss": 0.2748,
+      "step": 634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9433153173194074e-05,
+      "loss": 0.33,
+      "step": 635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9430873346326825e-05,
+      "loss": 0.2032,
+      "step": 636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9428589078262306e-05,
+      "loss": 0.2339,
+      "step": 637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9426300370076224e-05,
+      "loss": 0.2559,
+      "step": 638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424007222846388e-05,
+      "loss": 0.1395,
+      "step": 639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9421709637652683e-05,
+      "loss": 0.2033,
+      "step": 640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941940761557709e-05,
+      "loss": 0.2066,
+      "step": 641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417101157703688e-05,
+      "loss": 0.1853,
+      "step": 642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414790265118624e-05,
+      "loss": 0.2348,
+      "step": 643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412474938910157e-05,
+      "loss": 0.182,
+      "step": 644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9410155180168617e-05,
+      "loss": 0.2364,
+      "step": 645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.940783098998643e-05,
+      "loss": 0.2515,
+      "step": 646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9405502369458107e-05,
+      "loss": 0.2459,
+      "step": 647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940316931968024e-05,
+      "loss": 0.256,
+      "step": 648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9400831841751518e-05,
+      "loss": 0.2112,
+      "step": 649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.93984899367727e-05,
+      "loss": 0.2163,
+      "step": 650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396143605846648e-05,
+      "loss": 0.2151,
+      "step": 651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393792850078294e-05,
+      "loss": 0.1824,
+      "step": 652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391437670574656e-05,
+      "loss": 0.2915,
+      "step": 653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389078068444843e-05,
+      "loss": 0.2026,
+      "step": 654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938671404480004e-05,
+      "loss": 0.2107,
+      "step": 655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.938434560075351e-05,
+      "loss": 0.26,
+      "step": 656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9381972737420614e-05,
+      "loss": 0.1695,
+      "step": 657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9379595455918773e-05,
+      "loss": 0.2285,
+      "step": 658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377213757367503e-05,
+      "loss": 0.2467,
+      "step": 659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.93748276428884e-05,
+      "loss": 0.2381,
+      "step": 660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937243711360513e-05,
+      "loss": 0.1073,
+      "step": 661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370042170643445e-05,
+      "loss": 0.2558,
+      "step": 662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367642815131175e-05,
+      "loss": 0.2742,
+      "step": 663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9365239048198227e-05,
+      "loss": 0.2148,
+      "step": 664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362830870976586e-05,
+      "loss": 0.29,
+      "step": 665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9360418284600313e-05,
+      "loss": 0.2567,
+      "step": 666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358001290205542e-05,
+      "loss": 0.1811,
+      "step": 667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355579888930494e-05,
+      "loss": 0.2377,
+      "step": 668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9353154081915456e-05,
+      "loss": 0.1811,
+      "step": 669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935072387030279e-05,
+      "loss": 0.1768,
+      "step": 670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348289255236935e-05,
+      "loss": 0.2804,
+      "step": 671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93458502378644e-05,
+      "loss": 0.2087,
+      "step": 672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343406819333772e-05,
+      "loss": 0.2024,
+      "step": 673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340959000795707e-05,
+      "loss": 0.1409,
+      "step": 674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933850678340294e-05,
+      "loss": 0.1989,
+      "step": 675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933605016831026e-05,
+      "loss": 0.1857,
+      "step": 676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933358915667455e-05,
+      "loss": 0.1968,
+      "step": 677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331123749654748e-05,
+      "loss": 0.1624,
+      "step": 678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328653948411866e-05,
+      "loss": 0.2165,
+      "step": 679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9326179754108985e-05,
+      "loss": 0.2659,
+      "step": 680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932370116791126e-05,
+      "loss": 0.1517,
+      "step": 681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9321218190985906e-05,
+      "loss": 0.1319,
+      "step": 682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318730824502207e-05,
+      "loss": 0.2199,
+      "step": 683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9316239069631516e-05,
+      "loss": 0.1879,
+      "step": 684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931374292754726e-05,
+      "loss": 0.2677,
+      "step": 685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931124239942492e-05,
+      "loss": 0.209,
+      "step": 686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308737486442045e-05,
+      "loss": 0.2503,
+      "step": 687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306228189778255e-05,
+      "loss": 0.2394,
+      "step": 688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930371451061523e-05,
+      "loss": 0.2352,
+      "step": 689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930119645013672e-05,
+      "loss": 0.2172,
+      "step": 690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298674009528522e-05,
+      "loss": 0.1956,
+      "step": 691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929614718997851e-05,
+      "loss": 0.2571,
+      "step": 692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9293615992676626e-05,
+      "loss": 0.2443,
+      "step": 693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291080418814852e-05,
+      "loss": 0.2199,
+      "step": 694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928854046958725e-05,
+      "loss": 0.173,
+      "step": 695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9285996146189932e-05,
+      "loss": 0.1958,
+      "step": 696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9283447449821076e-05,
+      "loss": 0.1917,
+      "step": 697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9280894381680917e-05,
+      "loss": 0.1743,
+      "step": 698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9278336942971748e-05,
+      "loss": 0.2951,
+      "step": 699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927577513489792e-05,
+      "loss": 0.1729,
+      "step": 700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927320895866584e-05,
+      "loss": 0.2633,
+      "step": 701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927063841548398e-05,
+      "loss": 0.1601,
+      "step": 702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9268063506562865e-05,
+      "loss": 0.2568,
+      "step": 703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265484233115065e-05,
+      "loss": 0.2003,
+      "step": 704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9262900596355222e-05,
+      "loss": 0.1339,
+      "step": 705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.926031259750002e-05,
+      "loss": 0.2295,
+      "step": 706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9257720237768208e-05,
+      "loss": 0.2583,
+      "step": 707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9255123518380573e-05,
+      "loss": 0.242,
+      "step": 708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925252244055998e-05,
+      "loss": 0.2866,
+      "step": 709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249917005531315e-05,
+      "loss": 0.1983,
+      "step": 710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9247307214521543e-05,
+      "loss": 0.2867,
+      "step": 711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9244693068759668e-05,
+      "loss": 0.253,
+      "step": 712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924207456947674e-05,
+      "loss": 0.1197,
+      "step": 713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9239451717905878e-05,
+      "loss": 0.2439,
+      "step": 714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236824515282225e-05,
+      "loss": 0.2575,
+      "step": 715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234192962842996e-05,
+      "loss": 0.2119,
+      "step": 716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.923155706182744e-05,
+      "loss": 0.2254,
+      "step": 717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9228916813476855e-05,
+      "loss": 0.2883,
+      "step": 718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9226272219034595e-05,
+      "loss": 0.2242,
+      "step": 719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9223623279746054e-05,
+      "loss": 0.255,
+      "step": 720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922096999685868e-05,
+      "loss": 0.2675,
+      "step": 721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218312371621944e-05,
+      "loss": 0.2379,
+      "step": 722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9215650405287392e-05,
+      "loss": 0.2395,
+      "step": 723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9212984099108594e-05,
+      "loss": 0.2215,
+      "step": 724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9210313454341172e-05,
+      "loss": 0.2381,
+      "step": 725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207638472242785e-05,
+      "loss": 0.2039,
+      "step": 726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9204959154073143e-05,
+      "loss": 0.174,
+      "step": 727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9202275501093994e-05,
+      "loss": 0.1845,
+      "step": 728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9199587514569123e-05,
+      "loss": 0.1511,
+      "step": 729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196895195764363e-05,
+      "loss": 0.2334,
+      "step": 730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194198545947583e-05,
+      "loss": 0.1633,
+      "step": 731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9191497566388685e-05,
+      "loss": 0.2526,
+      "step": 732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918879225835963e-05,
+      "loss": 0.1512,
+      "step": 733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9186082623134397e-05,
+      "loss": 0.1765,
+      "step": 734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183368661989014e-05,
+      "loss": 0.2725,
+      "step": 735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9180650376201536e-05,
+      "loss": 0.3132,
+      "step": 736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9177927767052067e-05,
+      "loss": 0.1953,
+      "step": 737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917520083582274e-05,
+      "loss": 0.3097,
+      "step": 738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917246958379772e-05,
+      "loss": 0.2597,
+      "step": 739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916973401226322e-05,
+      "loss": 0.1449,
+      "step": 740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9166994122507474e-05,
+      "loss": 0.2563,
+      "step": 741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164249915820753e-05,
+      "loss": 0.1339,
+      "step": 742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916150139349536e-05,
+      "loss": 0.2648,
+      "step": 743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9158748556825637e-05,
+      "loss": 0.2707,
+      "step": 744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9155991407107947e-05,
+      "loss": 0.2359,
+      "step": 745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9153229945640695e-05,
+      "loss": 0.1754,
+      "step": 746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9150464173724313e-05,
+      "loss": 0.2046,
+      "step": 747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9147694092661254e-05,
+      "loss": 0.2399,
+      "step": 748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9144919703756012e-05,
+      "loss": 0.1558,
+      "step": 749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914214100831511e-05,
+      "loss": 0.1825,
+      "step": 750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9139358007647085e-05,
+      "loss": 0.223,
+      "step": 751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136570703062517e-05,
+      "loss": 0.3473,
+      "step": 752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9133779095874005e-05,
+      "loss": 0.1916,
+      "step": 753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9130983187396174e-05,
+      "loss": 0.205,
+      "step": 754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128182978945676e-05,
+      "loss": 0.1739,
+      "step": 755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9125378471841195e-05,
+      "loss": 0.2554,
+      "step": 756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9122569667403423e-05,
+      "loss": 0.2303,
+      "step": 757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119756566955092e-05,
+      "loss": 0.2325,
+      "step": 758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9116939171820945e-05,
+      "loss": 0.2239,
+      "step": 759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.911411748332776e-05,
+      "loss": 0.1511,
+      "step": 760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111291502804324e-05,
+      "loss": 0.2278,
+      "step": 761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9108461231581456e-05,
+      "loss": 0.1991,
+      "step": 762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105626670991982e-05,
+      "loss": 0.2435,
+      "step": 763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9102787822370764e-05,
+      "loss": 0.1997,
+      "step": 764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909994468705467e-05,
+      "loss": 0.2949,
+      "step": 765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9097097266382598e-05,
+      "loss": 0.2338,
+      "step": 766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094245561695456e-05,
+      "loss": 0.1431,
+      "step": 767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9091389574336173e-05,
+      "loss": 0.2446,
+      "step": 768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088529305649692e-05,
+      "loss": 0.2301,
+      "step": 769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9085664756982975e-05,
+      "loss": 0.1839,
+      "step": 770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082795929684995e-05,
+      "loss": 0.243,
+      "step": 771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907992282510675e-05,
+      "loss": 0.2199,
+      "step": 772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907704544460124e-05,
+      "loss": 0.1874,
+      "step": 773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907416378952349e-05,
+      "loss": 0.2215,
+      "step": 774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9071277861230522e-05,
+      "loss": 0.2163,
+      "step": 775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068387661081388e-05,
+      "loss": 0.2315,
+      "step": 776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9065493190437143e-05,
+      "loss": 0.1907,
+      "step": 777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062594450660857e-05,
+      "loss": 0.154,
+      "step": 778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9059691443117604e-05,
+      "loss": 0.2222,
+      "step": 779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056784169174472e-05,
+      "loss": 0.3018,
+      "step": 780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9053872630200557e-05,
+      "loss": 0.155,
+      "step": 781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905095682756697e-05,
+      "loss": 0.2707,
+      "step": 782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048036762646812e-05,
+      "loss": 0.1981,
+      "step": 783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9045112436815217e-05,
+      "loss": 0.2387,
+      "step": 784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90421838514493e-05,
+      "loss": 0.2239,
+      "step": 785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.90392510079282e-05,
+      "loss": 0.334,
+      "step": 786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036313907633057e-05,
+      "loss": 0.3155,
+      "step": 787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903337255194701e-05,
+      "loss": 0.1952,
+      "step": 788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030426942255203e-05,
+      "loss": 0.198,
+      "step": 789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902747707994479e-05,
+      "loss": 0.2053,
+      "step": 790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9024522966404922e-05,
+      "loss": 0.2842,
+      "step": 791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9021564603026756e-05,
+      "loss": 0.1668,
+      "step": 792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901860199120344e-05,
+      "loss": 0.2289,
+      "step": 793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901563513233014e-05,
+      "loss": 0.1892,
+      "step": 794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012664027804006e-05,
+      "loss": 0.2168,
+      "step": 795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.2065,
+      "step": 796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006709087391862e-05,
+      "loss": 0.2724,
+      "step": 797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900372525431016e-05,
+      "loss": 0.2222,
+      "step": 798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9000737181184237e-05,
+      "loss": 0.2567,
+      "step": 799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8997744869421248e-05,
+      "loss": 0.2176,
+      "step": 800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994748320430322e-05,
+      "loss": 0.2077,
+      "step": 801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8991747535622607e-05,
+      "loss": 0.2339,
+      "step": 802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8988742516411236e-05,
+      "loss": 0.2764,
+      "step": 803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8985733264211332e-05,
+      "loss": 0.221,
+      "step": 804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982719780440016e-05,
+      "loss": 0.2377,
+      "step": 805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979702066516402e-05,
+      "loss": 0.2091,
+      "step": 806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8976680123861593e-05,
+      "loss": 0.2178,
+      "step": 807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897365395389869e-05,
+      "loss": 0.1977,
+      "step": 808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8970623558052777e-05,
+      "loss": 0.1928,
+      "step": 809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967588937750933e-05,
+      "loss": 0.2522,
+      "step": 810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8964550094422223e-05,
+      "loss": 0.2217,
+      "step": 811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961507029497703e-05,
+      "loss": 0.2625,
+      "step": 812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895845974441042e-05,
+      "loss": 0.2639,
+      "step": 813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955408240595396e-05,
+      "loss": 0.2233,
+      "step": 814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8952352519489657e-05,
+      "loss": 0.2391,
+      "step": 815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.89492925825322e-05,
+      "loss": 0.2583,
+      "step": 816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8946228431164024e-05,
+      "loss": 0.2611,
+      "step": 817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8943160066828093e-05,
+      "loss": 0.1655,
+      "step": 818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894008749096937e-05,
+      "loss": 0.2456,
+      "step": 819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8937010705034788e-05,
+      "loss": 0.1727,
+      "step": 820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893392971047328e-05,
+      "loss": 0.2323,
+      "step": 821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8930844508735746e-05,
+      "loss": 0.3005,
+      "step": 822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927755101275076e-05,
+      "loss": 0.1498,
+      "step": 823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8924661489546136e-05,
+      "loss": 0.1889,
+      "step": 824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921563675005767e-05,
+      "loss": 0.2749,
+      "step": 825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8918461659112805e-05,
+      "loss": 0.201,
+      "step": 826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8915355443328044e-05,
+      "loss": 0.2413,
+      "step": 827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.891224502911428e-05,
+      "loss": 0.1898,
+      "step": 828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890913041793626e-05,
+      "loss": 0.2355,
+      "step": 829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890601161126073e-05,
+      "loss": 0.18,
+      "step": 830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902888610556394e-05,
+      "loss": 0.1571,
+      "step": 831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8899761417293944e-05,
+      "loss": 0.1948,
+      "step": 832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8896630032946034e-05,
+      "loss": 0.1827,
+      "step": 833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889349445898731e-05,
+      "loss": 0.25,
+      "step": 834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8890354696894374e-05,
+      "loss": 0.2897,
+      "step": 835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8887210748145806e-05,
+      "loss": 0.2799,
+      "step": 836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8884062614222158e-05,
+      "loss": 0.2419,
+      "step": 837
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880910296605956e-05,
+      "loss": 0.2256,
+      "step": 838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8877753796781686e-05,
+      "loss": 0.2675,
+      "step": 839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887459311623582e-05,
+      "loss": 0.1583,
+      "step": 840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871428256456776e-05,
+      "loss": 0.2585,
+      "step": 841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886825921893497e-05,
+      "loss": 0.2943,
+      "step": 842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8865086005162752e-05,
+      "loss": 0.2759,
+      "step": 843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861908616634465e-05,
+      "loss": 0.2863,
+      "step": 844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8858727054846407e-05,
+      "loss": 0.1721,
+      "step": 845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885554132129684e-05,
+      "loss": 0.2936,
+      "step": 846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852351417485994e-05,
+      "loss": 0.2732,
+      "step": 847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8849157344916066e-05,
+      "loss": 0.2513,
+      "step": 848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884595910509121e-05,
+      "loss": 0.2284,
+      "step": 849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884275669951754e-05,
+      "loss": 0.2717,
+      "step": 850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883955012970314e-05,
+      "loss": 0.1417,
+      "step": 851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8836339397158053e-05,
+      "loss": 0.1753,
+      "step": 852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883312450339428e-05,
+      "loss": 0.2414,
+      "step": 853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882990544992578e-05,
+      "loss": 0.2104,
+      "step": 854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8826682238268475e-05,
+      "loss": 0.1177,
+      "step": 855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8823454869940243e-05,
+      "loss": 0.2969,
+      "step": 856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882022334646092e-05,
+      "loss": 0.1657,
+      "step": 857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88169876693523e-05,
+      "loss": 0.2481,
+      "step": 858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813747840138127e-05,
+      "loss": 0.2225,
+      "step": 859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810503860344115e-05,
+      "loss": 0.179,
+      "step": 860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880725573149791e-05,
+      "loss": 0.2777,
+      "step": 861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880400345512913e-05,
+      "loss": 0.1529,
+      "step": 862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8800747032769343e-05,
+      "loss": 0.1918,
+      "step": 863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.879748646595206e-05,
+      "loss": 0.1635,
+      "step": 864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8794221756212758e-05,
+      "loss": 0.2872,
+      "step": 865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790952905088855e-05,
+      "loss": 0.2741,
+      "step": 866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8787679914119724e-05,
+      "loss": 0.21,
+      "step": 867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784402784846683e-05,
+      "loss": 0.2003,
+      "step": 868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8781121518813e-05,
+      "loss": 0.2486,
+      "step": 869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777836117563894e-05,
+      "loss": 0.2714,
+      "step": 870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877454658264653e-05,
+      "loss": 0.2087,
+      "step": 871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8771252915610022e-05,
+      "loss": 0.217,
+      "step": 872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8767955118005422e-05,
+      "loss": 0.2199,
+      "step": 873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8764653191385737e-05,
+      "loss": 0.241,
+      "step": 874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761347137305912e-05,
+      "loss": 0.1888,
+      "step": 875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8758036957322838e-05,
+      "loss": 0.2213,
+      "step": 876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8754722652995346e-05,
+      "loss": 0.1889,
+      "step": 877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8751404225884218e-05,
+      "loss": 0.2096,
+      "step": 878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8748081677552167e-05,
+      "loss": 0.3046,
+      "step": 879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874475500956385e-05,
+      "loss": 0.1992,
+      "step": 880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8741424223485868e-05,
+      "loss": 0.1841,
+      "step": 881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738089320886756e-05,
+      "loss": 0.3035,
+      "step": 882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.873475030333699e-05,
+      "loss": 0.3172,
+      "step": 883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873140717240899e-05,
+      "loss": 0.2376,
+      "step": 884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8728059929677097e-05,
+      "loss": 0.2366,
+      "step": 885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8724708576717607e-05,
+      "loss": 0.2627,
+      "step": 886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8721353115108736e-05,
+      "loss": 0.1721,
+      "step": 887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871799354643065e-05,
+      "loss": 0.2075,
+      "step": 888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871462987226543e-05,
+      "loss": 0.1778,
+      "step": 889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871126209419711e-05,
+      "loss": 0.251,
+      "step": 890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707890213811645e-05,
+      "loss": 0.2195,
+      "step": 891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870451423269692e-05,
+      "loss": 0.2323,
+      "step": 892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8701134152442765e-05,
+      "loss": 0.1749,
+      "step": 893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8697749974640924e-05,
+      "loss": 0.277,
+      "step": 894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8694361700885078e-05,
+      "loss": 0.175,
+      "step": 895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869096933277084e-05,
+      "loss": 0.1739,
+      "step": 896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8687572871895747e-05,
+      "loss": 0.2708,
+      "step": 897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8684172319859258e-05,
+      "loss": 0.2234,
+      "step": 898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8680767678262777e-05,
+      "loss": 0.1684,
+      "step": 899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867735894870961e-05,
+      "loss": 0.2434,
+      "step": 900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8673946132805005e-05,
+      "loss": 0.2692,
+      "step": 901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8670529232156128e-05,
+      "loss": 0.2459,
+      "step": 902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866710824837207e-05,
+      "loss": 0.2083,
+      "step": 903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8663683183063846e-05,
+      "loss": 0.1723,
+      "step": 904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.248,
+      "step": 905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656820814328557e-05,
+      "loss": 0.3038,
+      "step": 906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8653383514133122e-05,
+      "loss": 0.1775,
+      "step": 907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8649942138876792e-05,
+      "loss": 0.1314,
+      "step": 908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646496690180176e-05,
+      "loss": 0.2604,
+      "step": 909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864304716966581e-05,
+      "loss": 0.1166,
+      "step": 910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639593578958144e-05,
+      "loss": 0.2284,
+      "step": 911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863613591968355e-05,
+      "loss": 0.2474,
+      "step": 912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8632674193470312e-05,
+      "loss": 0.2478,
+      "step": 913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862920840194863e-05,
+      "loss": 0.2581,
+      "step": 914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625738546750617e-05,
+      "loss": 0.1997,
+      "step": 915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86222646295103e-05,
+      "loss": 0.2053,
+      "step": 916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8618786651863624e-05,
+      "loss": 0.2651,
+      "step": 917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861530461544844e-05,
+      "loss": 0.1789,
+      "step": 918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861181852190451e-05,
+      "loss": 0.3144,
+      "step": 919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860832837287351e-05,
+      "loss": 0.3016,
+      "step": 920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860483416999903e-05,
+      "loss": 0.2042,
+      "step": 921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601335914926558e-05,
+      "loss": 0.2322,
+      "step": 922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8597833609303498e-05,
+      "loss": 0.2527,
+      "step": 923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859432725477916e-05,
+      "loss": 0.2603,
+      "step": 924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859081685300476e-05,
+      "loss": 0.2567,
+      "step": 925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858730240563342e-05,
+      "loss": 0.1331,
+      "step": 926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8583783914320164e-05,
+      "loss": 0.186,
+      "step": 927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8580261380721932e-05,
+      "loss": 0.2263,
+      "step": 928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8576734806497553e-05,
+      "loss": 0.2849,
+      "step": 929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8573204193307767e-05,
+      "loss": 0.2365,
+      "step": 930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569669542815214e-05,
+      "loss": 0.2442,
+      "step": 931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8566130856684437e-05,
+      "loss": 0.1752,
+      "step": 932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856258813658188e-05,
+      "loss": 0.2461,
+      "step": 933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8559041384175874e-05,
+      "loss": 0.1732,
+      "step": 934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8555490601136675e-05,
+      "loss": 0.2511,
+      "step": 935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855193578913641e-05,
+      "loss": 0.257,
+      "step": 936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548376949849118e-05,
+      "loss": 0.2066,
+      "step": 937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854481408495073e-05,
+      "loss": 0.1873,
+      "step": 938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854124719611908e-05,
+      "loss": 0.1866,
+      "step": 939
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8537676285033886e-05,
+      "loss": 0.1633,
+      "step": 940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853410135337677e-05,
+      "loss": 0.1886,
+      "step": 941
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530522402831236e-05,
+      "loss": 0.2666,
+      "step": 942
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526939435082692e-05,
+      "loss": 0.3228,
+      "step": 943
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8523352451818435e-05,
+      "loss": 0.2558,
+      "step": 944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8519761454727647e-05,
+      "loss": 0.1822,
+      "step": 945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8516166445501405e-05,
+      "loss": 0.1559,
+      "step": 946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8512567425832674e-05,
+      "loss": 0.2212,
+      "step": 947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508964397416312e-05,
+      "loss": 0.1453,
+      "step": 948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.850535736194906e-05,
+      "loss": 0.151,
+      "step": 949
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501746321129542e-05,
+      "loss": 0.1495,
+      "step": 950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849813127665828e-05,
+      "loss": 0.1669,
+      "step": 951
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.849451223023767e-05,
+      "loss": 0.3217,
+      "step": 952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8490889183572002e-05,
+      "loss": 0.2902,
+      "step": 953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848726213836744e-05,
+      "loss": 0.1078,
+      "step": 954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848363109633204e-05,
+      "loss": 0.2159,
+      "step": 955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8479996059175733e-05,
+      "loss": 0.2091,
+      "step": 956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8476357028610337e-05,
+      "loss": 0.1812,
+      "step": 957
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8472714006349554e-05,
+      "loss": 0.1144,
+      "step": 958
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469066994108946e-05,
+      "loss": 0.2114,
+      "step": 959
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846541599360598e-05,
+      "loss": 0.1592,
+      "step": 960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461761006559982e-05,
+      "loss": 0.2239,
+      "step": 961
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845810203469217e-05,
+      "loss": 0.2338,
+      "step": 962
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845443907972562e-05,
+      "loss": 0.1533,
+      "step": 963
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845077214338531e-05,
+      "loss": 0.1817,
+      "step": 964
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8447101227398064e-05,
+      "loss": 0.1547,
+      "step": 965
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.84434263334926e-05,
+      "loss": 0.1786,
+      "step": 966
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439747463399502e-05,
+      "loss": 0.1741,
+      "step": 967
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8436064618851225e-05,
+      "loss": 0.1764,
+      "step": 968
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8432377801582105e-05,
+      "loss": 0.2297,
+      "step": 969
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8428687013328338e-05,
+      "loss": 0.179,
+      "step": 970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842499225582799e-05,
+      "loss": 0.1642,
+      "step": 971
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842129353082101e-05,
+      "loss": 0.1848,
+      "step": 972
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8417590840049192e-05,
+      "loss": 0.2083,
+      "step": 973
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413884185256225e-05,
+      "loss": 0.1362,
+      "step": 974
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8410173568187646e-05,
+      "loss": 0.1521,
+      "step": 975
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840645899059086e-05,
+      "loss": 0.2395,
+      "step": 976
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8402740454215142e-05,
+      "loss": 0.1949,
+      "step": 977
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8399017960811628e-05,
+      "loss": 0.1714,
+      "step": 978
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8395291512133323e-05,
+      "loss": 0.2095,
+      "step": 979
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8391561109935083e-05,
+      "loss": 0.1513,
+      "step": 980
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838782675597364e-05,
+      "loss": 0.1316,
+      "step": 981
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838408845200758e-05,
+      "loss": 0.1619,
+      "step": 982
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8380346199797342e-05,
+      "loss": 0.2289,
+      "step": 983
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837660000110524e-05,
+      "loss": 0.1923,
+      "step": 984
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837284985769543e-05,
+      "loss": 0.2156,
+      "step": 985
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8369095771333938e-05,
+      "loss": 0.2275,
+      "step": 986
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8365337743788647e-05,
+      "loss": 0.2615,
+      "step": 987
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836157577682928e-05,
+      "loss": 0.2084,
+      "step": 988
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8357809872227438e-05,
+      "loss": 0.168,
+      "step": 989
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835404003175656e-05,
+      "loss": 0.2349,
+      "step": 990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350266257191945e-05,
+      "loss": 0.1903,
+      "step": 991
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834648855031074e-05,
+      "loss": 0.1176,
+      "step": 992
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834270691289195e-05,
+      "loss": 0.2567,
+      "step": 993
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8338921346716426e-05,
+      "loss": 0.1809,
+      "step": 994
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833513185356687e-05,
+      "loss": 0.2336,
+      "step": 995
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8331338435227838e-05,
+      "loss": 0.1463,
+      "step": 996
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832754109348573e-05,
+      "loss": 0.2284,
+      "step": 997
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8323739830128787e-05,
+      "loss": 0.1043,
+      "step": 998
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8319934646947116e-05,
+      "loss": 0.2044,
+      "step": 999
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316125545732653e-05,
+      "loss": 0.2516,
+      "step": 1000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8312312528279183e-05,
+      "loss": 0.2281,
+      "step": 1001
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830849559638233e-05,
+      "loss": 0.1337,
+      "step": 1002
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8304674751839583e-05,
+      "loss": 0.146,
+      "step": 1003
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8300849996450244e-05,
+      "loss": 0.1596,
+      "step": 1004
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8297021332015477e-05,
+      "loss": 0.227,
+      "step": 1005
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8293188760338285e-05,
+      "loss": 0.1816,
+      "step": 1006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.82893522832235e-05,
+      "loss": 0.1425,
+      "step": 1007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8285511902477803e-05,
+      "loss": 0.1286,
+      "step": 1008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.828166761990971e-05,
+      "loss": 0.1988,
+      "step": 1009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277819437329577e-05,
+      "loss": 0.1711,
+      "step": 1010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8273967356549586e-05,
+      "loss": 0.1694,
+      "step": 1011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8270111379383773e-05,
+      "loss": 0.235,
+      "step": 1012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266251507647998e-05,
+      "loss": 0.1782,
+      "step": 1013
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.1727,
+      "step": 1014
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8258520087739162e-05,
+      "loss": 0.1667,
+      "step": 1015
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825464854320699e-05,
+      "loss": 0.1685,
+      "step": 1016
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8250773111386633e-05,
+      "loss": 0.2425,
+      "step": 1017
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8246893794103113e-05,
+      "loss": 0.1615,
+      "step": 1018
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.824301059318327e-05,
+      "loss": 0.1352,
+      "step": 1019
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82391235104558e-05,
+      "loss": 0.1298,
+      "step": 1020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8235232547751205e-05,
+      "loss": 0.1515,
+      "step": 1021
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8231337706901823e-05,
+      "loss": 0.1438,
+      "step": 1022
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822743898974182e-05,
+      "loss": 0.2162,
+      "step": 1023
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8223536398107177e-05,
+      "loss": 0.18,
+      "step": 1024
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821962993383571e-05,
+      "loss": 0.2566,
+      "step": 1025
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821571959876706e-05,
+      "loss": 0.2266,
+      "step": 1026
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8211805394742684e-05,
+      "loss": 0.2228,
+      "step": 1027
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8207887323605858e-05,
+      "loss": 0.1526,
+      "step": 1028
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820396538720169e-05,
+      "loss": 0.184,
+      "step": 1029
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.82000395873771e-05,
+      "loss": 0.2008,
+      "step": 1030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819610992598083e-05,
+      "loss": 0.1749,
+      "step": 1031
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8192176404863442e-05,
+      "loss": 0.142,
+      "step": 1032
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8188239025877314e-05,
+      "loss": 0.2354,
+      "step": 1033
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8184297790876638e-05,
+      "loss": 0.1534,
+      "step": 1034
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818035270171743e-05,
+      "loss": 0.2647,
+      "step": 1035
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817640376025751e-05,
+      "loss": 0.1839,
+      "step": 1036
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8172450968356525e-05,
+      "loss": 0.179,
+      "step": 1037
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8168494327875918e-05,
+      "loss": 0.1607,
+      "step": 1038
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8164533840678962e-05,
+      "loss": 0.1545,
+      "step": 1039
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8160569508630734e-05,
+      "loss": 0.1927,
+      "step": 1040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815660133359812e-05,
+      "loss": 0.2506,
+      "step": 1041
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8152629317449814e-05,
+      "loss": 0.1856,
+      "step": 1042
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.814865346205633e-05,
+      "loss": 0.1166,
+      "step": 1043
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8144673769289972e-05,
+      "loss": 0.1759,
+      "step": 1044
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8140690241024872e-05,
+      "loss": 0.102,
+      "step": 1045
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813670287913695e-05,
+      "loss": 0.1885,
+      "step": 1046
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813271168550394e-05,
+      "loss": 0.2423,
+      "step": 1047
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8128716662005385e-05,
+      "loss": 0.1462,
+      "step": 1048
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8124717810522615e-05,
+      "loss": 0.1958,
+      "step": 1049
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8120715132938784e-05,
+      "loss": 0.2131,
+      "step": 1050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8116708631138832e-05,
+      "loss": 0.2812,
+      "step": 1051
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8112698307009506e-05,
+      "loss": 0.1929,
+      "step": 1052
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8108684162439355e-05,
+      "loss": 0.2155,
+      "step": 1053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810466619931872e-05,
+      "loss": 0.2008,
+      "step": 1054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810064441953975e-05,
+      "loss": 0.1624,
+      "step": 1055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809661882499638e-05,
+      "loss": 0.1933,
+      "step": 1056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8092589417584353e-05,
+      "loss": 0.2056,
+      "step": 1057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8088556199201195e-05,
+      "loss": 0.2115,
+      "step": 1058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8084519171746238e-05,
+      "loss": 0.2382,
+      "step": 1059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8080478337120604e-05,
+      "loss": 0.2028,
+      "step": 1060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807643369722721e-05,
+      "loss": 0.1838,
+      "step": 1061
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8072385253970756e-05,
+      "loss": 0.1671,
+      "step": 1062
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8068333009257743e-05,
+      "loss": 0.166,
+      "step": 1063
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8064276964996458e-05,
+      "loss": 0.1328,
+      "step": 1064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8060217123096978e-05,
+      "loss": 0.2084,
+      "step": 1065
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8056153485471167e-05,
+      "loss": 0.2001,
+      "step": 1066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8052086054032677e-05,
+      "loss": 0.1572,
+      "step": 1067
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804801483069695e-05,
+      "loss": 0.186,
+      "step": 1068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804393981738121e-05,
+      "loss": 0.1918,
+      "step": 1069
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8039861016004464e-05,
+      "loss": 0.198,
+      "step": 1070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.803577842848751e-05,
+      "loss": 0.1193,
+      "step": 1071
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8031692056752926e-05,
+      "loss": 0.183,
+      "step": 1072
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8027601902725065e-05,
+      "loss": 0.2131,
+      "step": 1073
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8023507968330068e-05,
+      "loss": 0.2124,
+      "step": 1074
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019410255495853e-05,
+      "loss": 0.2205,
+      "step": 1075
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801530876615213e-05,
+      "loss": 0.1796,
+      "step": 1076
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8011203502230367e-05,
+      "loss": 0.1497,
+      "step": 1077
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8007094465663823e-05,
+      "loss": 0.2957,
+      "step": 1078
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8002981658387527e-05,
+      "loss": 0.1551,
+      "step": 1079
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.799886508233829e-05,
+      "loss": 0.2233,
+      "step": 1080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7994744739454694e-05,
+      "loss": 0.1808,
+      "step": 1081
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990620631677092e-05,
+      "loss": 0.1362,
+      "step": 1082
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798649276094762e-05,
+      "loss": 0.2352,
+      "step": 1083
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7982361129210172e-05,
+      "loss": 0.1662,
+      "step": 1084
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797822573841043e-05,
+      "loss": 0.2147,
+      "step": 1085
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7974086590495825e-05,
+      "loss": 0.1822,
+      "step": 1086
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7969943687415575e-05,
+      "loss": 0.1526,
+      "step": 1087
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7965797031120667e-05,
+      "loss": 0.123,
+      "step": 1088
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7961646623563836e-05,
+      "loss": 0.1602,
+      "step": 1089
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7957492466699606e-05,
+      "loss": 0.1471,
+      "step": 1090
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7953334562484252e-05,
+      "loss": 0.2228,
+      "step": 1091
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7949172912875824e-05,
+      "loss": 0.151,
+      "step": 1092
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794500751983413e-05,
+      "loss": 0.2184,
+      "step": 1093
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7940838385320732e-05,
+      "loss": 0.2351,
+      "step": 1094
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7936665511298975e-05,
+      "loss": 0.1664,
+      "step": 1095
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793248889973395e-05,
+      "loss": 0.1883,
+      "step": 1096
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928308552592505e-05,
+      "loss": 0.1517,
+      "step": 1097
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7924124471843263e-05,
+      "loss": 0.1717,
+      "step": 1098
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919936659456593e-05,
+      "loss": 0.2189,
+      "step": 1099
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.791574511740462e-05,
+      "loss": 0.1749,
+      "step": 1100
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911549847661236e-05,
+      "loss": 0.221,
+      "step": 1101
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7907350852202078e-05,
+      "loss": 0.1535,
+      "step": 1102
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790314813300454e-05,
+      "loss": 0.1858,
+      "step": 1103
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789894169204778e-05,
+      "loss": 0.1389,
+      "step": 1104
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7894731531312687e-05,
+      "loss": 0.1992,
+      "step": 1105
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7890517652781923e-05,
+      "loss": 0.1386,
+      "step": 1106
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7886300058439888e-05,
+      "loss": 0.1521,
+      "step": 1107
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788207875027274e-05,
+      "loss": 0.1687,
+      "step": 1108
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7877853730268387e-05,
+      "loss": 0.1489,
+      "step": 1109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787362500041647e-05,
+      "loss": 0.2221,
+      "step": 1110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786939256270839e-05,
+      "loss": 0.1386,
+      "step": 1111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786515641913729e-05,
+      "loss": 0.2491,
+      "step": 1112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.786091657169806e-05,
+      "loss": 0.1863,
+      "step": 1113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.785667302238734e-05,
+      "loss": 0.1729,
+      "step": 1114
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.78524257732035e-05,
+      "loss": 0.2967,
+      "step": 1115
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.784817482614666e-05,
+      "loss": 0.1469,
+      "step": 1116
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7843920183218683e-05,
+      "loss": 0.2097,
+      "step": 1117
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783966184642317e-05,
+      "loss": 0.1686,
+      "step": 1118
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.783539981776546e-05,
+      "loss": 0.1789,
+      "step": 1119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7831134099252633e-05,
+      "loss": 0.2391,
+      "step": 1120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7826864692893505e-05,
+      "loss": 0.1761,
+      "step": 1121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7822591600698632e-05,
+      "loss": 0.1705,
+      "step": 1122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.1633,
+      "step": 1123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7814034366852537e-05,
+      "loss": 0.1674,
+      "step": 1124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809750229231097e-05,
+      "loss": 0.1486,
+      "step": 1125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780546241383347e-05,
+      "loss": 0.1662,
+      "step": 1126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7801170922678884e-05,
+      "loss": 0.1908,
+      "step": 1127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779687575778829e-05,
+      "loss": 0.2384,
+      "step": 1128
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7792576921184374e-05,
+      "loss": 0.2198,
+      "step": 1129
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7788274414891543e-05,
+      "loss": 0.209,
+      "step": 1130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.778396824093594e-05,
+      "loss": 0.1556,
+      "step": 1131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7779658401345437e-05,
+      "loss": 0.168,
+      "step": 1132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7775344898149627e-05,
+      "loss": 0.1908,
+      "step": 1133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.777102773337983e-05,
+      "loss": 0.2064,
+      "step": 1134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7766706909069088e-05,
+      "loss": 0.1966,
+      "step": 1135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.776238242725217e-05,
+      "loss": 0.2211,
+      "step": 1136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775805428996556e-05,
+      "loss": 0.178,
+      "step": 1137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775372249924748e-05,
+      "loss": 0.2327,
+      "step": 1138
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7749387057137852e-05,
+      "loss": 0.167,
+      "step": 1139
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.774504796567833e-05,
+      "loss": 0.1215,
+      "step": 1140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7740705226912288e-05,
+      "loss": 0.2067,
+      "step": 1141
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773635884288481e-05,
+      "loss": 0.1834,
+      "step": 1142
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.77320088156427e-05,
+      "loss": 0.1623,
+      "step": 1143
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772765514723448e-05,
+      "loss": 0.2101,
+      "step": 1144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7723297839710377e-05,
+      "loss": 0.1461,
+      "step": 1145
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718936895122344e-05,
+      "loss": 0.1808,
+      "step": 1146
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7714572315524043e-05,
+      "loss": 0.1813,
+      "step": 1147
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7710204102970844e-05,
+      "loss": 0.1977,
+      "step": 1148
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7705832259519836e-05,
+      "loss": 0.1357,
+      "step": 1149
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7701456787229805e-05,
+      "loss": 0.263,
+      "step": 1150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769707768816126e-05,
+      "loss": 0.1906,
+      "step": 1151
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7692694964376404e-05,
+      "loss": 0.1784,
+      "step": 1152
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7688308617939158e-05,
+      "loss": 0.1637,
+      "step": 1153
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7683918650915148e-05,
+      "loss": 0.2326,
+      "step": 1154
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679525065371696e-05,
+      "loss": 0.1456,
+      "step": 1155
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767512786337784e-05,
+      "loss": 0.164,
+      "step": 1156
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.767072704700431e-05,
+      "loss": 0.2608,
+      "step": 1157
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7666322618323546e-05,
+      "loss": 0.1722,
+      "step": 1158
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7661914579409685e-05,
+      "loss": 0.1706,
+      "step": 1159
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7657502932338568e-05,
+      "loss": 0.1832,
+      "step": 1160
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7653087679187734e-05,
+      "loss": 0.1584,
+      "step": 1161
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764866882203641e-05,
+      "loss": 0.1614,
+      "step": 1162
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764424636296554e-05,
+      "loss": 0.1261,
+      "step": 1163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7639820304057745e-05,
+      "loss": 0.1498,
+      "step": 1164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7635390647397358e-05,
+      "loss": 0.2221,
+      "step": 1165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7630957395070387e-05,
+      "loss": 0.1589,
+      "step": 1166
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762652054916455e-05,
+      "loss": 0.2294,
+      "step": 1167
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622080111769257e-05,
+      "loss": 0.212,
+      "step": 1168
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7617636084975596e-05,
+      "loss": 0.1567,
+      "step": 1169
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7613188470876356e-05,
+      "loss": 0.1317,
+      "step": 1170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7608737271566004e-05,
+      "loss": 0.1504,
+      "step": 1171
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7604282489140718e-05,
+      "loss": 0.154,
+      "step": 1172
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759982412569834e-05,
+      "loss": 0.2057,
+      "step": 1173
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759536218333841e-05,
+      "loss": 0.214,
+      "step": 1174
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7590896664162153e-05,
+      "loss": 0.1425,
+      "step": 1175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.758642757027247e-05,
+      "loss": 0.218,
+      "step": 1176
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7581954903773956e-05,
+      "loss": 0.1811,
+      "step": 1177
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7577478666772886e-05,
+      "loss": 0.2114,
+      "step": 1178
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7572998861377203e-05,
+      "loss": 0.1814,
+      "step": 1179
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7568515489696558e-05,
+      "loss": 0.1954,
+      "step": 1180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7564028553842257e-05,
+      "loss": 0.2306,
+      "step": 1181
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7559538055927293e-05,
+      "loss": 0.1534,
+      "step": 1182
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7555043998066333e-05,
+      "loss": 0.1142,
+      "step": 1183
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7550546382375734e-05,
+      "loss": 0.1609,
+      "step": 1184
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754604521097351e-05,
+      "loss": 0.2223,
+      "step": 1185
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7541540485979357e-05,
+      "loss": 0.1113,
+      "step": 1186
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7537032209514646e-05,
+      "loss": 0.1834,
+      "step": 1187
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7532520383702425e-05,
+      "loss": 0.2105,
+      "step": 1188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.75280050106674e-05,
+      "loss": 0.1835,
+      "step": 1189
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7523486092535962e-05,
+      "loss": 0.2281,
+      "step": 1190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7518963631436162e-05,
+      "loss": 0.1854,
+      "step": 1191
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751443762949772e-05,
+      "loss": 0.2027,
+      "step": 1192
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7509908088852033e-05,
+      "loss": 0.1797,
+      "step": 1193
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750537501163215e-05,
+      "loss": 0.1138,
+      "step": 1194
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7500838399972794e-05,
+      "loss": 0.1661,
+      "step": 1195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7496298256010353e-05,
+      "loss": 0.218,
+      "step": 1196
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7491754581882875e-05,
+      "loss": 0.1635,
+      "step": 1197
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7487207379730078e-05,
+      "loss": 0.1984,
+      "step": 1198
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7482656651693327e-05,
+      "loss": 0.1902,
+      "step": 1199
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747810239991566e-05,
+      "loss": 0.1322,
+      "step": 1200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747354462654177e-05,
+      "loss": 0.1617,
+      "step": 1201
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468983333718006e-05,
+      "loss": 0.2222,
+      "step": 1202
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7464418523592384e-05,
+      "loss": 0.183,
+      "step": 1203
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7459850198314562e-05,
+      "loss": 0.1902,
+      "step": 1204
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455278360035862e-05,
+      "loss": 0.248,
+      "step": 1205
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7450703010909263e-05,
+      "loss": 0.2201,
+      "step": 1206
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7446124153089392e-05,
+      "loss": 0.2271,
+      "step": 1207
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7441541788732528e-05,
+      "loss": 0.2159,
+      "step": 1208
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7436955919996602e-05,
+      "loss": 0.1335,
+      "step": 1209
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7432366549041203e-05,
+      "loss": 0.2007,
+      "step": 1210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7427773678027556e-05,
+      "loss": 0.2066,
+      "step": 1211
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742317730911854e-05,
+      "loss": 0.2656,
+      "step": 1212
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.741857744447869e-05,
+      "loss": 0.1828,
+      "step": 1213
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741397408627417e-05,
+      "loss": 0.1668,
+      "step": 1214
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7409367236672806e-05,
+      "loss": 0.1909,
+      "step": 1215
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7404756897844054e-05,
+      "loss": 0.1846,
+      "step": 1216
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7400143071959027e-05,
+      "loss": 0.1694,
+      "step": 1217
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7395525761190466e-05,
+      "loss": 0.2094,
+      "step": 1218
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7390904967712764e-05,
+      "loss": 0.1806,
+      "step": 1219
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.738628069370195e-05,
+      "loss": 0.2186,
+      "step": 1220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7381652941335695e-05,
+      "loss": 0.1699,
+      "step": 1221
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.73770217127933e-05,
+      "loss": 0.2386,
+      "step": 1222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7372387010255707e-05,
+      "loss": 0.2214,
+      "step": 1223
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7367748835905504e-05,
+      "loss": 0.3277,
+      "step": 1224
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7363107191926897e-05,
+      "loss": 0.2157,
+      "step": 1225
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358462080505734e-05,
+      "loss": 0.1509,
+      "step": 1226
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.73538135038295e-05,
+      "loss": 0.1869,
+      "step": 1227
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7349161464087312e-05,
+      "loss": 0.1518,
+      "step": 1228
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7344505963469904e-05,
+      "loss": 0.2395,
+      "step": 1229
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339847004169653e-05,
+      "loss": 0.3239,
+      "step": 1230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335184588380563e-05,
+      "loss": 0.1955,
+      "step": 1231
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.1505,
+      "step": 1232
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7325849396120013e-05,
+      "loss": 0.2083,
+      "step": 1233
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732117662404469e-05,
+      "loss": 0.1654,
+      "step": 1234
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.73165004042728e-05,
+      "loss": 0.1448,
+      "step": 1235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731182073900648e-05,
+      "loss": 0.2076,
+      "step": 1236
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730713763044948e-05,
+      "loss": 0.213,
+      "step": 1237
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730245108080717e-05,
+      "loss": 0.1816,
+      "step": 1238
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7297761092286553e-05,
+      "loss": 0.188,
+      "step": 1239
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729306766709624e-05,
+      "loss": 0.2186,
+      "step": 1240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7288370807446463e-05,
+      "loss": 0.1349,
+      "step": 1241
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7283670515549073e-05,
+      "loss": 0.155,
+      "step": 1242
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727896679361754e-05,
+      "loss": 0.136,
+      "step": 1243
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727425964386694e-05,
+      "loss": 0.1484,
+      "step": 1244
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7269549068513973e-05,
+      "loss": 0.183,
+      "step": 1245
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7264835069776945e-05,
+      "loss": 0.1968,
+      "step": 1246
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7260117649875784e-05,
+      "loss": 0.1788,
+      "step": 1247
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7255396811032014e-05,
+      "loss": 0.1829,
+      "step": 1248
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7250672555468785e-05,
+      "loss": 0.1335,
+      "step": 1249
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7245944885410845e-05,
+      "loss": 0.2234,
+      "step": 1250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7241213803084557e-05,
+      "loss": 0.1784,
+      "step": 1251
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7236479310717878e-05,
+      "loss": 0.1774,
+      "step": 1252
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7231741410540395e-05,
+      "loss": 0.1581,
+      "step": 1253
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7227000104783274e-05,
+      "loss": 0.1017,
+      "step": 1254
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7222255395679298e-05,
+      "loss": 0.1933,
+      "step": 1255
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7217507285462854e-05,
+      "loss": 0.1719,
+      "step": 1256
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721275577636992e-05,
+      "loss": 0.1831,
+      "step": 1257
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7208000870638094e-05,
+      "loss": 0.182,
+      "step": 1258
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720324257050655e-05,
+      "loss": 0.2471,
+      "step": 1259
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719848087821608e-05,
+      "loss": 0.1915,
+      "step": 1260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7193715796009057e-05,
+      "loss": 0.1753,
+      "step": 1261
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718894732612947e-05,
+      "loss": 0.1314,
+      "step": 1262
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184175470822883e-05,
+      "loss": 0.2688,
+      "step": 1263
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7179400232336462e-05,
+      "loss": 0.2,
+      "step": 1264
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7174621612918977e-05,
+      "loss": 0.1616,
+      "step": 1265
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7169839614820775e-05,
+      "loss": 0.1922,
+      "step": 1266
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7165054240293798e-05,
+      "loss": 0.2015,
+      "step": 1267
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.716026549159158e-05,
+      "loss": 0.1614,
+      "step": 1268
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715547337096925e-05,
+      "loss": 0.1577,
+      "step": 1269
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7150677880683515e-05,
+      "loss": 0.1137,
+      "step": 1270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714587902299267e-05,
+      "loss": 0.1669,
+      "step": 1271
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.71410768001566e-05,
+      "loss": 0.2131,
+      "step": 1272
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7136271214436776e-05,
+      "loss": 0.1754,
+      "step": 1273
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7131462268096248e-05,
+      "loss": 0.1687,
+      "step": 1274
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7126649963399645e-05,
+      "loss": 0.1878,
+      "step": 1275
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712183430261319e-05,
+      "loss": 0.1697,
+      "step": 1276
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7117015288004677e-05,
+      "loss": 0.2599,
+      "step": 1277
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7112192921843477e-05,
+      "loss": 0.1904,
+      "step": 1278
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7107367206400555e-05,
+      "loss": 0.1678,
+      "step": 1279
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710253814394843e-05,
+      "loss": 0.2209,
+      "step": 1280
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7097705736761222e-05,
+      "loss": 0.2443,
+      "step": 1281
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70928699871146e-05,
+      "loss": 0.1838,
+      "step": 1282
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708803089728583e-05,
+      "loss": 0.1283,
+      "step": 1283
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7083188469553737e-05,
+      "loss": 0.1602,
+      "step": 1284
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7078342706198724e-05,
+      "loss": 0.1789,
+      "step": 1285
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7073493609502767e-05,
+      "loss": 0.1682,
+      "step": 1286
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70686411817494e-05,
+      "loss": 0.1705,
+      "step": 1287
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706378542522374e-05,
+      "loss": 0.1919,
+      "step": 1288
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7058926342212462e-05,
+      "loss": 0.1356,
+      "step": 1289
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7054063935003813e-05,
+      "loss": 0.1856,
+      "step": 1290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70491982058876e-05,
+      "loss": 0.1791,
+      "step": 1291
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70443291571552e-05,
+      "loss": 0.1631,
+      "step": 1292
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703945679109955e-05,
+      "loss": 0.1082,
+      "step": 1293
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7034581110015156e-05,
+      "loss": 0.2739,
+      "step": 1294
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7029702116198066e-05,
+      "loss": 0.2349,
+      "step": 1295
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.702481981194591e-05,
+      "loss": 0.1823,
+      "step": 1296
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7019934199557868e-05,
+      "loss": 0.1497,
+      "step": 1297
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7015045281334677e-05,
+      "loss": 0.1951,
+      "step": 1298
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.701015305957863e-05,
+      "loss": 0.1808,
+      "step": 1299
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7005257536593577e-05,
+      "loss": 0.1584,
+      "step": 1300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.700035871468493e-05,
+      "loss": 0.1879,
+      "step": 1301
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6995456596159637e-05,
+      "loss": 0.2168,
+      "step": 1302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6990551183326216e-05,
+      "loss": 0.133,
+      "step": 1303
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698564247849473e-05,
+      "loss": 0.1745,
+      "step": 1304
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6980730483976788e-05,
+      "loss": 0.1447,
+      "step": 1305
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6975815202085556e-05,
+      "loss": 0.1663,
+      "step": 1306
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6970896635135736e-05,
+      "loss": 0.1179,
+      "step": 1307
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6965974785443595e-05,
+      "loss": 0.1953,
+      "step": 1308
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.696104965532693e-05,
+      "loss": 0.0901,
+      "step": 1309
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6956121247105087e-05,
+      "loss": 0.1642,
+      "step": 1310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6951189563098966e-05,
+      "loss": 0.1768,
+      "step": 1311
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6946254605630995e-05,
+      "loss": 0.1676,
+      "step": 1312
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6941316377025146e-05,
+      "loss": 0.1531,
+      "step": 1313
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.693637487960694e-05,
+      "loss": 0.1799,
+      "step": 1314
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6931430115703434e-05,
+      "loss": 0.1688,
+      "step": 1315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6926482087643223e-05,
+      "loss": 0.1307,
+      "step": 1316
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6921530797756425e-05,
+      "loss": 0.1275,
+      "step": 1317
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.691657624837472e-05,
+      "loss": 0.1953,
+      "step": 1318
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6911618441831306e-05,
+      "loss": 0.2228,
+      "step": 1319
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690665738046092e-05,
+      "loss": 0.2167,
+      "step": 1320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6901693066599824e-05,
+      "loss": 0.2059,
+      "step": 1321
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689672550258583e-05,
+      "loss": 0.1808,
+      "step": 1322
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.689175469075826e-05,
+      "loss": 0.1979,
+      "step": 1323
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6886780633457975e-05,
+      "loss": 0.1913,
+      "step": 1324
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688180333302736e-05,
+      "loss": 0.1763,
+      "step": 1325
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687682279181034e-05,
+      "loss": 0.161,
+      "step": 1326
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6871839012152355e-05,
+      "loss": 0.2098,
+      "step": 1327
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6866851996400366e-05,
+      "loss": 0.1551,
+      "step": 1328
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6861861746902867e-05,
+      "loss": 0.1774,
+      "step": 1329
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6856868266009874e-05,
+      "loss": 0.1636,
+      "step": 1330
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.685187155607292e-05,
+      "loss": 0.1478,
+      "step": 1331
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684687161944506e-05,
+      "loss": 0.171,
+      "step": 1332
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6841868458480874e-05,
+      "loss": 0.1053,
+      "step": 1333
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6836862075536454e-05,
+      "loss": 0.1709,
+      "step": 1334
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6831852472969407e-05,
+      "loss": 0.1916,
+      "step": 1335
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6826839653138872e-05,
+      "loss": 0.1422,
+      "step": 1336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6821823618405475e-05,
+      "loss": 0.1534,
+      "step": 1337
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6816804371131388e-05,
+      "loss": 0.2082,
+      "step": 1338
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6811781913680273e-05,
+      "loss": 0.1547,
+      "step": 1339
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6806756248417307e-05,
+      "loss": 0.1882,
+      "step": 1340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6801727377709195e-05,
+      "loss": 0.1876,
+      "step": 1341
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679669530392413e-05,
+      "loss": 0.1895,
+      "step": 1342
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6791660029431818e-05,
+      "loss": 0.1979,
+      "step": 1343
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.678662155660348e-05,
+      "loss": 0.2217,
+      "step": 1344
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6781579887811842e-05,
+      "loss": 0.2152,
+      "step": 1345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.677653502543113e-05,
+      "loss": 0.1697,
+      "step": 1346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6771486971837077e-05,
+      "loss": 0.2025,
+      "step": 1347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6766435729406913e-05,
+      "loss": 0.1979,
+      "step": 1348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676138130051938e-05,
+      "loss": 0.2428,
+      "step": 1349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6756323687554713e-05,
+      "loss": 0.2021,
+      "step": 1350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6751262892894655e-05,
+      "loss": 0.2061,
+      "step": 1351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6746198918922426e-05,
+      "loss": 0.1943,
+      "step": 1352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6741131768022777e-05,
+      "loss": 0.0733,
+      "step": 1353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6736061442581922e-05,
+      "loss": 0.1689,
+      "step": 1354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673098794498759e-05,
+      "loss": 0.16,
+      "step": 1355
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6725911277628996e-05,
+      "loss": 0.1907,
+      "step": 1356
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672083144289685e-05,
+      "loss": 0.1852,
+      "step": 1357
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6715748443183353e-05,
+      "loss": 0.131,
+      "step": 1358
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6710662280882205e-05,
+      "loss": 0.1995,
+      "step": 1359
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6705572958388576e-05,
+      "loss": 0.216,
+      "step": 1360
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670048047809914e-05,
+      "loss": 0.1873,
+      "step": 1361
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695384842412053e-05,
+      "loss": 0.1583,
+      "step": 1362
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6690286053726958e-05,
+      "loss": 0.1758,
+      "step": 1363
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668518411444498e-05,
+      "loss": 0.2106,
+      "step": 1364
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6680079026968732e-05,
+      "loss": 0.2765,
+      "step": 1365
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667497079370231e-05,
+      "loss": 0.1627,
+      "step": 1366
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666985941705128e-05,
+      "loss": 0.1889,
+      "step": 1367
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6664744899422708e-05,
+      "loss": 0.1492,
+      "step": 1368
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6659627243225124e-05,
+      "loss": 0.1923,
+      "step": 1369
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6654506450868537e-05,
+      "loss": 0.1713,
+      "step": 1370
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6649382524764437e-05,
+      "loss": 0.2077,
+      "step": 1371
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6644255467325793e-05,
+      "loss": 0.2687,
+      "step": 1372
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6639125280967042e-05,
+      "loss": 0.1838,
+      "step": 1373
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6633991968104095e-05,
+      "loss": 0.1737,
+      "step": 1374
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6628855531154344e-05,
+      "loss": 0.1474,
+      "step": 1375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662371597253664e-05,
+      "loss": 0.2353,
+      "step": 1376
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6618573294671304e-05,
+      "loss": 0.1398,
+      "step": 1377
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6613427499980143e-05,
+      "loss": 0.1768,
+      "step": 1378
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6608278590886414e-05,
+      "loss": 0.1374,
+      "step": 1379
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6603126569814854e-05,
+      "loss": 0.1547,
+      "step": 1380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659797143919165e-05,
+      "loss": 0.1773,
+      "step": 1381
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6592813201444462e-05,
+      "loss": 0.2356,
+      "step": 1382
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6587651859002417e-05,
+      "loss": 0.1716,
+      "step": 1383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6582487414296097e-05,
+      "loss": 0.1681,
+      "step": 1384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6577319869757557e-05,
+      "loss": 0.1388,
+      "step": 1385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6572149227820286e-05,
+      "loss": 0.1674,
+      "step": 1386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656697549091926e-05,
+      "loss": 0.1673,
+      "step": 1387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6561798661490904e-05,
+      "loss": 0.2369,
+      "step": 1388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655661874197309e-05,
+      "loss": 0.1897,
+      "step": 1389
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.655143573480515e-05,
+      "loss": 0.1866,
+      "step": 1390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654624964242787e-05,
+      "loss": 0.1524,
+      "step": 1391
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6541060467283503e-05,
+      "loss": 0.1347,
+      "step": 1392
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6535868211815728e-05,
+      "loss": 0.1869,
+      "step": 1393
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653067287846969e-05,
+      "loss": 0.1921,
+      "step": 1394
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6525474469691985e-05,
+      "loss": 0.2526,
+      "step": 1395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6520272987930652e-05,
+      "loss": 0.2129,
+      "step": 1396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6515068435635177e-05,
+      "loss": 0.1647,
+      "step": 1397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6509860815256496e-05,
+      "loss": 0.1362,
+      "step": 1398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6504650129246983e-05,
+      "loss": 0.1899,
+      "step": 1399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6499436380060466e-05,
+      "loss": 0.1792,
+      "step": 1400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6494219570152203e-05,
+      "loss": 0.1693,
+      "step": 1401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6488999701978905e-05,
+      "loss": 0.239,
+      "step": 1402
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6483776777998716e-05,
+      "loss": 0.2378,
+      "step": 1403
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6478550800671224e-05,
+      "loss": 0.1992,
+      "step": 1404
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6473321772457444e-05,
+      "loss": 0.1594,
+      "step": 1405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6468089695819845e-05,
+      "loss": 0.1872,
+      "step": 1406
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6462854573222315e-05,
+      "loss": 0.1453,
+      "step": 1407
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645761640713019e-05,
+      "loss": 0.2088,
+      "step": 1408
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6452375200010232e-05,
+      "loss": 0.1722,
+      "step": 1409
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.644713095433063e-05,
+      "loss": 0.2405,
+      "step": 1410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6441883672561015e-05,
+      "loss": 0.1664,
+      "step": 1411
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6436633357172443e-05,
+      "loss": 0.1686,
+      "step": 1412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6431380010637394e-05,
+      "loss": 0.2153,
+      "step": 1413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6426123635429787e-05,
+      "loss": 0.172,
+      "step": 1414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6420864234024953e-05,
+      "loss": 0.2251,
+      "step": 1415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6415601808899658e-05,
+      "loss": 0.2115,
+      "step": 1416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641033636253209e-05,
+      "loss": 0.1537,
+      "step": 1417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640506789740186e-05,
+      "loss": 0.1984,
+      "step": 1418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6399796415989997e-05,
+      "loss": 0.2136,
+      "step": 1419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.639452192077895e-05,
+      "loss": 0.1554,
+      "step": 1420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6389244414252592e-05,
+      "loss": 0.1689,
+      "step": 1421
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6383963898896215e-05,
+      "loss": 0.1439,
+      "step": 1422
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6378680377196526e-05,
+      "loss": 0.2331,
+      "step": 1423
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6373393851641642e-05,
+      "loss": 0.1873,
+      "step": 1424
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63681043247211e-05,
+      "loss": 0.1439,
+      "step": 1425
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6362811798925852e-05,
+      "loss": 0.1894,
+      "step": 1426
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6357516276748264e-05,
+      "loss": 0.0934,
+      "step": 1427
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6352217760682098e-05,
+      "loss": 0.1983,
+      "step": 1428
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6346916253222547e-05,
+      "loss": 0.2015,
+      "step": 1429
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.63416117568662e-05,
+      "loss": 0.1721,
+      "step": 1430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6336304274111055e-05,
+      "loss": 0.1332,
+      "step": 1431
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633099380745652e-05,
+      "loss": 0.2697,
+      "step": 1432
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6325680359403405e-05,
+      "loss": 0.183,
+      "step": 1433
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6320363932453922e-05,
+      "loss": 0.1898,
+      "step": 1434
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.631504452911169e-05,
+      "loss": 0.1732,
+      "step": 1435
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630972215188173e-05,
+      "loss": 0.2253,
+      "step": 1436
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630439680327046e-05,
+      "loss": 0.1659,
+      "step": 1437
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.62990684857857e-05,
+      "loss": 0.1854,
+      "step": 1438
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6293737201936667e-05,
+      "loss": 0.2087,
+      "step": 1439
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6288402954233976e-05,
+      "loss": 0.2797,
+      "step": 1440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6283065745189633e-05,
+      "loss": 0.1503,
+      "step": 1441
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.627772557731705e-05,
+      "loss": 0.1586,
+      "step": 1442
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6272382453131017e-05,
+      "loss": 0.2064,
+      "step": 1443
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6267036375147728e-05,
+      "loss": 0.157,
+      "step": 1444
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626168734588476e-05,
+      "loss": 0.1544,
+      "step": 1445
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6256335367861088e-05,
+      "loss": 0.1638,
+      "step": 1446
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625098044359707e-05,
+      "loss": 0.2127,
+      "step": 1447
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6245622575614452e-05,
+      "loss": 0.1802,
+      "step": 1448
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624026176643637e-05,
+      "loss": 0.1439,
+      "step": 1449
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.2068,
+      "step": 1450
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6229531334593257e-05,
+      "loss": 0.239,
+      "step": 1451
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622416171698142e-05,
+      "loss": 0.1092,
+      "step": 1452
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6218789168280487e-05,
+      "loss": 0.1794,
+      "step": 1453
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6213413691020505e-05,
+      "loss": 0.2193,
+      "step": 1454
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6208035287732898e-05,
+      "loss": 0.1509,
+      "step": 1455
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6202653960950474e-05,
+      "loss": 0.16,
+      "step": 1456
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6197269713207413e-05,
+      "loss": 0.2759,
+      "step": 1457
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.619188254703927e-05,
+      "loss": 0.1794,
+      "step": 1458
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6186492464982968e-05,
+      "loss": 0.1797,
+      "step": 1459
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618109946957682e-05,
+      "loss": 0.1969,
+      "step": 1460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61757035633605e-05,
+      "loss": 0.2166,
+      "step": 1461
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617030474887505e-05,
+      "loss": 0.1347,
+      "step": 1462
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6164903028662887e-05,
+      "loss": 0.1729,
+      "step": 1463
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6159498405267794e-05,
+      "loss": 0.1882,
+      "step": 1464
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.615409088123493e-05,
+      "loss": 0.1843,
+      "step": 1465
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6148680459110803e-05,
+      "loss": 0.1442,
+      "step": 1466
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6143267141443296e-05,
+      "loss": 0.166,
+      "step": 1467
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.613785093078166e-05,
+      "loss": 0.1861,
+      "step": 1468
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6132431829676496e-05,
+      "loss": 0.1716,
+      "step": 1469
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612700984067978e-05,
+      "loss": 0.1887,
+      "step": 1470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6121584966344837e-05,
+      "loss": 0.1685,
+      "step": 1471
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6116157209226356e-05,
+      "loss": 0.1613,
+      "step": 1472
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6110726571880382e-05,
+      "loss": 0.2048,
+      "step": 1473
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6105293056864314e-05,
+      "loss": 0.1272,
+      "step": 1474
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6099856666736914e-05,
+      "loss": 0.1765,
+      "step": 1475
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609441740405829e-05,
+      "loss": 0.1854,
+      "step": 1476
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6088975271389905e-05,
+      "loss": 0.1412,
+      "step": 1477
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.608353027129457e-05,
+      "loss": 0.2498,
+      "step": 1478
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6078082406336465e-05,
+      "loss": 0.1577,
+      "step": 1479
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607263167908109e-05,
+      "loss": 0.1813,
+      "step": 1480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606717809209531e-05,
+      "loss": 0.1478,
+      "step": 1481
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6061721647947343e-05,
+      "loss": 0.1997,
+      "step": 1482
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6056262349206734e-05,
+      "loss": 0.1446,
+      "step": 1483
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6050800198444385e-05,
+      "loss": 0.1869,
+      "step": 1484
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.604533519823254e-05,
+      "loss": 0.2258,
+      "step": 1485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6039867351144778e-05,
+      "loss": 0.1967,
+      "step": 1486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6034396659756025e-05,
+      "loss": 0.2452,
+      "step": 1487
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6028923126642553e-05,
+      "loss": 0.1394,
+      "step": 1488
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6023446754381958e-05,
+      "loss": 0.1451,
+      "step": 1489
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6017967545553174e-05,
+      "loss": 0.2055,
+      "step": 1490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6012485502736486e-05,
+      "loss": 0.2184,
+      "step": 1491
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6007000628513498e-05,
+      "loss": 0.1789,
+      "step": 1492
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6001512925467153e-05,
+      "loss": 0.2231,
+      "step": 1493
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599602239618173e-05,
+      "loss": 0.2099,
+      "step": 1494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.599052904324283e-05,
+      "loss": 0.2097,
+      "step": 1495
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5985032869237392e-05,
+      "loss": 0.1384,
+      "step": 1496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5979533876753675e-05,
+      "loss": 0.1166,
+      "step": 1497
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597403206838128e-05,
+      "loss": 0.1723,
+      "step": 1498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596852744671112e-05,
+      "loss": 0.1241,
+      "step": 1499
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5963020014335437e-05,
+      "loss": 0.1687,
+      "step": 1500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.59575097738478e-05,
+      "loss": 0.2445,
+      "step": 1501
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5951996727843087e-05,
+      "loss": 0.1991,
+      "step": 1502
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.594648087891752e-05,
+      "loss": 0.2181,
+      "step": 1503
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5940962229668625e-05,
+      "loss": 0.168,
+      "step": 1504
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5935440782695248e-05,
+      "loss": 0.1616,
+      "step": 1505
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5929916540597557e-05,
+      "loss": 0.1277,
+      "step": 1506
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5924389505977038e-05,
+      "loss": 0.222,
+      "step": 1507
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591885968143648e-05,
+      "loss": 0.1929,
+      "step": 1508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591332706958e-05,
+      "loss": 0.2553,
+      "step": 1509
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5907791673013016e-05,
+      "loss": 0.1449,
+      "step": 1510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.590225349434227e-05,
+      "loss": 0.2003,
+      "step": 1511
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.58967125361758e-05,
+      "loss": 0.1836,
+      "step": 1512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891168801122966e-05,
+      "loss": 0.106,
+      "step": 1513
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588562229179443e-05,
+      "loss": 0.258,
+      "step": 1514
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5880073010802157e-05,
+      "loss": 0.1662,
+      "step": 1515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5874520960759423e-05,
+      "loss": 0.1427,
+      "step": 1516
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5868966144280804e-05,
+      "loss": 0.1892,
+      "step": 1517
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586340856398218e-05,
+      "loss": 0.1798,
+      "step": 1518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5857848222480733e-05,
+      "loss": 0.1901,
+      "step": 1519
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5852285122394943e-05,
+      "loss": 0.1924,
+      "step": 1520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5846719266344598e-05,
+      "loss": 0.2019,
+      "step": 1521
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584115065695077e-05,
+      "loss": 0.1339,
+      "step": 1522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583557929683584e-05,
+      "loss": 0.1803,
+      "step": 1523
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5830005188623476e-05,
+      "loss": 0.1366,
+      "step": 1524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582442833493864e-05,
+      "loss": 0.1899,
+      "step": 1525
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.58188487384076e-05,
+      "loss": 0.2371,
+      "step": 1526
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5813266401657898e-05,
+      "loss": 0.1326,
+      "step": 1527
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5807681327318372e-05,
+      "loss": 0.1981,
+      "step": 1528
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5802093518019157e-05,
+      "loss": 0.2208,
+      "step": 1529
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.579650297639167e-05,
+      "loss": 0.1475,
+      "step": 1530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5790909705068612e-05,
+      "loss": 0.1738,
+      "step": 1531
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578531370668397e-05,
+      "loss": 0.2146,
+      "step": 1532
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5779714983873023e-05,
+      "loss": 0.184,
+      "step": 1533
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5774113539272332e-05,
+      "loss": 0.1678,
+      "step": 1534
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5768509375519726e-05,
+      "loss": 0.2181,
+      "step": 1535
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576290249525433e-05,
+      "loss": 0.1196,
+      "step": 1536
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575729290111654e-05,
+      "loss": 0.2085,
+      "step": 1537
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575168059574803e-05,
+      "loss": 0.1293,
+      "step": 1538
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746065581791755e-05,
+      "loss": 0.2244,
+      "step": 1539
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5740447861891946e-05,
+      "loss": 0.1558,
+      "step": 1540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57348274386941e-05,
+      "loss": 0.2048,
+      "step": 1541
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5729204314845002e-05,
+      "loss": 0.145,
+      "step": 1542
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5723578492992687e-05,
+      "loss": 0.2052,
+      "step": 1543
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.571794997578649e-05,
+      "loss": 0.1937,
+      "step": 1544
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.571231876587698e-05,
+      "loss": 0.1766,
+      "step": 1545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5706684865916025e-05,
+      "loss": 0.2081,
+      "step": 1546
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5701048278556745e-05,
+      "loss": 0.2449,
+      "step": 1547
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5695409006453527e-05,
+      "loss": 0.156,
+      "step": 1548
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5689767052262028e-05,
+      "loss": 0.1596,
+      "step": 1549
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.568412241863915e-05,
+      "loss": 0.242,
+      "step": 1550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.567847510824308e-05,
+      "loss": 0.1989,
+      "step": 1551
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5672825123733257e-05,
+      "loss": 0.2583,
+      "step": 1552
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5667172467770372e-05,
+      "loss": 0.1126,
+      "step": 1553
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5661517143016386e-05,
+      "loss": 0.1939,
+      "step": 1554
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565585915213451e-05,
+      "loss": 0.2073,
+      "step": 1555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.56501984977892e-05,
+      "loss": 0.2157,
+      "step": 1556
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5644535182646197e-05,
+      "loss": 0.1645,
+      "step": 1557
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563886920937246e-05,
+      "loss": 0.1209,
+      "step": 1558
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563320058063622e-05,
+      "loss": 0.2077,
+      "step": 1559
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.562752929910696e-05,
+      "loss": 0.2079,
+      "step": 1560
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5621855367455395e-05,
+      "loss": 0.146,
+      "step": 1561
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5616178788353508e-05,
+      "loss": 0.1786,
+      "step": 1562
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5610499564474512e-05,
+      "loss": 0.1458,
+      "step": 1563
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5604817698492886e-05,
+      "loss": 0.2422,
+      "step": 1564
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559913319308433e-05,
+      "loss": 0.2082,
+      "step": 1565
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5593446050925797e-05,
+      "loss": 0.1943,
+      "step": 1566
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5587756274695488e-05,
+      "loss": 0.1934,
+      "step": 1567
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5582063867072835e-05,
+      "loss": 0.1603,
+      "step": 1568
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5576368830738515e-05,
+      "loss": 0.2684,
+      "step": 1569
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.557067116837444e-05,
+      "loss": 0.2037,
+      "step": 1570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.556497088266376e-05,
+      "loss": 0.1588,
+      "step": 1571
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5559267976290854e-05,
+      "loss": 0.1496,
+      "step": 1572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5553562451941353e-05,
+      "loss": 0.2466,
+      "step": 1573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5547854312302096e-05,
+      "loss": 0.1755,
+      "step": 1574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554214356006118e-05,
+      "loss": 0.1764,
+      "step": 1575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5536430197907904e-05,
+      "loss": 0.1251,
+      "step": 1576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5530714228532818e-05,
+      "loss": 0.1607,
+      "step": 1577
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5524995654627696e-05,
+      "loss": 0.1488,
+      "step": 1578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5519274478885535e-05,
+      "loss": 0.2439,
+      "step": 1579
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5513550704000548e-05,
+      "loss": 0.1484,
+      "step": 1580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5507824332668185e-05,
+      "loss": 0.1842,
+      "step": 1581
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5502095367585124e-05,
+      "loss": 0.2183,
+      "step": 1582
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5496363811449247e-05,
+      "loss": 0.1055,
+      "step": 1583
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5490629666959668e-05,
+      "loss": 0.1639,
+      "step": 1584
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548489293681671e-05,
+      "loss": 0.171,
+      "step": 1585
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.547915362372193e-05,
+      "loss": 0.1751,
+      "step": 1586
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5473411730378086e-05,
+      "loss": 0.1786,
+      "step": 1587
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5467667259489157e-05,
+      "loss": 0.2585,
+      "step": 1588
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.546192021376034e-05,
+      "loss": 0.1861,
+      "step": 1589
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5456170595898033e-05,
+      "loss": 0.1494,
+      "step": 1590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545041840860986e-05,
+      "loss": 0.2125,
+      "step": 1591
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5444663654604643e-05,
+      "loss": 0.1782,
+      "step": 1592
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5438906336592413e-05,
+      "loss": 0.131,
+      "step": 1593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543314645728442e-05,
+      "loss": 0.1497,
+      "step": 1594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542738401939311e-05,
+      "loss": 0.1329,
+      "step": 1595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.542161902563214e-05,
+      "loss": 0.1696,
+      "step": 1596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5415851478716356e-05,
+      "loss": 0.1871,
+      "step": 1597
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541008138136183e-05,
+      "loss": 0.205,
+      "step": 1598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5404308736285818e-05,
+      "loss": 0.1314,
+      "step": 1599
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539853354620678e-05,
+      "loss": 0.1764,
+      "step": 1600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.539275581384437e-05,
+      "loss": 0.2071,
+      "step": 1601
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5386975541919457e-05,
+      "loss": 0.219,
+      "step": 1602
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5381192733154083e-05,
+      "loss": 0.1517,
+      "step": 1603
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5375407390271492e-05,
+      "loss": 0.1413,
+      "step": 1604
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5369619515996132e-05,
+      "loss": 0.2314,
+      "step": 1605
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5363829113053633e-05,
+      "loss": 0.1697,
+      "step": 1606
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5358036184170817e-05,
+      "loss": 0.1962,
+      "step": 1607
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352240732075694e-05,
+      "loss": 0.0958,
+      "step": 1608
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534644275949747e-05,
+      "loss": 0.1661,
+      "step": 1609
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534064226916653e-05,
+      "loss": 0.1643,
+      "step": 1610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5334839263814447e-05,
+      "loss": 0.1769,
+      "step": 1611
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5329033746173975e-05,
+      "loss": 0.139,
+      "step": 1612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5323225718979063e-05,
+      "loss": 0.1761,
+      "step": 1613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317415184964824e-05,
+      "loss": 0.1851,
+      "step": 1614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531160214686757e-05,
+      "loss": 0.1944,
+      "step": 1615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.530578660742478e-05,
+      "loss": 0.1802,
+      "step": 1616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529996856937511e-05,
+      "loss": 0.2119,
+      "step": 1617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5294148035458406e-05,
+      "loss": 0.1913,
+      "step": 1618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5288325008415668e-05,
+      "loss": 0.1626,
+      "step": 1619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5282499490989094e-05,
+      "loss": 0.206,
+      "step": 1620
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5276671485922036e-05,
+      "loss": 0.2167,
+      "step": 1621
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5270840995959028e-05,
+      "loss": 0.1746,
+      "step": 1622
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526500802384577e-05,
+      "loss": 0.2625,
+      "step": 1623
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5259172572329132e-05,
+      "loss": 0.2009,
+      "step": 1624
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5253334644157147e-05,
+      "loss": 0.2644,
+      "step": 1625
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5247494242079024e-05,
+      "loss": 0.1315,
+      "step": 1626
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241651368845128e-05,
+      "loss": 0.1549,
+      "step": 1627
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5235806027206997e-05,
+      "loss": 0.1817,
+      "step": 1628
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5229958219917317e-05,
+      "loss": 0.1973,
+      "step": 1629
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5224107949729952e-05,
+      "loss": 0.1268,
+      "step": 1630
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.521825521939991e-05,
+      "loss": 0.2113,
+      "step": 1631
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5212400031683374e-05,
+      "loss": 0.1561,
+      "step": 1632
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520654238933767e-05,
+      "loss": 0.1513,
+      "step": 1633
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5200682295121286e-05,
+      "loss": 0.1868,
+      "step": 1634
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5194819751793861e-05,
+      "loss": 0.2038,
+      "step": 1635
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5188954762116197e-05,
+      "loss": 0.1954,
+      "step": 1636
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5183087328850237e-05,
+      "loss": 0.2361,
+      "step": 1637
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5177217454759081e-05,
+      "loss": 0.1918,
+      "step": 1638
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5171345142606973e-05,
+      "loss": 0.2057,
+      "step": 1639
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5165470395159314e-05,
+      "loss": 0.1468,
+      "step": 1640
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5159593215182646e-05,
+      "loss": 0.1152,
+      "step": 1641
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.515371360544465e-05,
+      "loss": 0.1938,
+      "step": 1642
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5147831568714164e-05,
+      "loss": 0.1561,
+      "step": 1643
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141947107761166e-05,
+      "loss": 0.2392,
+      "step": 1644
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5136060225356768e-05,
+      "loss": 0.1595,
+      "step": 1645
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5130170924273229e-05,
+      "loss": 0.1882,
+      "step": 1646
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5124279207283942e-05,
+      "loss": 0.1688,
+      "step": 1647
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5118385077163446e-05,
+      "loss": 0.1832,
+      "step": 1648
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5112488536687413e-05,
+      "loss": 0.1962,
+      "step": 1649
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5106589588632636e-05,
+      "loss": 0.155,
+      "step": 1650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5100688235777065e-05,
+      "loss": 0.1176,
+      "step": 1651
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5094784480899774e-05,
+      "loss": 0.2214,
+      "step": 1652
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5088878326780956e-05,
+      "loss": 0.1726,
+      "step": 1653
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5082969776201948e-05,
+      "loss": 0.1776,
+      "step": 1654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5077058831945212e-05,
+      "loss": 0.1592,
+      "step": 1655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5071145496794338e-05,
+      "loss": 0.1847,
+      "step": 1656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5065229773534035e-05,
+      "loss": 0.1763,
+      "step": 1657
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5059311664950146e-05,
+      "loss": 0.2262,
+      "step": 1658
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5053391173829632e-05,
+      "loss": 0.2489,
+      "step": 1659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5047468302960577e-05,
+      "loss": 0.1476,
+      "step": 1660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5041543055132186e-05,
+      "loss": 0.1086,
+      "step": 1661
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5035615433134787e-05,
+      "loss": 0.1635,
+      "step": 1662
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5029685439759813e-05,
+      "loss": 0.2045,
+      "step": 1663
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5023753077799834e-05,
+      "loss": 0.1745,
+      "step": 1664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5017818350048518e-05,
+      "loss": 0.1518,
+      "step": 1665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5011881259300654e-05,
+      "loss": 0.1769,
+      "step": 1666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5005941808352144e-05,
+      "loss": 0.2318,
+      "step": 1667
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.1662,
+      "step": 1668
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4994055837042347e-05,
+      "loss": 0.1482,
+      "step": 1669
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4988109322278417e-05,
+      "loss": 0.2079,
+      "step": 1670
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4982160458508546e-05,
+      "loss": 0.2055,
+      "step": 1671
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4976209248534183e-05,
+      "loss": 0.0938,
+      "step": 1672
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4970255695157876e-05,
+      "loss": 0.1448,
+      "step": 1673
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4964299801183278e-05,
+      "loss": 0.1296,
+      "step": 1674
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4958341569415149e-05,
+      "loss": 0.2053,
+      "step": 1675
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4952381002659344e-05,
+      "loss": 0.238,
+      "step": 1676
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4946418103722819e-05,
+      "loss": 0.1712,
+      "step": 1677
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4940452875413627e-05,
+      "loss": 0.1882,
+      "step": 1678
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4934485320540926e-05,
+      "loss": 0.2165,
+      "step": 1679
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.492851544191496e-05,
+      "loss": 0.1598,
+      "step": 1680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4922543242347067e-05,
+      "loss": 0.1681,
+      "step": 1681
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4916568724649688e-05,
+      "loss": 0.2373,
+      "step": 1682
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910591891636345e-05,
+      "loss": 0.1374,
+      "step": 1683
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4904612746121657e-05,
+      "loss": 0.1288,
+      "step": 1684
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4898631290921333e-05,
+      "loss": 0.1643,
+      "step": 1685
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.489264752885216e-05,
+      "loss": 0.1785,
+      "step": 1686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4886661462732021e-05,
+      "loss": 0.1587,
+      "step": 1687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4880673095379883e-05,
+      "loss": 0.1781,
+      "step": 1688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.487468242961579e-05,
+      "loss": 0.2579,
+      "step": 1689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4868689468260876e-05,
+      "loss": 0.2268,
+      "step": 1690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4862694214137352e-05,
+      "loss": 0.1348,
+      "step": 1691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4856696670068508e-05,
+      "loss": 0.2337,
+      "step": 1692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485069683887872e-05,
+      "loss": 0.2228,
+      "step": 1693
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4844694723393429e-05,
+      "loss": 0.1435,
+      "step": 1694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4838690326439161e-05,
+      "loss": 0.1242,
+      "step": 1695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.483268365084351e-05,
+      "loss": 0.1952,
+      "step": 1696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4826674699435146e-05,
+      "loss": 0.2042,
+      "step": 1697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4820663475043818e-05,
+      "loss": 0.1994,
+      "step": 1698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4814649980500333e-05,
+      "loss": 0.1427,
+      "step": 1699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.480863421863657e-05,
+      "loss": 0.2063,
+      "step": 1700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4802616192285486e-05,
+      "loss": 0.1645,
+      "step": 1701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479659590428109e-05,
+      "loss": 0.2398,
+      "step": 1702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4790573357458464e-05,
+      "loss": 0.1715,
+      "step": 1703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4784548554653756e-05,
+      "loss": 0.1652,
+      "step": 1704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4778521498704166e-05,
+      "loss": 0.1258,
+      "step": 1705
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4772492192447969e-05,
+      "loss": 0.1572,
+      "step": 1706
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766460638724486e-05,
+      "loss": 0.2146,
+      "step": 1707
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.476042684037411e-05,
+      "loss": 0.157,
+      "step": 1708
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4754390800238278e-05,
+      "loss": 0.1959,
+      "step": 1709
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4748352521159492e-05,
+      "loss": 0.1735,
+      "step": 1710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4742312005981303e-05,
+      "loss": 0.1877,
+      "step": 1711
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4736269257548316e-05,
+      "loss": 0.2456,
+      "step": 1712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4730224278706194e-05,
+      "loss": 0.1952,
+      "step": 1713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4724177072301642e-05,
+      "loss": 0.1267,
+      "step": 1714
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4718127641182414e-05,
+      "loss": 0.118,
+      "step": 1715
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4712075988197318e-05,
+      "loss": 0.1503,
+      "step": 1716
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4706022116196208e-05,
+      "loss": 0.1853,
+      "step": 1717
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4699966028029976e-05,
+      "loss": 0.1669,
+      "step": 1718
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.469390772655056e-05,
+      "loss": 0.1776,
+      "step": 1719
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4687847214610944e-05,
+      "loss": 0.2313,
+      "step": 1720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4681784495065154e-05,
+      "loss": 0.2098,
+      "step": 1721
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4675719570768252e-05,
+      "loss": 0.163,
+      "step": 1722
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4669652444576332e-05,
+      "loss": 0.1685,
+      "step": 1723
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.466358311934654e-05,
+      "loss": 0.202,
+      "step": 1724
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4657511597937046e-05,
+      "loss": 0.2338,
+      "step": 1725
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4651437883207056e-05,
+      "loss": 0.2134,
+      "step": 1726
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4645361978016811e-05,
+      "loss": 0.2031,
+      "step": 1727
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4639283885227586e-05,
+      "loss": 0.2195,
+      "step": 1728
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4633203607701683e-05,
+      "loss": 0.1818,
+      "step": 1729
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4627121148302428e-05,
+      "loss": 0.106,
+      "step": 1730
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621036509894184e-05,
+      "loss": 0.1962,
+      "step": 1731
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4614949695342335e-05,
+      "loss": 0.231,
+      "step": 1732
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4608860707513287e-05,
+      "loss": 0.2397,
+      "step": 1733
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4602769549274477e-05,
+      "loss": 0.1639,
+      "step": 1734
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4596676223494359e-05,
+      "loss": 0.2328,
+      "step": 1735
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4590580733042408e-05,
+      "loss": 0.293,
+      "step": 1736
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4584483080789117e-05,
+      "loss": 0.1383,
+      "step": 1737
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4578383269606004e-05,
+      "loss": 0.1651,
+      "step": 1738
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4572281302365596e-05,
+      "loss": 0.1918,
+      "step": 1739
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4566177181941439e-05,
+      "loss": 0.2282,
+      "step": 1740
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4560070911208088e-05,
+      "loss": 0.18,
+      "step": 1741
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.455396249304112e-05,
+      "loss": 0.1434,
+      "step": 1742
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547851930317119e-05,
+      "loss": 0.1649,
+      "step": 1743
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4541739225913669e-05,
+      "loss": 0.1347,
+      "step": 1744
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4535624382709382e-05,
+      "loss": 0.1644,
+      "step": 1745
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4529507403583862e-05,
+      "loss": 0.1827,
+      "step": 1746
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.452338829141772e-05,
+      "loss": 0.1695,
+      "step": 1747
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4517267049092585e-05,
+      "loss": 0.1703,
+      "step": 1748
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4511143679491069e-05,
+      "loss": 0.2357,
+      "step": 1749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4505018185496802e-05,
+      "loss": 0.1863,
+      "step": 1750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4498890569994407e-05,
+      "loss": 0.1573,
+      "step": 1751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4492760835869504e-05,
+      "loss": 0.1753,
+      "step": 1752
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4486628986008717e-05,
+      "loss": 0.1878,
+      "step": 1753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4480495023299665e-05,
+      "loss": 0.1671,
+      "step": 1754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4474358950630958e-05,
+      "loss": 0.1619,
+      "step": 1755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4468220770892208e-05,
+      "loss": 0.1323,
+      "step": 1756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4462080486974002e-05,
+      "loss": 0.1493,
+      "step": 1757
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445593810176794e-05,
+      "loss": 0.1357,
+      "step": 1758
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4449793618166594e-05,
+      "loss": 0.1512,
+      "step": 1759
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4443647039063534e-05,
+      "loss": 0.1961,
+      "step": 1760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4437498367353313e-05,
+      "loss": 0.1891,
+      "step": 1761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.443134760593147e-05,
+      "loss": 0.1277,
+      "step": 1762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442519475769453e-05,
+      "loss": 0.1035,
+      "step": 1763
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.441903982554e-05,
+      "loss": 0.172,
+      "step": 1764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.441288281236636e-05,
+      "loss": 0.1918,
+      "step": 1765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4406723721073088e-05,
+      "loss": 0.176,
+      "step": 1766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4400562554560624e-05,
+      "loss": 0.1895,
+      "step": 1767
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4394399315730389e-05,
+      "loss": 0.1597,
+      "step": 1768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438823400748479e-05,
+      "loss": 0.2043,
+      "step": 1769
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4382066632727192e-05,
+      "loss": 0.1555,
+      "step": 1770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.437589719436195e-05,
+      "loss": 0.2583,
+      "step": 1771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4369725695294377e-05,
+      "loss": 0.1824,
+      "step": 1772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4363552138430764e-05,
+      "loss": 0.1981,
+      "step": 1773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4357376526678368e-05,
+      "loss": 0.1987,
+      "step": 1774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4351198862945419e-05,
+      "loss": 0.1351,
+      "step": 1775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345019150141104e-05,
+      "loss": 0.2462,
+      "step": 1776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.2571,
+      "step": 1777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4332653588959974e-05,
+      "loss": 0.1706,
+      "step": 1778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4326467746406367e-05,
+      "loss": 0.145,
+      "step": 1779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4320279866427798e-05,
+      "loss": 0.1902,
+      "step": 1780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4314089951938273e-05,
+      "loss": 0.2346,
+      "step": 1781
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307898005852756e-05,
+      "loss": 0.2079,
+      "step": 1782
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.430170403108716e-05,
+      "loss": 0.184,
+      "step": 1783
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4295508030558361e-05,
+      "loss": 0.1399,
+      "step": 1784
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4289310007184187e-05,
+      "loss": 0.2113,
+      "step": 1785
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.428310996388342e-05,
+      "loss": 0.1775,
+      "step": 1786
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4276907903575788e-05,
+      "loss": 0.1466,
+      "step": 1787
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4270703829181969e-05,
+      "loss": 0.1232,
+      "step": 1788
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4264497743623598e-05,
+      "loss": 0.2081,
+      "step": 1789
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4258289649823252e-05,
+      "loss": 0.2204,
+      "step": 1790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4252079550704447e-05,
+      "loss": 0.1664,
+      "step": 1791
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424586744919166e-05,
+      "loss": 0.1727,
+      "step": 1792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423965334821029e-05,
+      "loss": 0.2227,
+      "step": 1793
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4233437250686695e-05,
+      "loss": 0.1778,
+      "step": 1794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4227219159548163e-05,
+      "loss": 0.1659,
+      "step": 1795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220999077722927e-05,
+      "loss": 0.2609,
+      "step": 1796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4214777008140155e-05,
+      "loss": 0.185,
+      "step": 1797
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4208552953729949e-05,
+      "loss": 0.1321,
+      "step": 1798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4202326917423344e-05,
+      "loss": 0.2735,
+      "step": 1799
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4196098902152323e-05,
+      "loss": 0.1367,
+      "step": 1800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4189868910849779e-05,
+      "loss": 0.1913,
+      "step": 1801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4183636946449547e-05,
+      "loss": 0.143,
+      "step": 1802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4177403011886397e-05,
+      "loss": 0.1549,
+      "step": 1803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4171167110096017e-05,
+      "loss": 0.117,
+      "step": 1804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4164929244015027e-05,
+      "loss": 0.1326,
+      "step": 1805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4158689416580965e-05,
+      "loss": 0.1759,
+      "step": 1806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4152447630732295e-05,
+      "loss": 0.1363,
+      "step": 1807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4146203889408418e-05,
+      "loss": 0.1658,
+      "step": 1808
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4139958195549635e-05,
+      "loss": 0.1234,
+      "step": 1809
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4133710552097175e-05,
+      "loss": 0.1489,
+      "step": 1810
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4127460961993184e-05,
+      "loss": 0.1893,
+      "step": 1811
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4121209428180731e-05,
+      "loss": 0.1913,
+      "step": 1812
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4114955953603793e-05,
+      "loss": 0.2063,
+      "step": 1813
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4108700541207259e-05,
+      "loss": 0.2135,
+      "step": 1814
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4102443193936941e-05,
+      "loss": 0.1086,
+      "step": 1815
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4096183914739554e-05,
+      "loss": 0.2029,
+      "step": 1816
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4089922706562722e-05,
+      "loss": 0.1571,
+      "step": 1817
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4083659572354982e-05,
+      "loss": 0.1706,
+      "step": 1818
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4077394515065777e-05,
+      "loss": 0.0999,
+      "step": 1819
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4071127537645453e-05,
+      "loss": 0.1645,
+      "step": 1820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064858643045264e-05,
+      "loss": 0.1791,
+      "step": 1821
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4058587834217356e-05,
+      "loss": 0.201,
+      "step": 1822
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4052315114114801e-05,
+      "loss": 0.2003,
+      "step": 1823
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4046040485691543e-05,
+      "loss": 0.2223,
+      "step": 1824
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.403976395190244e-05,
+      "loss": 0.1687,
+      "step": 1825
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4033485515703245e-05,
+      "loss": 0.1511,
+      "step": 1826
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4027205180050608e-05,
+      "loss": 0.1285,
+      "step": 1827
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4020922947902067e-05,
+      "loss": 0.2213,
+      "step": 1828
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4014638822216063e-05,
+      "loss": 0.2051,
+      "step": 1829
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4008352805951917e-05,
+      "loss": 0.2116,
+      "step": 1830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4002064902069854e-05,
+      "loss": 0.1755,
+      "step": 1831
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3995775113530974e-05,
+      "loss": 0.1742,
+      "step": 1832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3989483443297274e-05,
+      "loss": 0.1507,
+      "step": 1833
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3983189894331636e-05,
+      "loss": 0.1966,
+      "step": 1834
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3976894469597824e-05,
+      "loss": 0.187,
+      "step": 1835
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970597172060482e-05,
+      "loss": 0.1618,
+      "step": 1836
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3964298004685148e-05,
+      "loss": 0.1582,
+      "step": 1837
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3957996970438226e-05,
+      "loss": 0.1767,
+      "step": 1838
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3951694072287012e-05,
+      "loss": 0.2006,
+      "step": 1839
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3945389313199669e-05,
+      "loss": 0.259,
+      "step": 1840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3939082696145243e-05,
+      "loss": 0.1664,
+      "step": 1841
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3932774224093653e-05,
+      "loss": 0.1414,
+      "step": 1842
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.392646390001569e-05,
+      "loss": 0.1576,
+      "step": 1843
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3920151726883021e-05,
+      "loss": 0.1569,
+      "step": 1844
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3913837707668179e-05,
+      "loss": 0.1709,
+      "step": 1845
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3907521845344571e-05,
+      "loss": 0.1379,
+      "step": 1846
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3901204142886473e-05,
+      "loss": 0.1283,
+      "step": 1847
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3894884603269016e-05,
+      "loss": 0.2103,
+      "step": 1848
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3888563229468213e-05,
+      "loss": 0.1247,
+      "step": 1849
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3882240024460928e-05,
+      "loss": 0.1551,
+      "step": 1850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3875914991224888e-05,
+      "loss": 0.1609,
+      "step": 1851
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3869588132738696e-05,
+      "loss": 0.2249,
+      "step": 1852
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.386325945198179e-05,
+      "loss": 0.1525,
+      "step": 1853
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3856928951934487e-05,
+      "loss": 0.1608,
+      "step": 1854
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3850596635577955e-05,
+      "loss": 0.1505,
+      "step": 1855
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3844262505894205e-05,
+      "loss": 0.263,
+      "step": 1856
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3837926565866127e-05,
+      "loss": 0.1844,
+      "step": 1857
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3831588818477437e-05,
+      "loss": 0.1902,
+      "step": 1858
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.382524926671272e-05,
+      "loss": 0.13,
+      "step": 1859
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.38189079135574e-05,
+      "loss": 0.2812,
+      "step": 1860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3812564761997758e-05,
+      "loss": 0.1198,
+      "step": 1861
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806219815020916e-05,
+      "loss": 0.1713,
+      "step": 1862
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3799873075614847e-05,
+      "loss": 0.146,
+      "step": 1863
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3793524546768358e-05,
+      "loss": 0.1821,
+      "step": 1864
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3787174231471112e-05,
+      "loss": 0.1811,
+      "step": 1865
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.37808221327136e-05,
+      "loss": 0.1405,
+      "step": 1866
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3774468253487165e-05,
+      "loss": 0.1424,
+      "step": 1867
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.376811259678398e-05,
+      "loss": 0.186,
+      "step": 1868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3761755165597057e-05,
+      "loss": 0.1454,
+      "step": 1869
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.375539596292025e-05,
+      "loss": 0.2157,
+      "step": 1870
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3749034991748233e-05,
+      "loss": 0.1438,
+      "step": 1871
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3742672255076528e-05,
+      "loss": 0.1802,
+      "step": 1872
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373630775590148e-05,
+      "loss": 0.2257,
+      "step": 1873
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3729941497220267e-05,
+      "loss": 0.2059,
+      "step": 1874
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3723573482030892e-05,
+      "loss": 0.2307,
+      "step": 1875
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.371720371333219e-05,
+      "loss": 0.2534,
+      "step": 1876
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3710832194123817e-05,
+      "loss": 0.2017,
+      "step": 1877
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704458927406261e-05,
+      "loss": 0.172,
+      "step": 1878
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3698083916180823e-05,
+      "loss": 0.1564,
+      "step": 1879
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3691707163449631e-05,
+      "loss": 0.1719,
+      "step": 1880
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3685328672215634e-05,
+      "loss": 0.2163,
+      "step": 1881
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3678948445482598e-05,
+      "loss": 0.1708,
+      "step": 1882
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3672566486255105e-05,
+      "loss": 0.2087,
+      "step": 1883
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3666182797538554e-05,
+      "loss": 0.2143,
+      "step": 1884
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3659797382339162e-05,
+      "loss": 0.1983,
+      "step": 1885
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.1771,
+      "step": 1886
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3647021384520761e-05,
+      "loss": 0.2204,
+      "step": 1887
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3640630807918246e-05,
+      "loss": 0.1538,
+      "step": 1888
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3634238516865858e-05,
+      "loss": 0.1738,
+      "step": 1889
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.362784451437386e-05,
+      "loss": 0.142,
+      "step": 1890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3621448803453325e-05,
+      "loss": 0.14,
+      "step": 1891
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3615051387116131e-05,
+      "loss": 0.1585,
+      "step": 1892
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3608652268374952e-05,
+      "loss": 0.1341,
+      "step": 1893
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3602251450243273e-05,
+      "loss": 0.1235,
+      "step": 1894
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3595848935735365e-05,
+      "loss": 0.1473,
+      "step": 1895
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.358944472786632e-05,
+      "loss": 0.173,
+      "step": 1896
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3583038829652005e-05,
+      "loss": 0.1611,
+      "step": 1897
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3576631244109093e-05,
+      "loss": 0.1193,
+      "step": 1898
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.357022197425506e-05,
+      "loss": 0.1391,
+      "step": 1899
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3563811023108157e-05,
+      "loss": 0.1768,
+      "step": 1900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355739839368744e-05,
+      "loss": 0.1928,
+      "step": 1901
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355098408901275e-05,
+      "loss": 0.1611,
+      "step": 1902
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.354456811210472e-05,
+      "loss": 0.1416,
+      "step": 1903
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3538150465984771e-05,
+      "loss": 0.1566,
+      "step": 1904
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3531731153675103e-05,
+      "loss": 0.1598,
+      "step": 1905
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3525310178198707e-05,
+      "loss": 0.1535,
+      "step": 1906
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.351888754257936e-05,
+      "loss": 0.1725,
+      "step": 1907
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3512463249841613e-05,
+      "loss": 0.1147,
+      "step": 1908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3506037303010799e-05,
+      "loss": 0.1643,
+      "step": 1909
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3499609705113034e-05,
+      "loss": 0.1549,
+      "step": 1910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3493180459175212e-05,
+      "loss": 0.1337,
+      "step": 1911
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3486749568225002e-05,
+      "loss": 0.1683,
+      "step": 1912
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3480317035290837e-05,
+      "loss": 0.1188,
+      "step": 1913
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3473882863401941e-05,
+      "loss": 0.1577,
+      "step": 1914
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3467447055588298e-05,
+      "loss": 0.1597,
+      "step": 1915
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3461009614880668e-05,
+      "loss": 0.1733,
+      "step": 1916
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3454570544310575e-05,
+      "loss": 0.1405,
+      "step": 1917
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3448129846910312e-05,
+      "loss": 0.1519,
+      "step": 1918
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3441687525712943e-05,
+      "loss": 0.1237,
+      "step": 1919
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3435243583752294e-05,
+      "loss": 0.1496,
+      "step": 1920
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3428798024062946e-05,
+      "loss": 0.1511,
+      "step": 1921
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3422350849680258e-05,
+      "loss": 0.2133,
+      "step": 1922
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3415902063640333e-05,
+      "loss": 0.1105,
+      "step": 1923
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3409451668980047e-05,
+      "loss": 0.1752,
+      "step": 1924
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3402999668737022e-05,
+      "loss": 0.1046,
+      "step": 1925
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3396546065949643e-05,
+      "loss": 0.1561,
+      "step": 1926
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3390090863657048e-05,
+      "loss": 0.1278,
+      "step": 1927
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3383634064899127e-05,
+      "loss": 0.1901,
+      "step": 1928
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3377175672716522e-05,
+      "loss": 0.1414,
+      "step": 1929
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3370715690150631e-05,
+      "loss": 0.1474,
+      "step": 1930
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3364254120243589e-05,
+      "loss": 0.13,
+      "step": 1931
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3357790966038292e-05,
+      "loss": 0.1106,
+      "step": 1932
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335132623057837e-05,
+      "loss": 0.1293,
+      "step": 1933
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3344859916908206e-05,
+      "loss": 0.1207,
+      "step": 1934
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3338392028072926e-05,
+      "loss": 0.1569,
+      "step": 1935
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3331922567118394e-05,
+      "loss": 0.1297,
+      "step": 1936
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3325451537091212e-05,
+      "loss": 0.1221,
+      "step": 1937
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.331897894103873e-05,
+      "loss": 0.1776,
+      "step": 1938
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3312504782009026e-05,
+      "loss": 0.1317,
+      "step": 1939
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.330602906305092e-05,
+      "loss": 0.1734,
+      "step": 1940
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3299551787213964e-05,
+      "loss": 0.1103,
+      "step": 1941
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3293072957548443e-05,
+      "loss": 0.132,
+      "step": 1942
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3286592577105379e-05,
+      "loss": 0.1702,
+      "step": 1943
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280110648936512e-05,
+      "loss": 0.1473,
+      "step": 1944
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3273627176094326e-05,
+      "loss": 0.1231,
+      "step": 1945
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267142161632025e-05,
+      "loss": 0.1492,
+      "step": 1946
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3260655608603534e-05,
+      "loss": 0.1154,
+      "step": 1947
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.325416752006351e-05,
+      "loss": 0.1523,
+      "step": 1948
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3247677899067332e-05,
+      "loss": 0.1199,
+      "step": 1949
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.32411867486711e-05,
+      "loss": 0.1336,
+      "step": 1950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3234694071931631e-05,
+      "loss": 0.1717,
+      "step": 1951
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3228199871906465e-05,
+      "loss": 0.1573,
+      "step": 1952
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3221704151653856e-05,
+      "loss": 0.147,
+      "step": 1953
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.321520691423278e-05,
+      "loss": 0.1826,
+      "step": 1954
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3208708162702921e-05,
+      "loss": 0.1361,
+      "step": 1955
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3202207900124677e-05,
+      "loss": 0.1762,
+      "step": 1956
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.319570612955916e-05,
+      "loss": 0.1036,
+      "step": 1957
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.318920285406819e-05,
+      "loss": 0.1907,
+      "step": 1958
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.31826980767143e-05,
+      "loss": 0.1636,
+      "step": 1959
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.317619180056072e-05,
+      "loss": 0.0865,
+      "step": 1960
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.31696840286714e-05,
+      "loss": 0.1869,
+      "step": 1961
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3163174764110985e-05,
+      "loss": 0.1532,
+      "step": 1962
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3156664009944818e-05,
+      "loss": 0.1431,
+      "step": 1963
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3150151769238964e-05,
+      "loss": 0.1664,
+      "step": 1964
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3143638045060164e-05,
+      "loss": 0.1176,
+      "step": 1965
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313712284047587e-05,
+      "loss": 0.1013,
+      "step": 1966
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3130606158554233e-05,
+      "loss": 0.116,
+      "step": 1967
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3124088002364093e-05,
+      "loss": 0.1723,
+      "step": 1968
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.311756837497499e-05,
+      "loss": 0.1176,
+      "step": 1969
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311104727945715e-05,
+      "loss": 0.1631,
+      "step": 1970
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3104524718881494e-05,
+      "loss": 0.1046,
+      "step": 1971
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3098000696319642e-05,
+      "loss": 0.151,
+      "step": 1972
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3091475214843885e-05,
+      "loss": 0.1267,
+      "step": 1973
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3084948277527211e-05,
+      "loss": 0.1233,
+      "step": 1974
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3078419887443299e-05,
+      "loss": 0.1271,
+      "step": 1975
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3071890047666498e-05,
+      "loss": 0.1504,
+      "step": 1976
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3065358761271853e-05,
+      "loss": 0.2374,
+      "step": 1977
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.305882603133508e-05,
+      "loss": 0.1234,
+      "step": 1978
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3052291860932577e-05,
+      "loss": 0.1774,
+      "step": 1979
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3045756253141435e-05,
+      "loss": 0.1388,
+      "step": 1980
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3039219211039398e-05,
+      "loss": 0.1542,
+      "step": 1981
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3032680737704902e-05,
+      "loss": 0.1302,
+      "step": 1982
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.302614083621705e-05,
+      "loss": 0.1357,
+      "step": 1983
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.301959950965562e-05,
+      "loss": 0.1108,
+      "step": 1984
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3013056761101064e-05,
+      "loss": 0.1432,
+      "step": 1985
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3006512593634492e-05,
+      "loss": 0.1683,
+      "step": 1986
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2999967010337702e-05,
+      "loss": 0.0961,
+      "step": 1987
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2993420014293139e-05,
+      "loss": 0.1425,
+      "step": 1988
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2986871608583921e-05,
+      "loss": 0.1922,
+      "step": 1989
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2980321796293838e-05,
+      "loss": 0.1593,
+      "step": 1990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2973770580507328e-05,
+      "loss": 0.2113,
+      "step": 1991
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.29672179643095e-05,
+      "loss": 0.1302,
+      "step": 1992
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2960663950786121e-05,
+      "loss": 0.0911,
+      "step": 1993
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2954108543023605e-05,
+      "loss": 0.1479,
+      "step": 1994
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2947551744109044e-05,
+      "loss": 0.0934,
+      "step": 1995
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2940993557130166e-05,
+      "loss": 0.1248,
+      "step": 1996
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2934433985175359e-05,
+      "loss": 0.0987,
+      "step": 1997
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2927873031333666e-05,
+      "loss": 0.174,
+      "step": 1998
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2921310698694783e-05,
+      "loss": 0.1139,
+      "step": 1999
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2914746990349041e-05,
+      "loss": 0.1275,
+      "step": 2000
+    }
+  ],
+  "max_steps": 4720,
+  "num_train_epochs": 5,
+  "total_flos": 1.4359173018733773e+17,
+  "trial_name": null,
+  "trial_params": null
+}