diff --git "a/checkpoint-3546/trainer_state.json" "b/checkpoint-3546/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3546/trainer_state.json"
@@ -0,0 +1,21416 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9987315010570823,
+  "eval_steps": 236,
+  "global_step": 3546,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507045e-08,
+      "loss": 1.3418,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-08,
+      "loss": 1.6801,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352113e-08,
+      "loss": 1.3628,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802818e-07,
+      "loss": 1.3222,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253522e-07,
+      "loss": 1.519,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6901408450704225e-07,
+      "loss": 1.5033,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9718309859154932e-07,
+      "loss": 1.2227,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2535211267605636e-07,
+      "loss": 1.5744,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.535211267605634e-07,
+      "loss": 1.1132,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 1.2757,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0985915492957747e-07,
+      "loss": 1.6722,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.380281690140845e-07,
+      "loss": 1.5187,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.661971830985916e-07,
+      "loss": 1.2149,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9436619718309864e-07,
+      "loss": 1.9633,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.225352112676057e-07,
+      "loss": 1.5469,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.507042253521127e-07,
+      "loss": 1.2916,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788732394366198e-07,
+      "loss": 1.3984,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.070422535211268e-07,
+      "loss": 1.3255,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.352112676056338e-07,
+      "loss": 1.3907,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 1.3508,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.915492957746479e-07,
+      "loss": 1.3875,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.197183098591549e-07,
+      "loss": 1.2283,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.47887323943662e-07,
+      "loss": 1.432,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.76056338028169e-07,
+      "loss": 1.4333,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.042253521126762e-07,
+      "loss": 1.4205,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.323943661971832e-07,
+      "loss": 1.2151,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.605633802816901e-07,
+      "loss": 1.3993,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.887323943661973e-07,
+      "loss": 1.7116,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.169014084507043e-07,
+      "loss": 1.6577,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 1.439,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.732394366197183e-07,
+      "loss": 1.5013,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.014084507042254e-07,
+      "loss": 1.514,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.295774647887325e-07,
+      "loss": 1.5555,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.577464788732395e-07,
+      "loss": 1.6068,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.859154929577465e-07,
+      "loss": 1.3398,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0140845070422536e-06,
+      "loss": 1.4168,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0422535211267606e-06,
+      "loss": 1.5454,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0704225352112677e-06,
+      "loss": 1.4684,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.098591549295775e-06,
+      "loss": 1.3311,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 1.2699,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1549295774647888e-06,
+      "loss": 1.4445,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1830985915492958e-06,
+      "loss": 1.221,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.211267605633803e-06,
+      "loss": 1.4705,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2394366197183099e-06,
+      "loss": 1.4435,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.267605633802817e-06,
+      "loss": 1.3422,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.295774647887324e-06,
+      "loss": 1.1821,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.323943661971831e-06,
+      "loss": 1.2586,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.352112676056338e-06,
+      "loss": 1.5441,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3802816901408453e-06,
+      "loss": 1.2419,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 1.3732,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4366197183098594e-06,
+      "loss": 1.3104,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4647887323943664e-06,
+      "loss": 1.2385,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4929577464788732e-06,
+      "loss": 1.4739,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5211267605633803e-06,
+      "loss": 1.1976,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5492957746478873e-06,
+      "loss": 1.3504,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5774647887323946e-06,
+      "loss": 1.6489,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6056338028169016e-06,
+      "loss": 1.2824,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6338028169014086e-06,
+      "loss": 1.3271,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6619718309859157e-06,
+      "loss": 1.2428,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 1.2441,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7183098591549297e-06,
+      "loss": 1.3945,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7464788732394366e-06,
+      "loss": 1.4281,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.774647887323944e-06,
+      "loss": 1.3462,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8028169014084509e-06,
+      "loss": 1.3179,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8309859154929579e-06,
+      "loss": 1.3657,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.859154929577465e-06,
+      "loss": 1.4192,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.887323943661972e-06,
+      "loss": 1.258,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.915492957746479e-06,
+      "loss": 1.3094,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.943661971830986e-06,
+      "loss": 1.2587,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 1.279,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.2552,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.028169014084507e-06,
+      "loss": 1.2978,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0563380281690144e-06,
+      "loss": 1.333,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0845070422535212e-06,
+      "loss": 1.2979,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1126760563380285e-06,
+      "loss": 1.1694,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1408450704225353e-06,
+      "loss": 1.487,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.169014084507042e-06,
+      "loss": 1.1773,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19718309859155e-06,
+      "loss": 1.1597,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2253521126760566e-06,
+      "loss": 1.2153,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 1.4597,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2816901408450707e-06,
+      "loss": 1.2858,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3098591549295775e-06,
+      "loss": 1.2605,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.338028169014085e-06,
+      "loss": 1.2524,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3661971830985916e-06,
+      "loss": 1.2135,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3943661971830984e-06,
+      "loss": 1.2342,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.422535211267606e-06,
+      "loss": 1.2146,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.450704225352113e-06,
+      "loss": 1.2575,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4788732394366198e-06,
+      "loss": 1.3696,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.507042253521127e-06,
+      "loss": 1.221,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 1.2187,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.563380281690141e-06,
+      "loss": 1.3197,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.591549295774648e-06,
+      "loss": 1.2436,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.619718309859155e-06,
+      "loss": 1.2079,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.647887323943662e-06,
+      "loss": 1.1906,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.676056338028169e-06,
+      "loss": 1.2254,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.704225352112676e-06,
+      "loss": 1.1684,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7323943661971837e-06,
+      "loss": 1.2176,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7605633802816906e-06,
+      "loss": 1.2144,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7887323943661974e-06,
+      "loss": 1.1904,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 1.4412,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8450704225352115e-06,
+      "loss": 1.403,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8732394366197187e-06,
+      "loss": 1.2991,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9014084507042255e-06,
+      "loss": 1.177,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.929577464788733e-06,
+      "loss": 1.2982,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9577464788732396e-06,
+      "loss": 1.1565,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9859154929577465e-06,
+      "loss": 1.3319,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0140845070422537e-06,
+      "loss": 1.4427,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0422535211267605e-06,
+      "loss": 1.2611,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0704225352112678e-06,
+      "loss": 1.1248,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 1.2654,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1267605633802823e-06,
+      "loss": 1.1998,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.154929577464789e-06,
+      "loss": 1.4276,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1830985915492964e-06,
+      "loss": 1.2273,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.211267605633803e-06,
+      "loss": 1.1384,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2394366197183104e-06,
+      "loss": 1.2032,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2676056338028173e-06,
+      "loss": 1.0634,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.295774647887324e-06,
+      "loss": 1.2516,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3239436619718313e-06,
+      "loss": 1.2097,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.352112676056338e-06,
+      "loss": 1.2057,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 1.0051,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4084507042253522e-06,
+      "loss": 1.3015,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4366197183098595e-06,
+      "loss": 1.1629,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4647887323943663e-06,
+      "loss": 1.0463,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.492957746478873e-06,
+      "loss": 1.4034,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5211267605633804e-06,
+      "loss": 1.3264,
+      "step": 125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.549295774647888e-06,
+      "loss": 1.129,
+      "step": 126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.577464788732395e-06,
+      "loss": 1.3082,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6056338028169017e-06,
+      "loss": 1.2496,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.633802816901409e-06,
+      "loss": 1.35,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 1.1036,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.690140845070423e-06,
+      "loss": 1.1417,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.71830985915493e-06,
+      "loss": 1.2943,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.746478873239437e-06,
+      "loss": 1.3206,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.774647887323944e-06,
+      "loss": 1.1397,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8028169014084508e-06,
+      "loss": 1.2328,
+      "step": 135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.830985915492958e-06,
+      "loss": 1.3558,
+      "step": 136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.859154929577465e-06,
+      "loss": 1.1015,
+      "step": 137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.887323943661972e-06,
+      "loss": 1.2189,
+      "step": 138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.915492957746479e-06,
+      "loss": 1.3142,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 1.2039,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.971830985915493e-06,
+      "loss": 1.0501,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3026,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.028169014084508e-06,
+      "loss": 1.1499,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.056338028169014e-06,
+      "loss": 1.3819,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0845070422535216e-06,
+      "loss": 1.2836,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.112676056338029e-06,
+      "loss": 1.1178,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.140845070422535e-06,
+      "loss": 1.1516,
+      "step": 147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.1690140845070425e-06,
+      "loss": 1.1855,
+      "step": 148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.19718309859155e-06,
+      "loss": 1.1717,
+      "step": 149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 1.3517,
+      "step": 150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.253521126760563e-06,
+      "loss": 1.1338,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.281690140845071e-06,
+      "loss": 1.1674,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.309859154929578e-06,
+      "loss": 1.3618,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.338028169014084e-06,
+      "loss": 1.1007,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3661971830985915e-06,
+      "loss": 1.1097,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3943661971831e-06,
+      "loss": 1.1588,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.422535211267606e-06,
+      "loss": 1.0512,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.450704225352113e-06,
+      "loss": 1.2985,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4788732394366205e-06,
+      "loss": 1.2587,
+      "step": 159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 1.4696,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.535211267605634e-06,
+      "loss": 1.2118,
+      "step": 161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5633802816901414e-06,
+      "loss": 1.2436,
+      "step": 162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.591549295774648e-06,
+      "loss": 1.2545,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.619718309859155e-06,
+      "loss": 1.21,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.647887323943662e-06,
+      "loss": 1.0979,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.67605633802817e-06,
+      "loss": 1.2912,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.704225352112676e-06,
+      "loss": 1.1633,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.732394366197183e-06,
+      "loss": 1.3976,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7605633802816905e-06,
+      "loss": 1.2334,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 1.4072,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.816901408450705e-06,
+      "loss": 1.4666,
+      "step": 171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.845070422535212e-06,
+      "loss": 1.2881,
+      "step": 172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.873239436619719e-06,
+      "loss": 1.1987,
+      "step": 173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.901408450704226e-06,
+      "loss": 1.4005,
+      "step": 174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.929577464788733e-06,
+      "loss": 1.2795,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9577464788732395e-06,
+      "loss": 1.4463,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.985915492957747e-06,
+      "loss": 1.2518,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.014084507042254e-06,
+      "loss": 1.2853,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.042253521126761e-06,
+      "loss": 1.2258,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 1.1489,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.098591549295775e-06,
+      "loss": 1.229,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.126760563380282e-06,
+      "loss": 1.2392,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.154929577464789e-06,
+      "loss": 1.339,
+      "step": 183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.183098591549296e-06,
+      "loss": 1.1776,
+      "step": 184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.211267605633803e-06,
+      "loss": 1.2591,
+      "step": 185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.23943661971831e-06,
+      "loss": 1.144,
+      "step": 186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.267605633802817e-06,
+      "loss": 1.3478,
+      "step": 187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.295774647887324e-06,
+      "loss": 1.2036,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.323943661971831e-06,
+      "loss": 1.1938,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 1.2662,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.380281690140845e-06,
+      "loss": 1.235,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.408450704225352e-06,
+      "loss": 1.1768,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.43661971830986e-06,
+      "loss": 1.1144,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4647887323943675e-06,
+      "loss": 1.4197,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.492957746478874e-06,
+      "loss": 1.2688,
+      "step": 195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.521126760563381e-06,
+      "loss": 1.2287,
+      "step": 196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.549295774647888e-06,
+      "loss": 1.2163,
+      "step": 197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.577464788732395e-06,
+      "loss": 1.0609,
+      "step": 198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.605633802816902e-06,
+      "loss": 1.1017,
+      "step": 199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 1.185,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.6619718309859165e-06,
+      "loss": 1.0727,
+      "step": 201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.690140845070423e-06,
+      "loss": 1.5206,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.71830985915493e-06,
+      "loss": 1.0901,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7464788732394374e-06,
+      "loss": 1.3183,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.774647887323944e-06,
+      "loss": 1.3105,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.802816901408451e-06,
+      "loss": 1.3785,
+      "step": 206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.830985915492958e-06,
+      "loss": 1.1496,
+      "step": 207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.859154929577466e-06,
+      "loss": 1.167,
+      "step": 208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.887323943661972e-06,
+      "loss": 1.2458,
+      "step": 209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 1.1596,
+      "step": 210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9436619718309865e-06,
+      "loss": 1.1357,
+      "step": 211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971830985915493e-06,
+      "loss": 0.9843,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6e-06,
+      "loss": 1.2245,
+      "step": 213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.028169014084507e-06,
+      "loss": 1.0025,
+      "step": 214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.056338028169015e-06,
+      "loss": 1.1183,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.084507042253521e-06,
+      "loss": 1.5397,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.112676056338028e-06,
+      "loss": 1.0552,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.1408450704225356e-06,
+      "loss": 1.3601,
+      "step": 218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.169014084507042e-06,
+      "loss": 1.2845,
+      "step": 219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 1.0935,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2253521126760565e-06,
+      "loss": 1.146,
+      "step": 221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.2535211267605646e-06,
+      "loss": 1.1768,
+      "step": 222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.281690140845072e-06,
+      "loss": 1.1167,
+      "step": 223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.309859154929578e-06,
+      "loss": 1.3253,
+      "step": 224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.3380281690140855e-06,
+      "loss": 1.1376,
+      "step": 225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.366197183098593e-06,
+      "loss": 1.3858,
+      "step": 226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.394366197183099e-06,
+      "loss": 1.0459,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.422535211267606e-06,
+      "loss": 1.1447,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.450704225352114e-06,
+      "loss": 1.1114,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 1.464,
+      "step": 230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.507042253521127e-06,
+      "loss": 1.2157,
+      "step": 231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.5352112676056345e-06,
+      "loss": 1.122,
+      "step": 232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.563380281690142e-06,
+      "loss": 1.1739,
+      "step": 233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.591549295774648e-06,
+      "loss": 1.3235,
+      "step": 234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.619718309859155e-06,
+      "loss": 1.2631,
+      "step": 235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.647887323943663e-06,
+      "loss": 1.2404,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.1165865659713745,
+      "eval_runtime": 269.3914,
+      "eval_samples_per_second": 3.708,
+      "eval_steps_per_second": 0.464,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.67605633802817e-06,
+      "loss": 1.1415,
+      "step": 237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.704225352112676e-06,
+      "loss": 1.2497,
+      "step": 238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.7323943661971836e-06,
+      "loss": 1.1512,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 1.0767,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.788732394366197e-06,
+      "loss": 0.9569,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.8169014084507045e-06,
+      "loss": 1.35,
+      "step": 242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.845070422535212e-06,
+      "loss": 1.2382,
+      "step": 243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.873239436619719e-06,
+      "loss": 1.1095,
+      "step": 244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.901408450704225e-06,
+      "loss": 1.2218,
+      "step": 245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.929577464788733e-06,
+      "loss": 1.382,
+      "step": 246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.95774647887324e-06,
+      "loss": 1.369,
+      "step": 247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.985915492957746e-06,
+      "loss": 1.2502,
+      "step": 248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.0140845070422535e-06,
+      "loss": 1.0344,
+      "step": 249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 1.2101,
+      "step": 250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.070422535211268e-06,
+      "loss": 1.2263,
+      "step": 251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.098591549295776e-06,
+      "loss": 1.1012,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.1267605633802825e-06,
+      "loss": 1.2301,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.15492957746479e-06,
+      "loss": 1.2226,
+      "step": 254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.183098591549297e-06,
+      "loss": 1.2694,
+      "step": 255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.211267605633803e-06,
+      "loss": 1.4335,
+      "step": 256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.239436619718311e-06,
+      "loss": 1.1342,
+      "step": 257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.267605633802818e-06,
+      "loss": 1.2146,
+      "step": 258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.295774647887325e-06,
+      "loss": 1.1116,
+      "step": 259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 1.2331,
+      "step": 260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.352112676056339e-06,
+      "loss": 1.0549,
+      "step": 261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.380281690140846e-06,
+      "loss": 1.222,
+      "step": 262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.4084507042253525e-06,
+      "loss": 1.1478,
+      "step": 263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.43661971830986e-06,
+      "loss": 1.2954,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.464788732394367e-06,
+      "loss": 1.196,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.492957746478874e-06,
+      "loss": 1.0139,
+      "step": 266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.521126760563381e-06,
+      "loss": 1.5008,
+      "step": 267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.549295774647888e-06,
+      "loss": 1.1822,
+      "step": 268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.577464788732395e-06,
+      "loss": 1.103,
+      "step": 269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 1.1599,
+      "step": 270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.633802816901409e-06,
+      "loss": 1.1041,
+      "step": 271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.661971830985916e-06,
+      "loss": 1.4274,
+      "step": 272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.690140845070423e-06,
+      "loss": 1.1177,
+      "step": 273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.71830985915493e-06,
+      "loss": 1.3761,
+      "step": 274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.746478873239436e-06,
+      "loss": 1.1913,
+      "step": 275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.774647887323943e-06,
+      "loss": 1.0957,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.80281690140845e-06,
+      "loss": 1.0218,
+      "step": 277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.830985915492958e-06,
+      "loss": 1.2873,
+      "step": 278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.859154929577465e-06,
+      "loss": 1.2246,
+      "step": 279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 1.2641,
+      "step": 280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.91549295774648e-06,
+      "loss": 1.2165,
+      "step": 281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.943661971830987e-06,
+      "loss": 1.1298,
+      "step": 282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 7.971830985915494e-06,
+      "loss": 1.2814,
+      "step": 283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.0617,
+      "step": 284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.028169014084509e-06,
+      "loss": 1.1663,
+      "step": 285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.056338028169016e-06,
+      "loss": 1.1186,
+      "step": 286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.084507042253521e-06,
+      "loss": 1.3063,
+      "step": 287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.112676056338029e-06,
+      "loss": 1.2076,
+      "step": 288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.140845070422536e-06,
+      "loss": 1.084,
+      "step": 289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 1.3851,
+      "step": 290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.19718309859155e-06,
+      "loss": 1.3329,
+      "step": 291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.225352112676058e-06,
+      "loss": 1.1447,
+      "step": 292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.253521126760565e-06,
+      "loss": 1.1243,
+      "step": 293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.28169014084507e-06,
+      "loss": 1.1835,
+      "step": 294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.309859154929578e-06,
+      "loss": 1.0693,
+      "step": 295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.338028169014085e-06,
+      "loss": 1.0549,
+      "step": 296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.366197183098592e-06,
+      "loss": 1.0551,
+      "step": 297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.3943661971831e-06,
+      "loss": 1.3071,
+      "step": 298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.422535211267607e-06,
+      "loss": 1.2693,
+      "step": 299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 1.3027,
+      "step": 300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.47887323943662e-06,
+      "loss": 1.0142,
+      "step": 301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.507042253521127e-06,
+      "loss": 1.3342,
+      "step": 302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.535211267605634e-06,
+      "loss": 1.1116,
+      "step": 303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.563380281690141e-06,
+      "loss": 1.1299,
+      "step": 304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.591549295774648e-06,
+      "loss": 1.2015,
+      "step": 305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.619718309859156e-06,
+      "loss": 1.2035,
+      "step": 306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647887323943663e-06,
+      "loss": 1.1743,
+      "step": 307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676056338028169e-06,
+      "loss": 1.1202,
+      "step": 308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704225352112676e-06,
+      "loss": 1.4374,
+      "step": 309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 1.3139,
+      "step": 310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.760563380281692e-06,
+      "loss": 1.1208,
+      "step": 311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7887323943662e-06,
+      "loss": 1.2294,
+      "step": 312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.816901408450705e-06,
+      "loss": 1.2346,
+      "step": 313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.845070422535212e-06,
+      "loss": 1.2432,
+      "step": 314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.87323943661972e-06,
+      "loss": 1.1142,
+      "step": 315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.901408450704227e-06,
+      "loss": 1.1793,
+      "step": 316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.929577464788734e-06,
+      "loss": 1.2635,
+      "step": 317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.957746478873241e-06,
+      "loss": 1.2721,
+      "step": 318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.985915492957748e-06,
+      "loss": 1.1499,
+      "step": 319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 1.414,
+      "step": 320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.042253521126761e-06,
+      "loss": 1.3243,
+      "step": 321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.070422535211268e-06,
+      "loss": 1.2385,
+      "step": 322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.098591549295776e-06,
+      "loss": 1.4159,
+      "step": 323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.126760563380283e-06,
+      "loss": 1.1442,
+      "step": 324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.15492957746479e-06,
+      "loss": 1.1827,
+      "step": 325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.183098591549296e-06,
+      "loss": 1.0153,
+      "step": 326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.211267605633803e-06,
+      "loss": 1.1522,
+      "step": 327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.23943661971831e-06,
+      "loss": 1.0683,
+      "step": 328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.267605633802817e-06,
+      "loss": 1.0879,
+      "step": 329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 1.0873,
+      "step": 330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.323943661971832e-06,
+      "loss": 1.3289,
+      "step": 331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.35211267605634e-06,
+      "loss": 1.224,
+      "step": 332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.380281690140845e-06,
+      "loss": 1.2185,
+      "step": 333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.408450704225352e-06,
+      "loss": 1.1307,
+      "step": 334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.43661971830986e-06,
+      "loss": 1.1897,
+      "step": 335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.464788732394366e-06,
+      "loss": 1.3596,
+      "step": 336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.492957746478874e-06,
+      "loss": 1.055,
+      "step": 337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.521126760563381e-06,
+      "loss": 1.2561,
+      "step": 338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.549295774647888e-06,
+      "loss": 1.0939,
+      "step": 339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 1.2203,
+      "step": 340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.605633802816903e-06,
+      "loss": 1.2119,
+      "step": 341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.63380281690141e-06,
+      "loss": 1.1143,
+      "step": 342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.661971830985917e-06,
+      "loss": 1.1352,
+      "step": 343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.690140845070424e-06,
+      "loss": 0.9916,
+      "step": 344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.71830985915493e-06,
+      "loss": 1.131,
+      "step": 345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.746478873239437e-06,
+      "loss": 1.2705,
+      "step": 346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.774647887323945e-06,
+      "loss": 1.0525,
+      "step": 347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.802816901408452e-06,
+      "loss": 1.147,
+      "step": 348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.830985915492959e-06,
+      "loss": 1.2475,
+      "step": 349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 1.3519,
+      "step": 350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.887323943661974e-06,
+      "loss": 1.3495,
+      "step": 351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.915492957746479e-06,
+      "loss": 1.088,
+      "step": 352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.943661971830986e-06,
+      "loss": 1.2713,
+      "step": 353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.971830985915494e-06,
+      "loss": 1.305,
+      "step": 354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1e-05,
+      "loss": 1.2236,
+      "step": 355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.999997576817574e-06,
+      "loss": 1.3538,
+      "step": 356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.999990307272647e-06,
+      "loss": 1.1192,
+      "step": 357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.999978191372262e-06,
+      "loss": 1.6531,
+      "step": 358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.999961229128163e-06,
+      "loss": 1.2085,
+      "step": 359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.999939420556794e-06,
+      "loss": 1.1289,
+      "step": 360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.99991276567929e-06,
+      "loss": 1.2806,
+      "step": 361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.99988126452149e-06,
+      "loss": 1.2833,
+      "step": 362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999844917113923e-06,
+      "loss": 1.2652,
+      "step": 363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999803723491823e-06,
+      "loss": 1.2422,
+      "step": 364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999757683695118e-06,
+      "loss": 1.1305,
+      "step": 365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999706797768432e-06,
+      "loss": 1.0564,
+      "step": 366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999651065761086e-06,
+      "loss": 1.3385,
+      "step": 367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999590487727103e-06,
+      "loss": 1.1687,
+      "step": 368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999525063725197e-06,
+      "loss": 1.1614,
+      "step": 369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999454793818782e-06,
+      "loss": 1.0145,
+      "step": 370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.99937967807597e-06,
+      "loss": 1.2096,
+      "step": 371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.999299716569567e-06,
+      "loss": 1.1507,
+      "step": 372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.999214909377078e-06,
+      "loss": 1.1685,
+      "step": 373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.999125256580707e-06,
+      "loss": 1.0961,
+      "step": 374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.999030758267348e-06,
+      "loss": 1.2691,
+      "step": 375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998931414528597e-06,
+      "loss": 1.1904,
+      "step": 376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998827225460745e-06,
+      "loss": 1.0868,
+      "step": 377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.99871819116478e-06,
+      "loss": 0.9782,
+      "step": 378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998604311746387e-06,
+      "loss": 1.0768,
+      "step": 379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998485587315945e-06,
+      "loss": 1.2106,
+      "step": 380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998362017988532e-06,
+      "loss": 1.1554,
+      "step": 381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998233603883917e-06,
+      "loss": 1.1672,
+      "step": 382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.998100345126573e-06,
+      "loss": 1.3399,
+      "step": 383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.99796224184566e-06,
+      "loss": 1.2215,
+      "step": 384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.99781929417504e-06,
+      "loss": 1.1036,
+      "step": 385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.997671502253268e-06,
+      "loss": 1.1551,
+      "step": 386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.997518866223593e-06,
+      "loss": 1.2181,
+      "step": 387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.997361386233963e-06,
+      "loss": 1.2378,
+      "step": 388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.99719906243702e-06,
+      "loss": 1.2212,
+      "step": 389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.997031894990097e-06,
+      "loss": 1.1687,
+      "step": 390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.996859884055226e-06,
+      "loss": 1.1656,
+      "step": 391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.996683029799135e-06,
+      "loss": 1.266,
+      "step": 392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.996501332393241e-06,
+      "loss": 1.1556,
+      "step": 393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.996314792013659e-06,
+      "loss": 1.1053,
+      "step": 394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.996123408841198e-06,
+      "loss": 1.077,
+      "step": 395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.995927183061361e-06,
+      "loss": 1.0333,
+      "step": 396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.995726114864343e-06,
+      "loss": 1.4151,
+      "step": 397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.995520204445036e-06,
+      "loss": 1.171,
+      "step": 398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.995309452003022e-06,
+      "loss": 1.1679,
+      "step": 399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.995093857742579e-06,
+      "loss": 1.1301,
+      "step": 400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.994873421872673e-06,
+      "loss": 1.1813,
+      "step": 401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.994648144606971e-06,
+      "loss": 1.1935,
+      "step": 402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.994418026163826e-06,
+      "loss": 1.1786,
+      "step": 403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.994183066766285e-06,
+      "loss": 1.3293,
+      "step": 404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.99394326664209e-06,
+      "loss": 1.008,
+      "step": 405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.99369862602367e-06,
+      "loss": 1.1245,
+      "step": 406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.993449145148151e-06,
+      "loss": 1.1093,
+      "step": 407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.993194824257346e-06,
+      "loss": 1.192,
+      "step": 408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.992935663597764e-06,
+      "loss": 1.1796,
+      "step": 409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.9926716634206e-06,
+      "loss": 1.35,
+      "step": 410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.992402823981743e-06,
+      "loss": 1.1076,
+      "step": 411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.992129145541771e-06,
+      "loss": 1.2868,
+      "step": 412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.991850628365956e-06,
+      "loss": 1.034,
+      "step": 413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.991567272724254e-06,
+      "loss": 1.171,
+      "step": 414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.991279078891313e-06,
+      "loss": 1.1568,
+      "step": 415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.990986047146477e-06,
+      "loss": 1.3956,
+      "step": 416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.990688177773769e-06,
+      "loss": 1.0784,
+      "step": 417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.990385471061908e-06,
+      "loss": 1.1396,
+      "step": 418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.990077927304296e-06,
+      "loss": 1.0309,
+      "step": 419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.989765546799032e-06,
+      "loss": 1.1868,
+      "step": 420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.989448329848895e-06,
+      "loss": 1.1238,
+      "step": 421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.989126276761355e-06,
+      "loss": 1.3391,
+      "step": 422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.988799387848567e-06,
+      "loss": 1.2333,
+      "step": 423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.988467663427381e-06,
+      "loss": 1.3088,
+      "step": 424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.988131103819326e-06,
+      "loss": 1.2754,
+      "step": 425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.987789709350618e-06,
+      "loss": 1.05,
+      "step": 426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.987443480352164e-06,
+      "loss": 1.098,
+      "step": 427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.987092417159552e-06,
+      "loss": 1.1603,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.986736520113062e-06,
+      "loss": 1.2035,
+      "step": 429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.986375789557652e-06,
+      "loss": 1.2774,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.98601022584297e-06,
+      "loss": 1.1506,
+      "step": 431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.985639829323345e-06,
+      "loss": 1.1668,
+      "step": 432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.985264600357795e-06,
+      "loss": 1.2493,
+      "step": 433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.984884539310017e-06,
+      "loss": 1.3186,
+      "step": 434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.984499646548395e-06,
+      "loss": 1.0414,
+      "step": 435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.984109922445998e-06,
+      "loss": 1.3101,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.983715367380568e-06,
+      "loss": 1.1813,
+      "step": 437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.983315981734544e-06,
+      "loss": 1.112,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.982911765895035e-06,
+      "loss": 1.2157,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.982502720253838e-06,
+      "loss": 1.3154,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.98208884520743e-06,
+      "loss": 1.279,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.98167014115697e-06,
+      "loss": 1.2873,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.981246608508293e-06,
+      "loss": 0.9583,
+      "step": 443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.98081824767192e-06,
+      "loss": 1.1099,
+      "step": 444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.980385059063051e-06,
+      "loss": 1.1703,
+      "step": 445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.979947043101562e-06,
+      "loss": 1.2188,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.979504200212011e-06,
+      "loss": 1.2011,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.979056530823632e-06,
+      "loss": 1.1022,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.978604035370341e-06,
+      "loss": 1.1497,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.978146714290729e-06,
+      "loss": 1.2801,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.977684568028064e-06,
+      "loss": 1.0909,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.977217597030293e-06,
+      "loss": 1.144,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.976745801750038e-06,
+      "loss": 1.1655,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.976269182644596e-06,
+      "loss": 1.2935,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.975787740175944e-06,
+      "loss": 1.2087,
+      "step": 455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.975301474810727e-06,
+      "loss": 1.1307,
+      "step": 456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.974810387020276e-06,
+      "loss": 1.1211,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.97431447728058e-06,
+      "loss": 1.0913,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.973813746072319e-06,
+      "loss": 1.2474,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.97330819388083e-06,
+      "loss": 1.6276,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.972797821196139e-06,
+      "loss": 1.2004,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.972282628512933e-06,
+      "loss": 1.1322,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.971762616330574e-06,
+      "loss": 1.2291,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.971237785153096e-06,
+      "loss": 1.1639,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.970708135489204e-06,
+      "loss": 1.3482,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.970173667852275e-06,
+      "loss": 1.0859,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.96963438276035e-06,
+      "loss": 1.3802,
+      "step": 467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.969090280736145e-06,
+      "loss": 1.1761,
+      "step": 468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.968541362307045e-06,
+      "loss": 1.1965,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.9679876280051e-06,
+      "loss": 1.1966,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.96742907836703e-06,
+      "loss": 1.175,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.966865713934224e-06,
+      "loss": 1.2103,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.1101255416870117,
+      "eval_runtime": 268.0616,
+      "eval_samples_per_second": 3.727,
+      "eval_steps_per_second": 0.466,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.966297535252733e-06,
+      "loss": 1.056,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.96572454287328e-06,
+      "loss": 1.3044,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.965146737351249e-06,
+      "loss": 1.1039,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.964564119246692e-06,
+      "loss": 1.2216,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.963976689124324e-06,
+      "loss": 1.1563,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.963384447553527e-06,
+      "loss": 1.3368,
+      "step": 478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.962787395108344e-06,
+      "loss": 1.27,
+      "step": 479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.96218553236748e-06,
+      "loss": 1.2054,
+      "step": 480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.961578859914307e-06,
+      "loss": 1.2318,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.960967378336854e-06,
+      "loss": 0.975,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.960351088227817e-06,
+      "loss": 0.9899,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.959729990184546e-06,
+      "loss": 1.1608,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.959104084809055e-06,
+      "loss": 1.2532,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.958473372708018e-06,
+      "loss": 1.0833,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.957837854492766e-06,
+      "loss": 1.2435,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.95719753077929e-06,
+      "loss": 1.0572,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.95655240218824e-06,
+      "loss": 1.0929,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.955902469344923e-06,
+      "loss": 1.2556,
+      "step": 490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.955247732879296e-06,
+      "loss": 1.0618,
+      "step": 491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.95458819342598e-06,
+      "loss": 1.167,
+      "step": 492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.953923851624251e-06,
+      "loss": 1.081,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.953254708118035e-06,
+      "loss": 0.9734,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.952580763555916e-06,
+      "loss": 1.1847,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.951902018591129e-06,
+      "loss": 1.2165,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.951218473881564e-06,
+      "loss": 1.2169,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.950530130089762e-06,
+      "loss": 1.212,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.949836987882916e-06,
+      "loss": 1.3074,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.949139047932872e-06,
+      "loss": 1.0401,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.948436310916121e-06,
+      "loss": 1.0383,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.947728777513809e-06,
+      "loss": 1.1037,
+      "step": 502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.947016448411728e-06,
+      "loss": 1.529,
+      "step": 503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.94629932430032e-06,
+      "loss": 0.9924,
+      "step": 504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.945577405874675e-06,
+      "loss": 1.1132,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.944850693834527e-06,
+      "loss": 1.1532,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.94411918888426e-06,
+      "loss": 1.0351,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.9433828917329e-06,
+      "loss": 1.295,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.942641803094122e-06,
+      "loss": 1.2403,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.941895923686243e-06,
+      "loss": 1.1403,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.941145254232223e-06,
+      "loss": 1.2117,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.940389795459665e-06,
+      "loss": 1.293,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.939629548100815e-06,
+      "loss": 1.0857,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.938864512892562e-06,
+      "loss": 1.144,
+      "step": 514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.938094690576432e-06,
+      "loss": 1.1896,
+      "step": 515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.937320081898594e-06,
+      "loss": 1.2558,
+      "step": 516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.936540687609854e-06,
+      "loss": 1.0021,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.935756508465659e-06,
+      "loss": 1.2673,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.934967545226093e-06,
+      "loss": 1.325,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.934173798655877e-06,
+      "loss": 1.2052,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.933375269524366e-06,
+      "loss": 1.1101,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.932571958605555e-06,
+      "loss": 1.2369,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.931763866678068e-06,
+      "loss": 1.1627,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.93095099452517e-06,
+      "loss": 1.1519,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.930133342934757e-06,
+      "loss": 0.973,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.929310912699353e-06,
+      "loss": 1.328,
+      "step": 526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.928483704616119e-06,
+      "loss": 1.3404,
+      "step": 527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.927651719486847e-06,
+      "loss": 1.136,
+      "step": 528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.926814958117955e-06,
+      "loss": 1.0129,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.925973421320496e-06,
+      "loss": 1.0048,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.925127109910145e-06,
+      "loss": 1.2779,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.924276024707213e-06,
+      "loss": 1.2642,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.92342016653663e-06,
+      "loss": 1.209,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.92255953622796e-06,
+      "loss": 1.1691,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.921694134615384e-06,
+      "loss": 1.1225,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.920823962537719e-06,
+      "loss": 1.1844,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.919949020838395e-06,
+      "loss": 1.2377,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.919069310365467e-06,
+      "loss": 1.1053,
+      "step": 538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.91818483197162e-06,
+      "loss": 1.0328,
+      "step": 539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.917295586514151e-06,
+      "loss": 1.2847,
+      "step": 540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.916401574854983e-06,
+      "loss": 1.1179,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.91550279786066e-06,
+      "loss": 1.3723,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.914599256402337e-06,
+      "loss": 1.345,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.913690951355794e-06,
+      "loss": 1.2219,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.91277788360143e-06,
+      "loss": 1.1517,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.911860054024254e-06,
+      "loss": 1.125,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.910937463513893e-06,
+      "loss": 1.3486,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.910010112964589e-06,
+      "loss": 1.3102,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.909078003275198e-06,
+      "loss": 1.081,
+      "step": 549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.908141135349192e-06,
+      "loss": 1.1321,
+      "step": 550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.907199510094647e-06,
+      "loss": 1.0972,
+      "step": 551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906253128424257e-06,
+      "loss": 1.1546,
+      "step": 552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905301991255325e-06,
+      "loss": 0.8754,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.904346099509763e-06,
+      "loss": 1.0774,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.903385454114088e-06,
+      "loss": 1.0208,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90242005599943e-06,
+      "loss": 1.1422,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.901449906101522e-06,
+      "loss": 1.2779,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.900475005360705e-06,
+      "loss": 1.1482,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.899495354721923e-06,
+      "loss": 1.2687,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.898510955134728e-06,
+      "loss": 1.3523,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.897521807553267e-06,
+      "loss": 1.197,
+      "step": 561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.896527912936298e-06,
+      "loss": 1.3878,
+      "step": 562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.895529272247173e-06,
+      "loss": 1.1694,
+      "step": 563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.894525886453852e-06,
+      "loss": 1.0472,
+      "step": 564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.893517756528884e-06,
+      "loss": 1.1021,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.892504883449424e-06,
+      "loss": 1.1179,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.891487268197226e-06,
+      "loss": 1.1236,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.890464911758636e-06,
+      "loss": 1.2332,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.889437815124591e-06,
+      "loss": 1.1649,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.888405979290634e-06,
+      "loss": 1.0852,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.887369405256893e-06,
+      "loss": 1.1948,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.886328094028091e-06,
+      "loss": 1.1706,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.885282046613544e-06,
+      "loss": 1.2262,
+      "step": 573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.884231264027155e-06,
+      "loss": 1.0444,
+      "step": 574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.883175747287424e-06,
+      "loss": 1.1165,
+      "step": 575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.882115497417431e-06,
+      "loss": 1.2076,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.881050515444848e-06,
+      "loss": 1.2477,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.879980802401933e-06,
+      "loss": 1.3205,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.878906359325531e-06,
+      "loss": 1.164,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.877827187257071e-06,
+      "loss": 1.0069,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.876743287242565e-06,
+      "loss": 1.1968,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.875654660332608e-06,
+      "loss": 0.9908,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.874561307582375e-06,
+      "loss": 1.1746,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.873463230051624e-06,
+      "loss": 1.1264,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.872360428804693e-06,
+      "loss": 1.1619,
+      "step": 585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.871252904910497e-06,
+      "loss": 0.8161,
+      "step": 586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.870140659442529e-06,
+      "loss": 1.1386,
+      "step": 587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.869023693478857e-06,
+      "loss": 1.0719,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.867902008102129e-06,
+      "loss": 1.0873,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.866775604399562e-06,
+      "loss": 1.2557,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.865644483462948e-06,
+      "loss": 1.2357,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.864508646388654e-06,
+      "loss": 1.2229,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.863368094277615e-06,
+      "loss": 1.1834,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.862222828235337e-06,
+      "loss": 1.1732,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.861072849371898e-06,
+      "loss": 1.2159,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.859918158801938e-06,
+      "loss": 1.2349,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.858758757644669e-06,
+      "loss": 1.0245,
+      "step": 597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.857594647023866e-06,
+      "loss": 1.0835,
+      "step": 598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.856425828067872e-06,
+      "loss": 1.0209,
+      "step": 599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.85525230190959e-06,
+      "loss": 1.1577,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.854074069686488e-06,
+      "loss": 1.1819,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.852891132540594e-06,
+      "loss": 1.2076,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.851703491618498e-06,
+      "loss": 1.0536,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.850511148071349e-06,
+      "loss": 1.2979,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.84931410305485e-06,
+      "loss": 1.2886,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.848112357729267e-06,
+      "loss": 1.3502,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.846905913259419e-06,
+      "loss": 1.1714,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.84569477081468e-06,
+      "loss": 1.3495,
+      "step": 608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844478931568977e-06,
+      "loss": 1.2452,
+      "step": 609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84325839670079e-06,
+      "loss": 1.3248,
+      "step": 610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842033167393152e-06,
+      "loss": 1.0893,
+      "step": 611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84080324483364e-06,
+      "loss": 1.2009,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839568630214394e-06,
+      "loss": 1.1399,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.838329324732082e-06,
+      "loss": 1.3179,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837085329587936e-06,
+      "loss": 1.1108,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835836645987726e-06,
+      "loss": 1.3108,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834583275141764e-06,
+      "loss": 1.1305,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833325218264913e-06,
+      "loss": 1.1946,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.832062476576569e-06,
+      "loss": 1.0692,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830795051300678e-06,
+      "loss": 1.1116,
+      "step": 620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.829522943665717e-06,
+      "loss": 1.0437,
+      "step": 621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.828246154904709e-06,
+      "loss": 1.1046,
+      "step": 622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.826964686255207e-06,
+      "loss": 1.0374,
+      "step": 623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.825678538959307e-06,
+      "loss": 1.1186,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.824387714263636e-06,
+      "loss": 0.9855,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.823092213419354e-06,
+      "loss": 1.1694,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.82179203768216e-06,
+      "loss": 1.0869,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.820487188312271e-06,
+      "loss": 1.3015,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.81917766657445e-06,
+      "loss": 1.3485,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.817863473737974e-06,
+      "loss": 1.3147,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.816544611076662e-06,
+      "loss": 1.1984,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.815221079868847e-06,
+      "loss": 1.1805,
+      "step": 632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.813892881397392e-06,
+      "loss": 1.2298,
+      "step": 633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.812560016949685e-06,
+      "loss": 1.2681,
+      "step": 634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.811222487817638e-06,
+      "loss": 1.1365,
+      "step": 635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.809880295297677e-06,
+      "loss": 1.1779,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.808533440690757e-06,
+      "loss": 1.1847,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.807181925302347e-06,
+      "loss": 1.052,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.805825750442431e-06,
+      "loss": 1.171,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.804464917425518e-06,
+      "loss": 1.3062,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.803099427570622e-06,
+      "loss": 1.1465,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.801729282201277e-06,
+      "loss": 1.234,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.800354482645529e-06,
+      "loss": 1.0271,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.79897503023593e-06,
+      "loss": 1.181,
+      "step": 644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.797590926309553e-06,
+      "loss": 1.196,
+      "step": 645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.796202172207966e-06,
+      "loss": 1.1196,
+      "step": 646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.794808769277256e-06,
+      "loss": 1.1135,
+      "step": 647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.793410718868007e-06,
+      "loss": 1.1422,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.792008022335312e-06,
+      "loss": 1.1302,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.790600681038768e-06,
+      "loss": 1.377,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.789188696342472e-06,
+      "loss": 1.0816,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.787772069615024e-06,
+      "loss": 1.1914,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.786350802229519e-06,
+      "loss": 1.2652,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.784924895563557e-06,
+      "loss": 1.0834,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.783494350999227e-06,
+      "loss": 1.1317,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.782059169923122e-06,
+      "loss": 1.1723,
+      "step": 656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.780619353726318e-06,
+      "loss": 1.0223,
+      "step": 657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.779174903804394e-06,
+      "loss": 1.0725,
+      "step": 658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.777725821557416e-06,
+      "loss": 1.3925,
+      "step": 659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.77627210838994e-06,
+      "loss": 1.2944,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.77481376571101e-06,
+      "loss": 1.3887,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.773350794934158e-06,
+      "loss": 1.2,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.771883197477403e-06,
+      "loss": 1.0415,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.770410974763249e-06,
+      "loss": 1.3197,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.768934128218678e-06,
+      "loss": 1.1447,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.76745265927516e-06,
+      "loss": 1.0206,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.765966569368642e-06,
+      "loss": 1.0944,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.76447585993955e-06,
+      "loss": 1.0766,
+      "step": 668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.762980532432791e-06,
+      "loss": 1.0156,
+      "step": 669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.761480588297743e-06,
+      "loss": 1.3631,
+      "step": 670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.759976028988263e-06,
+      "loss": 1.1277,
+      "step": 671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.758466855962679e-06,
+      "loss": 1.1512,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.75695307068379e-06,
+      "loss": 1.1474,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.755434674618869e-06,
+      "loss": 0.9991,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.753911669239656e-06,
+      "loss": 1.2495,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.752384056022359e-06,
+      "loss": 1.3236,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.750851836447652e-06,
+      "loss": 0.9799,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.749315012000674e-06,
+      "loss": 1.1405,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.747773584171027e-06,
+      "loss": 1.3172,
+      "step": 679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.746227554452777e-06,
+      "loss": 1.1769,
+      "step": 680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.744676924344447e-06,
+      "loss": 1.1341,
+      "step": 681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.74312169534902e-06,
+      "loss": 1.151,
+      "step": 682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.741561868973942e-06,
+      "loss": 1.1281,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.739997446731104e-06,
+      "loss": 1.2655,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.738428430136864e-06,
+      "loss": 1.2719,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.736854820712023e-06,
+      "loss": 1.2859,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.735276619981841e-06,
+      "loss": 1.1492,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.733693829476025e-06,
+      "loss": 1.0478,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.73210645072873e-06,
+      "loss": 1.0658,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.73051448527856e-06,
+      "loss": 1.1349,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.728917934668562e-06,
+      "loss": 1.1609,
+      "step": 691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727316800446233e-06,
+      "loss": 1.0521,
+      "step": 692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725711084163507e-06,
+      "loss": 1.0287,
+      "step": 693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724100787376761e-06,
+      "loss": 1.1164,
+      "step": 694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.722485911646812e-06,
+      "loss": 1.1328,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720866458538918e-06,
+      "loss": 1.3098,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71924242962277e-06,
+      "loss": 1.118,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717613826472494e-06,
+      "loss": 1.2226,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71598065066665e-06,
+      "loss": 1.3669,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.714342903788233e-06,
+      "loss": 1.0277,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712700587424668e-06,
+      "loss": 0.9084,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.711053703167807e-06,
+      "loss": 1.2625,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709402252613928e-06,
+      "loss": 1.2487,
+      "step": 703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.707746237363742e-06,
+      "loss": 1.1017,
+      "step": 704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.706085659022374e-06,
+      "loss": 1.1921,
+      "step": 705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.704420519199382e-06,
+      "loss": 0.9304,
+      "step": 706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.702750819508742e-06,
+      "loss": 1.0809,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.701076561568845e-06,
+      "loss": 1.0357,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.0738600492477417,
+      "eval_runtime": 268.3516,
+      "eval_samples_per_second": 3.723,
+      "eval_steps_per_second": 0.466,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.699397747002504e-06,
+      "loss": 1.2802,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.697714377436952e-06,
+      "loss": 1.114,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.69602645450383e-06,
+      "loss": 1.1755,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.694333979839199e-06,
+      "loss": 0.9908,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.692636955083527e-06,
+      "loss": 0.9812,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.690935381881697e-06,
+      "loss": 1.0437,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.689229261882993e-06,
+      "loss": 1.1983,
+      "step": 715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.687518596741118e-06,
+      "loss": 0.9838,
+      "step": 716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.685803388114165e-06,
+      "loss": 1.052,
+      "step": 717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.684083637664646e-06,
+      "loss": 1.0882,
+      "step": 718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.682359347059467e-06,
+      "loss": 1.4096,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.680630517969935e-06,
+      "loss": 1.1297,
+      "step": 720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.678897152071757e-06,
+      "loss": 1.1865,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.67715925104504e-06,
+      "loss": 1.1132,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.67541681657428e-06,
+      "loss": 1.1738,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.673669850348379e-06,
+      "loss": 0.9507,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.67191835406062e-06,
+      "loss": 1.2744,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.670162329408678e-06,
+      "loss": 1.1879,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.668401778094625e-06,
+      "loss": 1.1947,
+      "step": 727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.666636701824913e-06,
+      "loss": 1.2646,
+      "step": 728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.664867102310385e-06,
+      "loss": 1.1888,
+      "step": 729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.663092981266262e-06,
+      "loss": 1.2276,
+      "step": 730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.661314340412157e-06,
+      "loss": 1.1809,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.659531181472055e-06,
+      "loss": 1.1176,
+      "step": 732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.657743506174324e-06,
+      "loss": 1.0581,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.65595131625171e-06,
+      "loss": 1.3788,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.654154613441336e-06,
+      "loss": 1.1932,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.652353399484693e-06,
+      "loss": 1.2889,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.650547676127654e-06,
+      "loss": 1.3505,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.648737445120453e-06,
+      "loss": 1.1538,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.6469227082177e-06,
+      "loss": 1.1364,
+      "step": 739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.645103467178374e-06,
+      "loss": 1.1804,
+      "step": 740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.64327972376581e-06,
+      "loss": 1.1698,
+      "step": 741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.641451479747718e-06,
+      "loss": 1.0781,
+      "step": 742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.639618736896163e-06,
+      "loss": 1.2016,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.637781496987575e-06,
+      "loss": 1.1431,
+      "step": 744
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.635939761802739e-06,
+      "loss": 1.029,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.6340935331268e-06,
+      "loss": 1.1025,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.632242812749257e-06,
+      "loss": 1.2688,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.630387602463964e-06,
+      "loss": 1.0155,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.628527904069125e-06,
+      "loss": 1.1089,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.626663719367299e-06,
+      "loss": 1.0812,
+      "step": 750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.624795050165385e-06,
+      "loss": 0.9655,
+      "step": 751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.622921898274635e-06,
+      "loss": 1.2333,
+      "step": 752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.621044265510648e-06,
+      "loss": 1.2071,
+      "step": 753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.619162153693358e-06,
+      "loss": 0.924,
+      "step": 754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.617275564647047e-06,
+      "loss": 1.1042,
+      "step": 755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.615384500200336e-06,
+      "loss": 1.0014,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.61348896218618e-06,
+      "loss": 1.2376,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.611588952441874e-06,
+      "loss": 1.4647,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.609684472809046e-06,
+      "loss": 1.0991,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.607775525133657e-06,
+      "loss": 1.1941,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.605862111266e-06,
+      "loss": 1.1336,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.603944233060694e-06,
+      "loss": 1.4626,
+      "step": 762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.602021892376684e-06,
+      "loss": 1.2229,
+      "step": 763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.600095091077246e-06,
+      "loss": 1.2506,
+      "step": 764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.598163831029974e-06,
+      "loss": 1.2708,
+      "step": 765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.59622811410679e-06,
+      "loss": 1.0561,
+      "step": 766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.59428794218393e-06,
+      "loss": 1.0878,
+      "step": 767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.592343317141945e-06,
+      "loss": 1.1305,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.590394240865717e-06,
+      "loss": 1.1386,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.588440715244427e-06,
+      "loss": 1.05,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.586482742171576e-06,
+      "loss": 1.1249,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.584520323544974e-06,
+      "loss": 1.1213,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.58255346126674e-06,
+      "loss": 1.1652,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.580582157243302e-06,
+      "loss": 1.1854,
+      "step": 774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.57860641338539e-06,
+      "loss": 0.8986,
+      "step": 775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.57662623160804e-06,
+      "loss": 1.2885,
+      "step": 776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.57464161383059e-06,
+      "loss": 1.0786,
+      "step": 777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.572652561976673e-06,
+      "loss": 1.1266,
+      "step": 778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.570659077974227e-06,
+      "loss": 1.0,
+      "step": 779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.568661163755479e-06,
+      "loss": 1.2158,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.566658821256953e-06,
+      "loss": 1.1925,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.56465205241947e-06,
+      "loss": 1.1023,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.562640859188133e-06,
+      "loss": 1.171,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.560625243512336e-06,
+      "loss": 0.983,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.558605207345765e-06,
+      "loss": 0.977,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.556580752646383e-06,
+      "loss": 1.2945,
+      "step": 786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.554551881376441e-06,
+      "loss": 1.0642,
+      "step": 787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.55251859550247e-06,
+      "loss": 1.3097,
+      "step": 788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.550480896995275e-06,
+      "loss": 1.0686,
+      "step": 789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.548438787829946e-06,
+      "loss": 1.3291,
+      "step": 790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.546392269985842e-06,
+      "loss": 1.1261,
+      "step": 791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.544341345446599e-06,
+      "loss": 1.132,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.542286016200122e-06,
+      "loss": 1.3352,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.540226284238585e-06,
+      "loss": 1.2645,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.538162151558434e-06,
+      "loss": 1.0958,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.536093620160372e-06,
+      "loss": 1.1581,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.534020692049374e-06,
+      "loss": 1.2218,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.531943369234674e-06,
+      "loss": 1.1858,
+      "step": 798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.529861653729763e-06,
+      "loss": 1.3329,
+      "step": 799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.52777554755239e-06,
+      "loss": 1.2955,
+      "step": 800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.525685052724565e-06,
+      "loss": 1.2115,
+      "step": 801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.523590171272546e-06,
+      "loss": 1.0408,
+      "step": 802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.521490905226844e-06,
+      "loss": 1.2728,
+      "step": 803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.519387256622223e-06,
+      "loss": 0.8925,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.517279227497692e-06,
+      "loss": 1.021,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.515166819896506e-06,
+      "loss": 1.0705,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.513050035866165e-06,
+      "loss": 1.0632,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.510928877458412e-06,
+      "loss": 1.1912,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.508803346729226e-06,
+      "loss": 1.1943,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.506673445738828e-06,
+      "loss": 1.0524,
+      "step": 810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.504539176551672e-06,
+      "loss": 1.1134,
+      "step": 811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.502400541236451e-06,
+      "loss": 1.0985,
+      "step": 812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.500257541866082e-06,
+      "loss": 1.1162,
+      "step": 813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.498110180517719e-06,
+      "loss": 1.3487,
+      "step": 814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.49595845927274e-06,
+      "loss": 1.0569,
+      "step": 815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.49380238021675e-06,
+      "loss": 1.2644,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.49164194543958e-06,
+      "loss": 1.0979,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.48947715703528e-06,
+      "loss": 1.231,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.487308017102122e-06,
+      "loss": 1.1796,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.485134527742592e-06,
+      "loss": 1.2198,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.482956691063396e-06,
+      "loss": 1.1496,
+      "step": 821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.480774509175453e-06,
+      "loss": 1.4714,
+      "step": 822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.478587984193892e-06,
+      "loss": 1.1229,
+      "step": 823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.476397118238054e-06,
+      "loss": 1.1059,
+      "step": 824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.474201913431483e-06,
+      "loss": 1.2096,
+      "step": 825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.472002371901935e-06,
+      "loss": 1.1935,
+      "step": 826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.469798495781364e-06,
+      "loss": 1.0894,
+      "step": 827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.46759028720593e-06,
+      "loss": 1.2556,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.465377748315987e-06,
+      "loss": 1.2763,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.463160881256089e-06,
+      "loss": 1.0416,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.460939688174989e-06,
+      "loss": 1.0384,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.458714171225626e-06,
+      "loss": 1.1414,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.456484332565135e-06,
+      "loss": 1.3063,
+      "step": 833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.454250174354838e-06,
+      "loss": 1.0993,
+      "step": 834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.452011698760242e-06,
+      "loss": 0.9598,
+      "step": 835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.449768907951045e-06,
+      "loss": 1.186,
+      "step": 836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.447521804101121e-06,
+      "loss": 1.2067,
+      "step": 837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.445270389388527e-06,
+      "loss": 1.2154,
+      "step": 838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.4430146659955e-06,
+      "loss": 1.0513,
+      "step": 839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.44075463610845e-06,
+      "loss": 1.1081,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.438490301917964e-06,
+      "loss": 1.0729,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.436221665618799e-06,
+      "loss": 1.1742,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.433948729409882e-06,
+      "loss": 1.0925,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.431671495494312e-06,
+      "loss": 1.1372,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.429389966079347e-06,
+      "loss": 1.027,
+      "step": 845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.427104143376414e-06,
+      "loss": 1.0007,
+      "step": 846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.424814029601096e-06,
+      "loss": 1.1123,
+      "step": 847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.42251962697314e-06,
+      "loss": 1.1191,
+      "step": 848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.420220937716452e-06,
+      "loss": 1.063,
+      "step": 849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.417917964059085e-06,
+      "loss": 1.2922,
+      "step": 850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.415610708233252e-06,
+      "loss": 1.003,
+      "step": 851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.41329917247531e-06,
+      "loss": 1.1363,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.41098335902577e-06,
+      "loss": 1.2522,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.408663270129289e-06,
+      "loss": 1.2031,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.406338908034665e-06,
+      "loss": 0.9692,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.40401027499484e-06,
+      "loss": 1.101,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.401677373266892e-06,
+      "loss": 1.3211,
+      "step": 857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.399340205112045e-06,
+      "loss": 1.0759,
+      "step": 858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.396998772795647e-06,
+      "loss": 1.1563,
+      "step": 859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.39465307858719e-06,
+      "loss": 1.1472,
+      "step": 860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.39230312476029e-06,
+      "loss": 1.0359,
+      "step": 861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.389948913592694e-06,
+      "loss": 1.3914,
+      "step": 862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.387590447366274e-06,
+      "loss": 1.1398,
+      "step": 863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.38522772836703e-06,
+      "loss": 1.1828,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.38286075888508e-06,
+      "loss": 1.2239,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.380489541214664e-06,
+      "loss": 1.0,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.378114077654138e-06,
+      "loss": 1.0441,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.375734370505976e-06,
+      "loss": 1.149,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.373350422076764e-06,
+      "loss": 1.1879,
+      "step": 869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.370962234677198e-06,
+      "loss": 1.167,
+      "step": 870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.368569810622083e-06,
+      "loss": 1.4329,
+      "step": 871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.366173152230333e-06,
+      "loss": 1.1271,
+      "step": 872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.363772261824962e-06,
+      "loss": 1.1377,
+      "step": 873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.36136714173309e-06,
+      "loss": 1.3837,
+      "step": 874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.358957794285933e-06,
+      "loss": 1.0266,
+      "step": 875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.356544221818806e-06,
+      "loss": 1.187,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.354126426671124e-06,
+      "loss": 1.1631,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.351704411186384e-06,
+      "loss": 1.216,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.349278177712187e-06,
+      "loss": 1.0426,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.34684772860021e-06,
+      "loss": 1.1462,
+      "step": 880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.344413066206223e-06,
+      "loss": 1.216,
+      "step": 881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.34197419289008e-06,
+      "loss": 0.9109,
+      "step": 882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.339531111015713e-06,
+      "loss": 1.0075,
+      "step": 883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.337083822951139e-06,
+      "loss": 1.1929,
+      "step": 884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.334632331068445e-06,
+      "loss": 1.1048,
+      "step": 885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.332176637743796e-06,
+      "loss": 1.3165,
+      "step": 886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.329716745357427e-06,
+      "loss": 0.9306,
+      "step": 887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.32725265629365e-06,
+      "loss": 1.1577,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.324784372940836e-06,
+      "loss": 1.271,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.322311897691427e-06,
+      "loss": 0.9726,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.319835232941929e-06,
+      "loss": 1.149,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.317354381092902e-06,
+      "loss": 1.315,
+      "step": 892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.314869344548969e-06,
+      "loss": 1.0514,
+      "step": 893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.312380125718811e-06,
+      "loss": 1.2799,
+      "step": 894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.30988672701516e-06,
+      "loss": 0.9576,
+      "step": 895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.3073891508548e-06,
+      "loss": 1.3859,
+      "step": 896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.304887399658562e-06,
+      "loss": 1.2188,
+      "step": 897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.302381475851328e-06,
+      "loss": 1.0086,
+      "step": 898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.29987138186202e-06,
+      "loss": 1.2477,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.297357120123606e-06,
+      "loss": 1.0726,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.294838693073093e-06,
+      "loss": 1.1705,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.292316103151524e-06,
+      "loss": 1.1084,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.289789352803975e-06,
+      "loss": 1.1051,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.287258444479558e-06,
+      "loss": 1.105,
+      "step": 904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.284723380631415e-06,
+      "loss": 1.2598,
+      "step": 905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.282184163716712e-06,
+      "loss": 0.9921,
+      "step": 906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.279640796196647e-06,
+      "loss": 1.0443,
+      "step": 907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.277093280536435e-06,
+      "loss": 1.293,
+      "step": 908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.274541619205314e-06,
+      "loss": 1.1358,
+      "step": 909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.271985814676542e-06,
+      "loss": 1.1665,
+      "step": 910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.26942586942739e-06,
+      "loss": 1.0294,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.266861785939145e-06,
+      "loss": 0.965,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.264293566697103e-06,
+      "loss": 1.0344,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.261721214190567e-06,
+      "loss": 1.2077,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.259144730912853e-06,
+      "loss": 0.9496,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.256564119361276e-06,
+      "loss": 1.1303,
+      "step": 916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.253979382037149e-06,
+      "loss": 1.2029,
+      "step": 917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.251390521445792e-06,
+      "loss": 1.1028,
+      "step": 918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.248797540096515e-06,
+      "loss": 1.4242,
+      "step": 919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.246200440502628e-06,
+      "loss": 1.2078,
+      "step": 920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.243599225181428e-06,
+      "loss": 1.1104,
+      "step": 921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.2409938966542e-06,
+      "loss": 1.1803,
+      "step": 922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.238384457446222e-06,
+      "loss": 1.1503,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.235770910086753e-06,
+      "loss": 1.0537,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.23315325710903e-06,
+      "loss": 1.2245,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.230531501050277e-06,
+      "loss": 1.0824,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.22790564445169e-06,
+      "loss": 1.1142,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.22527568985844e-06,
+      "loss": 0.9585,
+      "step": 928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.222641639819672e-06,
+      "loss": 1.1498,
+      "step": 929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.220003496888502e-06,
+      "loss": 1.1592,
+      "step": 930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.217361263622004e-06,
+      "loss": 1.0368,
+      "step": 931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.214714942581228e-06,
+      "loss": 1.0401,
+      "step": 932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.212064536331182e-06,
+      "loss": 0.9658,
+      "step": 933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.20941004744083e-06,
+      "loss": 1.1442,
+      "step": 934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.2067514784831e-06,
+      "loss": 1.2851,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.204088832034868e-06,
+      "loss": 0.9718,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.201422110676967e-06,
+      "loss": 1.1958,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.198751316994178e-06,
+      "loss": 0.9945,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.196076453575226e-06,
+      "loss": 1.2643,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.19339752301279e-06,
+      "loss": 1.1821,
+      "step": 940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.190714527903477e-06,
+      "loss": 1.1818,
+      "step": 941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.18802747084785e-06,
+      "loss": 1.0277,
+      "step": 942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.185336354450393e-06,
+      "loss": 1.1213,
+      "step": 943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.182641181319538e-06,
+      "loss": 1.27,
+      "step": 944
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.0539518594741821,
+      "eval_runtime": 268.6045,
+      "eval_samples_per_second": 3.719,
+      "eval_steps_per_second": 0.465,
+      "step": 944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.179941954067641e-06,
+      "loss": 1.1262,
+      "step": 945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.177238675310991e-06,
+      "loss": 1.0239,
+      "step": 946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.174531347669801e-06,
+      "loss": 1.0628,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.171819973768213e-06,
+      "loss": 1.2047,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.169104556234288e-06,
+      "loss": 1.2009,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.166385097700005e-06,
+      "loss": 1.0352,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.163661600801264e-06,
+      "loss": 1.1939,
+      "step": 951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.160934068177875e-06,
+      "loss": 0.9893,
+      "step": 952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.158202502473563e-06,
+      "loss": 1.1192,
+      "step": 953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.155466906335961e-06,
+      "loss": 1.1379,
+      "step": 954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.152727282416607e-06,
+      "loss": 1.0479,
+      "step": 955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.149983633370945e-06,
+      "loss": 1.1274,
+      "step": 956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.147235961858321e-06,
+      "loss": 1.1024,
+      "step": 957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.144484270541976e-06,
+      "loss": 0.9003,
+      "step": 958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.141728562089054e-06,
+      "loss": 1.122,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.138968839170585e-06,
+      "loss": 1.0116,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.136205104461495e-06,
+      "loss": 1.4711,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.133437360640597e-06,
+      "loss": 1.2094,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.130665610390593e-06,
+      "loss": 1.163,
+      "step": 963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.127889856398063e-06,
+      "loss": 1.205,
+      "step": 964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.12511010135347e-06,
+      "loss": 1.0823,
+      "step": 965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.122326347951155e-06,
+      "loss": 1.0909,
+      "step": 966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.11953859888934e-06,
+      "loss": 0.9585,
+      "step": 967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.116746856870106e-06,
+      "loss": 1.2888,
+      "step": 968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.113951124599423e-06,
+      "loss": 1.1315,
+      "step": 969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.111151404787111e-06,
+      "loss": 1.2,
+      "step": 970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.108347700146866e-06,
+      "loss": 1.0591,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.105540013396243e-06,
+      "loss": 1.2939,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.102728347256656e-06,
+      "loss": 1.1627,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.099912704453377e-06,
+      "loss": 1.1061,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.097093087715534e-06,
+      "loss": 1.2272,
+      "step": 975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.094269499776106e-06,
+      "loss": 1.0806,
+      "step": 976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.091441943371914e-06,
+      "loss": 1.4281,
+      "step": 977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.08861042124364e-06,
+      "loss": 1.0855,
+      "step": 978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.085774936135798e-06,
+      "loss": 1.0476,
+      "step": 979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.08293549079675e-06,
+      "loss": 1.0993,
+      "step": 980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.080092087978688e-06,
+      "loss": 1.1712,
+      "step": 981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.07724473043765e-06,
+      "loss": 1.0634,
+      "step": 982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.0743934209335e-06,
+      "loss": 1.1836,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.071538162229939e-06,
+      "loss": 1.0579,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.068678957094487e-06,
+      "loss": 1.1601,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.0658158082985e-06,
+      "loss": 1.1144,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.062948718617147e-06,
+      "loss": 1.0275,
+      "step": 987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.060077690829421e-06,
+      "loss": 1.118,
+      "step": 988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.057202727718134e-06,
+      "loss": 1.1003,
+      "step": 989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.054323832069904e-06,
+      "loss": 1.3383,
+      "step": 990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.051441006675172e-06,
+      "loss": 1.1776,
+      "step": 991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.048554254328184e-06,
+      "loss": 1.0921,
+      "step": 992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.045663577826987e-06,
+      "loss": 1.1837,
+      "step": 993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.042768979973435e-06,
+      "loss": 0.9844,
+      "step": 994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.039870463573188e-06,
+      "loss": 0.9682,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.036968031435697e-06,
+      "loss": 0.9711,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.03406168637421e-06,
+      "loss": 1.1953,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.031151431205771e-06,
+      "loss": 1.0843,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.02823726875121e-06,
+      "loss": 1.0642,
+      "step": 999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.025319201835146e-06,
+      "loss": 1.0711,
+      "step": 1000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.022397233285983e-06,
+      "loss": 1.1597,
+      "step": 1001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.019471365935907e-06,
+      "loss": 1.2079,
+      "step": 1002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.01654160262088e-06,
+      "loss": 0.9484,
+      "step": 1003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.013607946180644e-06,
+      "loss": 0.9462,
+      "step": 1004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.010670399458711e-06,
+      "loss": 1.1408,
+      "step": 1005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.007728965302369e-06,
+      "loss": 1.0795,
+      "step": 1006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.004783646562669e-06,
+      "loss": 1.0688,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.001834446094427e-06,
+      "loss": 1.0623,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.998881366756224e-06,
+      "loss": 1.3446,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.995924411410405e-06,
+      "loss": 1.0849,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.992963582923057e-06,
+      "loss": 1.1452,
+      "step": 1011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.989998884164039e-06,
+      "loss": 1.1624,
+      "step": 1012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.987030318006952e-06,
+      "loss": 1.0914,
+      "step": 1013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.984057887329145e-06,
+      "loss": 1.1951,
+      "step": 1014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.981081595011717e-06,
+      "loss": 1.2496,
+      "step": 1015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.978101443939506e-06,
+      "loss": 1.0267,
+      "step": 1016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.97511743700109e-06,
+      "loss": 1.1496,
+      "step": 1017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.972129577088791e-06,
+      "loss": 1.1033,
+      "step": 1018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.969137867098658e-06,
+      "loss": 1.3013,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.966142309930474e-06,
+      "loss": 1.0801,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.963142908487755e-06,
+      "loss": 1.1514,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.960139665677734e-06,
+      "loss": 1.116,
+      "step": 1022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.957132584411378e-06,
+      "loss": 1.1385,
+      "step": 1023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.954121667603368e-06,
+      "loss": 1.0869,
+      "step": 1024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.951106918172105e-06,
+      "loss": 1.1665,
+      "step": 1025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.948088339039703e-06,
+      "loss": 1.0617,
+      "step": 1026
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.94506593313199e-06,
+      "loss": 0.9255,
+      "step": 1027
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.942039703378502e-06,
+      "loss": 1.2122,
+      "step": 1028
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.939009652712481e-06,
+      "loss": 1.1291,
+      "step": 1029
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.935975784070875e-06,
+      "loss": 1.2508,
+      "step": 1030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.932938100394329e-06,
+      "loss": 1.1186,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.929896604627186e-06,
+      "loss": 1.1314,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926851299717492e-06,
+      "loss": 1.0752,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.923802188616973e-06,
+      "loss": 1.2764,
+      "step": 1034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.92074927428105e-06,
+      "loss": 1.2321,
+      "step": 1035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.917692559668835e-06,
+      "loss": 0.9983,
+      "step": 1036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.914632047743114e-06,
+      "loss": 1.0619,
+      "step": 1037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.911567741470361e-06,
+      "loss": 1.0057,
+      "step": 1038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.908499643820726e-06,
+      "loss": 1.2063,
+      "step": 1039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.905427757768031e-06,
+      "loss": 1.157,
+      "step": 1040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.902352086289774e-06,
+      "loss": 0.9421,
+      "step": 1041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.899272632367117e-06,
+      "loss": 1.2596,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.896189398984895e-06,
+      "loss": 1.2484,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.8931023891316e-06,
+      "loss": 1.2569,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.89001160579939e-06,
+      "loss": 1.0117,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.886917051984077e-06,
+      "loss": 1.0436,
+      "step": 1046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.883818730685126e-06,
+      "loss": 1.2294,
+      "step": 1047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.880716644905658e-06,
+      "loss": 1.2142,
+      "step": 1048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.877610797652441e-06,
+      "loss": 0.9556,
+      "step": 1049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.87450119193589e-06,
+      "loss": 1.1449,
+      "step": 1050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.871387830770058e-06,
+      "loss": 1.1524,
+      "step": 1051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.868270717172645e-06,
+      "loss": 1.0689,
+      "step": 1052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.865149854164983e-06,
+      "loss": 0.964,
+      "step": 1053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.862025244772043e-06,
+      "loss": 1.3596,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.858896892022421e-06,
+      "loss": 0.9729,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.855764798948348e-06,
+      "loss": 1.112,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.852628968585674e-06,
+      "loss": 1.2246,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.849489403973879e-06,
+      "loss": 1.0653,
+      "step": 1058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.846346108156053e-06,
+      "loss": 1.1116,
+      "step": 1059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.84319908417891e-06,
+      "loss": 1.0794,
+      "step": 1060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.840048335092774e-06,
+      "loss": 1.0809,
+      "step": 1061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.836893863951584e-06,
+      "loss": 1.0488,
+      "step": 1062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.833735673812882e-06,
+      "loss": 1.0564,
+      "step": 1063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.830573767737815e-06,
+      "loss": 1.1996,
+      "step": 1064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.827408148791134e-06,
+      "loss": 1.0043,
+      "step": 1065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.824238820041189e-06,
+      "loss": 1.0391,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.821065784559924e-06,
+      "loss": 1.0803,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.817889045422876e-06,
+      "loss": 1.182,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.814708605709172e-06,
+      "loss": 1.0138,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.811524468501527e-06,
+      "loss": 1.1156,
+      "step": 1070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.808336636886238e-06,
+      "loss": 1.1226,
+      "step": 1071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.805145113953188e-06,
+      "loss": 1.0384,
+      "step": 1072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.80194990279583e-06,
+      "loss": 1.309,
+      "step": 1073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.798751006511195e-06,
+      "loss": 1.0248,
+      "step": 1074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.795548428199889e-06,
+      "loss": 1.2012,
+      "step": 1075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.792342170966086e-06,
+      "loss": 0.9943,
+      "step": 1076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.78913223791752e-06,
+      "loss": 1.2125,
+      "step": 1077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.785918632165497e-06,
+      "loss": 1.1148,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.782701356824875e-06,
+      "loss": 1.1898,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.779480415014074e-06,
+      "loss": 1.0791,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.776255809855065e-06,
+      "loss": 0.9168,
+      "step": 1081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.773027544473369e-06,
+      "loss": 1.1382,
+      "step": 1082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.769795621998058e-06,
+      "loss": 1.3949,
+      "step": 1083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.766560045561749e-06,
+      "loss": 1.0926,
+      "step": 1084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.763320818300596e-06,
+      "loss": 1.0953,
+      "step": 1085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.760077943354295e-06,
+      "loss": 1.2448,
+      "step": 1086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.756831423866077e-06,
+      "loss": 1.1668,
+      "step": 1087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.753581262982707e-06,
+      "loss": 0.9305,
+      "step": 1088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.750327463854475e-06,
+      "loss": 1.2713,
+      "step": 1089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.747070029635204e-06,
+      "loss": 1.0204,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.743808963482235e-06,
+      "loss": 1.1212,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.740544268556432e-06,
+      "loss": 1.0726,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.737275948022175e-06,
+      "loss": 1.1245,
+      "step": 1093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.734004005047358e-06,
+      "loss": 1.2051,
+      "step": 1094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.73072844280339e-06,
+      "loss": 0.9471,
+      "step": 1095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.727449264465182e-06,
+      "loss": 1.1714,
+      "step": 1096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.724166473211154e-06,
+      "loss": 0.8268,
+      "step": 1097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.720880072223225e-06,
+      "loss": 0.9204,
+      "step": 1098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.717590064686819e-06,
+      "loss": 1.1903,
+      "step": 1099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.714296453790845e-06,
+      "loss": 1.1292,
+      "step": 1100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.710999242727719e-06,
+      "loss": 1.0532,
+      "step": 1101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.70769843469333e-06,
+      "loss": 1.1348,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.704394032887068e-06,
+      "loss": 1.1642,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.7010860405118e-06,
+      "loss": 1.1413,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.69777446077387e-06,
+      "loss": 1.0836,
+      "step": 1105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.694459296883107e-06,
+      "loss": 1.208,
+      "step": 1106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.691140552052807e-06,
+      "loss": 0.9945,
+      "step": 1107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.68781822949974e-06,
+      "loss": 1.2102,
+      "step": 1108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.684492332444143e-06,
+      "loss": 1.196,
+      "step": 1109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.68116286410972e-06,
+      "loss": 0.9378,
+      "step": 1110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.677829827723633e-06,
+      "loss": 1.1607,
+      "step": 1111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.674493226516506e-06,
+      "loss": 1.2279,
+      "step": 1112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.671153063722413e-06,
+      "loss": 1.1437,
+      "step": 1113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.667809342578886e-06,
+      "loss": 1.3294,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.664462066326902e-06,
+      "loss": 1.0363,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.661111238210889e-06,
+      "loss": 1.1603,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.65775686147871e-06,
+      "loss": 1.0391,
+      "step": 1117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.654398939381672e-06,
+      "loss": 1.0092,
+      "step": 1118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.651037475174521e-06,
+      "loss": 1.1966,
+      "step": 1119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.647672472115433e-06,
+      "loss": 1.2529,
+      "step": 1120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.644303933466011e-06,
+      "loss": 1.0756,
+      "step": 1121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.640931862491294e-06,
+      "loss": 1.2182,
+      "step": 1122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.637556262459731e-06,
+      "loss": 1.2276,
+      "step": 1123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.634177136643208e-06,
+      "loss": 1.0768,
+      "step": 1124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.630794488317017e-06,
+      "loss": 1.1027,
+      "step": 1125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.627408320759868e-06,
+      "loss": 1.1422,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.62401863725388e-06,
+      "loss": 0.8542,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.620625441084584e-06,
+      "loss": 1.1349,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.617228735540912e-06,
+      "loss": 1.1749,
+      "step": 1129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.6138285239152e-06,
+      "loss": 1.1928,
+      "step": 1130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.61042480950318e-06,
+      "loss": 1.0909,
+      "step": 1131
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.607017595603979e-06,
+      "loss": 1.0666,
+      "step": 1132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.603606885520122e-06,
+      "loss": 1.0361,
+      "step": 1133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.600192682557515e-06,
+      "loss": 1.0654,
+      "step": 1134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.596774990025452e-06,
+      "loss": 1.0341,
+      "step": 1135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.593353811236611e-06,
+      "loss": 0.9078,
+      "step": 1136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.589929149507048e-06,
+      "loss": 1.0794,
+      "step": 1137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.586501008156195e-06,
+      "loss": 1.067,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.583069390506857e-06,
+      "loss": 0.914,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.579634299885208e-06,
+      "loss": 1.0398,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.576195739620788e-06,
+      "loss": 1.0567,
+      "step": 1141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.572753713046503e-06,
+      "loss": 1.0994,
+      "step": 1142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.569308223498613e-06,
+      "loss": 0.9963,
+      "step": 1143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.56585927431674e-06,
+      "loss": 1.2382,
+      "step": 1144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.562406868843855e-06,
+      "loss": 1.0334,
+      "step": 1145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.558951010426284e-06,
+      "loss": 1.0346,
+      "step": 1146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.555491702413695e-06,
+      "loss": 1.0372,
+      "step": 1147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.552028948159103e-06,
+      "loss": 1.003,
+      "step": 1148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.548562751018863e-06,
+      "loss": 1.1725,
+      "step": 1149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.545093114352665e-06,
+      "loss": 1.2225,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.541620041523533e-06,
+      "loss": 1.0574,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.538143535897825e-06,
+      "loss": 1.1119,
+      "step": 1152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.534663600845222e-06,
+      "loss": 1.1181,
+      "step": 1153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.53118023973873e-06,
+      "loss": 1.1781,
+      "step": 1154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.52769345595468e-06,
+      "loss": 1.1527,
+      "step": 1155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.524203252872715e-06,
+      "loss": 0.9467,
+      "step": 1156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.520709633875796e-06,
+      "loss": 1.248,
+      "step": 1157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.51721260235019e-06,
+      "loss": 1.1148,
+      "step": 1158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.513712161685482e-06,
+      "loss": 1.0607,
+      "step": 1159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.510208315274545e-06,
+      "loss": 1.1193,
+      "step": 1160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.506701066513574e-06,
+      "loss": 1.1711,
+      "step": 1161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.50319041880204e-06,
+      "loss": 0.9829,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.499676375542725e-06,
+      "loss": 1.1226,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.496158940141694e-06,
+      "loss": 1.2635,
+      "step": 1164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.492638116008302e-06,
+      "loss": 1.0496,
+      "step": 1165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.48911390655519e-06,
+      "loss": 0.9152,
+      "step": 1166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.485586315198277e-06,
+      "loss": 0.8973,
+      "step": 1167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.482055345356764e-06,
+      "loss": 1.2606,
+      "step": 1168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.478521000453123e-06,
+      "loss": 1.0187,
+      "step": 1169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.4749832839131e-06,
+      "loss": 1.092,
+      "step": 1170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.471442199165708e-06,
+      "loss": 1.0946,
+      "step": 1171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.467897749643224e-06,
+      "loss": 0.9934,
+      "step": 1172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.464349938781186e-06,
+      "loss": 1.0903,
+      "step": 1173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.460798770018394e-06,
+      "loss": 1.2764,
+      "step": 1174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.457244246796898e-06,
+      "loss": 1.0176,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.453686372562004e-06,
+      "loss": 1.142,
+      "step": 1176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.45012515076226e-06,
+      "loss": 1.1121,
+      "step": 1177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.446560584849462e-06,
+      "loss": 1.1098,
+      "step": 1178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.44299267827865e-06,
+      "loss": 1.2095,
+      "step": 1179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.439421434508097e-06,
+      "loss": 1.3557,
+      "step": 1180
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.0330462455749512,
+      "eval_runtime": 268.4891,
+      "eval_samples_per_second": 3.721,
+      "eval_steps_per_second": 0.466,
+      "step": 1180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.435846856999315e-06,
+      "loss": 1.0335,
+      "step": 1181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.432268949217043e-06,
+      "loss": 1.1255,
+      "step": 1182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.428687714629254e-06,
+      "loss": 0.9004,
+      "step": 1183
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.425103156707138e-06,
+      "loss": 0.7679,
+      "step": 1184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.421515278925112e-06,
+      "loss": 0.6991,
+      "step": 1185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.417924084760808e-06,
+      "loss": 0.7962,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.414329577695075e-06,
+      "loss": 0.8281,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.41073176121197e-06,
+      "loss": 0.6528,
+      "step": 1188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.407130638798762e-06,
+      "loss": 0.7819,
+      "step": 1189
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.403526213945916e-06,
+      "loss": 0.8075,
+      "step": 1190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.399918490147109e-06,
+      "loss": 0.8032,
+      "step": 1191
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.396307470899209e-06,
+      "loss": 0.5828,
+      "step": 1192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.392693159702278e-06,
+      "loss": 0.8597,
+      "step": 1193
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.389075560059571e-06,
+      "loss": 0.8054,
+      "step": 1194
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.385454675477529e-06,
+      "loss": 0.799,
+      "step": 1195
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.381830509465778e-06,
+      "loss": 0.8236,
+      "step": 1196
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.378203065537123e-06,
+      "loss": 0.84,
+      "step": 1197
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.374572347207551e-06,
+      "loss": 0.8597,
+      "step": 1198
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.370938357996214e-06,
+      "loss": 0.8883,
+      "step": 1199
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.367301101425443e-06,
+      "loss": 0.9171,
+      "step": 1200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.363660581020732e-06,
+      "loss": 0.8965,
+      "step": 1201
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.360016800310737e-06,
+      "loss": 0.9698,
+      "step": 1202
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.35636976282728e-06,
+      "loss": 0.6694,
+      "step": 1203
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.35271947210533e-06,
+      "loss": 0.828,
+      "step": 1204
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.349065931683023e-06,
+      "loss": 0.8741,
+      "step": 1205
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.345409145101629e-06,
+      "loss": 0.6502,
+      "step": 1206
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.341749115905579e-06,
+      "loss": 0.8031,
+      "step": 1207
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.338085847642435e-06,
+      "loss": 0.6399,
+      "step": 1208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.334419343862908e-06,
+      "loss": 0.7349,
+      "step": 1209
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.330749608120839e-06,
+      "loss": 0.7399,
+      "step": 1210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.327076643973205e-06,
+      "loss": 0.8795,
+      "step": 1211
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.323400454980108e-06,
+      "loss": 0.9148,
+      "step": 1212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.319721044704781e-06,
+      "loss": 0.6226,
+      "step": 1213
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.316038416713576e-06,
+      "loss": 0.795,
+      "step": 1214
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.312352574575966e-06,
+      "loss": 0.7841,
+      "step": 1215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.308663521864537e-06,
+      "loss": 0.6682,
+      "step": 1216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.304971262154988e-06,
+      "loss": 0.7727,
+      "step": 1217
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.301275799026128e-06,
+      "loss": 0.7568,
+      "step": 1218
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.297577136059868e-06,
+      "loss": 0.792,
+      "step": 1219
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.293875276841223e-06,
+      "loss": 0.7481,
+      "step": 1220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.290170224958305e-06,
+      "loss": 0.9504,
+      "step": 1221
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.286461984002318e-06,
+      "loss": 0.9522,
+      "step": 1222
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.282750557567565e-06,
+      "loss": 0.8079,
+      "step": 1223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.279035949251425e-06,
+      "loss": 0.9811,
+      "step": 1224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.275318162654373e-06,
+      "loss": 0.8199,
+      "step": 1225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.271597201379958e-06,
+      "loss": 0.8016,
+      "step": 1226
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.267873069034803e-06,
+      "loss": 0.8766,
+      "step": 1227
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.264145769228613e-06,
+      "loss": 1.0084,
+      "step": 1228
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.260415305574158e-06,
+      "loss": 0.7803,
+      "step": 1229
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.256681681687276e-06,
+      "loss": 0.7794,
+      "step": 1230
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.252944901186866e-06,
+      "loss": 0.7943,
+      "step": 1231
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.24920496769489e-06,
+      "loss": 0.8437,
+      "step": 1232
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.245461884836364e-06,
+      "loss": 0.7792,
+      "step": 1233
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.241715656239356e-06,
+      "loss": 0.9584,
+      "step": 1234
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.237966285534984e-06,
+      "loss": 0.7377,
+      "step": 1235
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.234213776357416e-06,
+      "loss": 0.8634,
+      "step": 1236
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.230458132343853e-06,
+      "loss": 0.9165,
+      "step": 1237
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.226699357134538e-06,
+      "loss": 0.7243,
+      "step": 1238
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.222937454372757e-06,
+      "loss": 1.0427,
+      "step": 1239
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.219172427704814e-06,
+      "loss": 0.7074,
+      "step": 1240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.21540428078005e-06,
+      "loss": 0.9002,
+      "step": 1241
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.211633017250829e-06,
+      "loss": 0.6826,
+      "step": 1242
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.207858640772534e-06,
+      "loss": 0.7906,
+      "step": 1243
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.204081155003565e-06,
+      "loss": 0.8716,
+      "step": 1244
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.200300563605339e-06,
+      "loss": 0.8657,
+      "step": 1245
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.19651687024228e-06,
+      "loss": 1.0615,
+      "step": 1246
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.192730078581817e-06,
+      "loss": 0.8344,
+      "step": 1247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.18894019229439e-06,
+      "loss": 0.6587,
+      "step": 1248
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.18514721505343e-06,
+      "loss": 0.811,
+      "step": 1249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.181351150535367e-06,
+      "loss": 0.6639,
+      "step": 1250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.177552002419626e-06,
+      "loss": 0.9059,
+      "step": 1251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.173749774388617e-06,
+      "loss": 0.5774,
+      "step": 1252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.169944470127738e-06,
+      "loss": 0.8222,
+      "step": 1253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.166136093325365e-06,
+      "loss": 0.9749,
+      "step": 1254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.162324647672857e-06,
+      "loss": 0.7724,
+      "step": 1255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.158510136864546e-06,
+      "loss": 0.8193,
+      "step": 1256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.154692564597732e-06,
+      "loss": 0.9054,
+      "step": 1257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.150871934572683e-06,
+      "loss": 0.7646,
+      "step": 1258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.14704825049264e-06,
+      "loss": 0.7423,
+      "step": 1259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.143221516063789e-06,
+      "loss": 0.9064,
+      "step": 1260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.139391734995282e-06,
+      "loss": 0.7705,
+      "step": 1261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.135558910999224e-06,
+      "loss": 0.9946,
+      "step": 1262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.131723047790668e-06,
+      "loss": 0.8563,
+      "step": 1263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.127884149087607e-06,
+      "loss": 0.781,
+      "step": 1264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.12404221861099e-06,
+      "loss": 0.7517,
+      "step": 1265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.120197260084693e-06,
+      "loss": 0.8545,
+      "step": 1266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.116349277235526e-06,
+      "loss": 0.7673,
+      "step": 1267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.112498273793241e-06,
+      "loss": 0.8758,
+      "step": 1268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.10864425349051e-06,
+      "loss": 0.9505,
+      "step": 1269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.10478722006293e-06,
+      "loss": 0.7377,
+      "step": 1270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.100927177249016e-06,
+      "loss": 0.8697,
+      "step": 1271
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.097064128790208e-06,
+      "loss": 0.8201,
+      "step": 1272
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.093198078430853e-06,
+      "loss": 0.7965,
+      "step": 1273
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.08932902991821e-06,
+      "loss": 0.7078,
+      "step": 1274
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.08545698700244e-06,
+      "loss": 0.9246,
+      "step": 1275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.08158195343661e-06,
+      "loss": 0.7999,
+      "step": 1276
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.07770393297669e-06,
+      "loss": 0.7969,
+      "step": 1277
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.073822929381537e-06,
+      "loss": 0.5789,
+      "step": 1278
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.069938946412902e-06,
+      "loss": 0.8024,
+      "step": 1279
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.066051987835424e-06,
+      "loss": 0.8179,
+      "step": 1280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.062162057416631e-06,
+      "loss": 0.9319,
+      "step": 1281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.058269158926924e-06,
+      "loss": 0.7582,
+      "step": 1282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.054373296139586e-06,
+      "loss": 0.7034,
+      "step": 1283
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.05047447283077e-06,
+      "loss": 0.6827,
+      "step": 1284
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.046572692779501e-06,
+      "loss": 0.8734,
+      "step": 1285
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.042667959767669e-06,
+      "loss": 0.9446,
+      "step": 1286
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.038760277580028e-06,
+      "loss": 0.8456,
+      "step": 1287
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.034849650004184e-06,
+      "loss": 0.8128,
+      "step": 1288
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.030936080830607e-06,
+      "loss": 0.672,
+      "step": 1289
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.02701957385261e-06,
+      "loss": 0.7987,
+      "step": 1290
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.023100132866363e-06,
+      "loss": 0.7923,
+      "step": 1291
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.019177761670869e-06,
+      "loss": 0.8528,
+      "step": 1292
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.015252464067977e-06,
+      "loss": 0.8589,
+      "step": 1293
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.011324243862373e-06,
+      "loss": 0.9414,
+      "step": 1294
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.007393104861576e-06,
+      "loss": 0.9105,
+      "step": 1295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.00345905087593e-06,
+      "loss": 0.9244,
+      "step": 1296
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.99952208571861e-06,
+      "loss": 0.8747,
+      "step": 1297
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.995582213205609e-06,
+      "loss": 0.7196,
+      "step": 1298
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.991639437155737e-06,
+      "loss": 1.0153,
+      "step": 1299
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.98769376139062e-06,
+      "loss": 0.8517,
+      "step": 1300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.9837451897347e-06,
+      "loss": 0.7054,
+      "step": 1301
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.979793726015216e-06,
+      "loss": 0.8094,
+      "step": 1302
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.975839374062215e-06,
+      "loss": 0.728,
+      "step": 1303
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.971882137708546e-06,
+      "loss": 0.8033,
+      "step": 1304
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.967922020789852e-06,
+      "loss": 0.9154,
+      "step": 1305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.963959027144561e-06,
+      "loss": 0.7133,
+      "step": 1306
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.959993160613904e-06,
+      "loss": 1.0074,
+      "step": 1307
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.956024425041882e-06,
+      "loss": 0.8653,
+      "step": 1308
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.952052824275284e-06,
+      "loss": 0.9813,
+      "step": 1309
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.948078362163678e-06,
+      "loss": 0.7448,
+      "step": 1310
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.944101042559403e-06,
+      "loss": 0.8158,
+      "step": 1311
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.940120869317563e-06,
+      "loss": 0.7748,
+      "step": 1312
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.936137846296037e-06,
+      "loss": 0.7427,
+      "step": 1313
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.932151977355457e-06,
+      "loss": 0.8562,
+      "step": 1314
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.928163266359221e-06,
+      "loss": 0.9136,
+      "step": 1315
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.92417171717348e-06,
+      "loss": 1.0584,
+      "step": 1316
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.920177333667133e-06,
+      "loss": 0.759,
+      "step": 1317
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 7.916180119711826e-06,
+      "loss": 0.8964,
+      "step": 1318
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.912180079181955e-06,
+      "loss": 0.8855,
+      "step": 1319
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.908177215954647e-06,
+      "loss": 0.8903,
+      "step": 1320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.90417153390977e-06,
+      "loss": 0.9178,
+      "step": 1321
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.900163036929925e-06,
+      "loss": 1.1118,
+      "step": 1322
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.89615172890044e-06,
+      "loss": 0.7354,
+      "step": 1323
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.892137613709365e-06,
+      "loss": 0.8816,
+      "step": 1324
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.888120695247474e-06,
+      "loss": 0.881,
+      "step": 1325
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.884100977408258e-06,
+      "loss": 0.753,
+      "step": 1326
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.880078464087923e-06,
+      "loss": 0.7867,
+      "step": 1327
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.876053159185379e-06,
+      "loss": 0.9282,
+      "step": 1328
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.872025066602248e-06,
+      "loss": 0.8503,
+      "step": 1329
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 7.867994190242848e-06,
+      "loss": 0.6758,
+      "step": 1330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.8639605340142e-06,
+      "loss": 0.8734,
+      "step": 1331
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.859924101826021e-06,
+      "loss": 0.8385,
+      "step": 1332
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.855884897590713e-06,
+      "loss": 0.7796,
+      "step": 1333
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.851842925223365e-06,
+      "loss": 0.9565,
+      "step": 1334
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.847798188641755e-06,
+      "loss": 0.8402,
+      "step": 1335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.843750691766337e-06,
+      "loss": 0.8208,
+      "step": 1336
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.839700438520238e-06,
+      "loss": 0.7438,
+      "step": 1337
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.83564743282926e-06,
+      "loss": 0.8494,
+      "step": 1338
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.831591678621872e-06,
+      "loss": 0.8482,
+      "step": 1339
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.827533179829207e-06,
+      "loss": 0.741,
+      "step": 1340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.82347194038506e-06,
+      "loss": 0.9508,
+      "step": 1341
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 7.819407964225877e-06,
+      "loss": 0.831,
+      "step": 1342
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.815341255290763e-06,
+      "loss": 0.8623,
+      "step": 1343
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.811271817521467e-06,
+      "loss": 0.8215,
+      "step": 1344
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.807199654862387e-06,
+      "loss": 0.948,
+      "step": 1345
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.80312477126056e-06,
+      "loss": 0.862,
+      "step": 1346
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.79904717066566e-06,
+      "loss": 0.954,
+      "step": 1347
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.79496685702999e-06,
+      "loss": 0.8394,
+      "step": 1348
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.790883834308499e-06,
+      "loss": 0.6857,
+      "step": 1349
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.78679810645874e-06,
+      "loss": 0.8154,
+      "step": 1350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.782709677440906e-06,
+      "loss": 0.7856,
+      "step": 1351
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.778618551217793e-06,
+      "loss": 0.6972,
+      "step": 1352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 7.774524731754828e-06,
+      "loss": 0.7651,
+      "step": 1353
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.770428223020032e-06,
+      "loss": 0.7907,
+      "step": 1354
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.766329028984046e-06,
+      "loss": 0.8141,
+      "step": 1355
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.762227153620103e-06,
+      "loss": 0.7577,
+      "step": 1356
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.758122600904043e-06,
+      "loss": 0.8165,
+      "step": 1357
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.754015374814297e-06,
+      "loss": 1.0484,
+      "step": 1358
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.74990547933189e-06,
+      "loss": 0.7713,
+      "step": 1359
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.745792918440428e-06,
+      "loss": 0.8515,
+      "step": 1360
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.741677696126109e-06,
+      "loss": 1.0008,
+      "step": 1361
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.737559816377707e-06,
+      "loss": 0.7889,
+      "step": 1362
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.73343928318657e-06,
+      "loss": 0.7361,
+      "step": 1363
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.72931610054662e-06,
+      "loss": 0.884,
+      "step": 1364
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.725190272454346e-06,
+      "loss": 0.9249,
+      "step": 1365
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.721061802908802e-06,
+      "loss": 0.7989,
+      "step": 1366
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.716930695911601e-06,
+      "loss": 0.7336,
+      "step": 1367
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.712796955466916e-06,
+      "loss": 1.0,
+      "step": 1368
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.708660585581468e-06,
+      "loss": 0.7851,
+      "step": 1369
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.704521590264528e-06,
+      "loss": 0.7862,
+      "step": 1370
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.700379973527913e-06,
+      "loss": 0.8644,
+      "step": 1371
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.69623573938598e-06,
+      "loss": 0.813,
+      "step": 1372
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.692088891855624e-06,
+      "loss": 0.7961,
+      "step": 1373
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.687939434956272e-06,
+      "loss": 0.9151,
+      "step": 1374
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.68378737270988e-06,
+      "loss": 0.6837,
+      "step": 1375
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.679632709140926e-06,
+      "loss": 1.0069,
+      "step": 1376
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.675475448276419e-06,
+      "loss": 0.6677,
+      "step": 1377
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.671315594145877e-06,
+      "loss": 0.9099,
+      "step": 1378
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.667153150781336e-06,
+      "loss": 0.7294,
+      "step": 1379
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.662988122217337e-06,
+      "loss": 0.7783,
+      "step": 1380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.65882051249093e-06,
+      "loss": 0.7734,
+      "step": 1381
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.654650325641669e-06,
+      "loss": 0.7689,
+      "step": 1382
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.6504775657116e-06,
+      "loss": 0.781,
+      "step": 1383
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.64630223674527e-06,
+      "loss": 0.8659,
+      "step": 1384
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.64212434278971e-06,
+      "loss": 0.7793,
+      "step": 1385
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.637943887894441e-06,
+      "loss": 0.8332,
+      "step": 1386
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.633760876111465e-06,
+      "loss": 0.7817,
+      "step": 1387
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.629575311495259e-06,
+      "loss": 0.7154,
+      "step": 1388
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.625387198102781e-06,
+      "loss": 0.7349,
+      "step": 1389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.621196539993456e-06,
+      "loss": 0.9055,
+      "step": 1390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.617003341229174e-06,
+      "loss": 0.9576,
+      "step": 1391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.612807605874292e-06,
+      "loss": 0.8434,
+      "step": 1392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.608609337995621e-06,
+      "loss": 0.7587,
+      "step": 1393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.604408541662427e-06,
+      "loss": 0.8904,
+      "step": 1394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.600205220946429e-06,
+      "loss": 0.8067,
+      "step": 1395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.595999379921795e-06,
+      "loss": 0.7751,
+      "step": 1396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5917910226651315e-06,
+      "loss": 0.7369,
+      "step": 1397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.587580153255482e-06,
+      "loss": 0.7441,
+      "step": 1398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.583366775774336e-06,
+      "loss": 0.6865,
+      "step": 1399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.579150894305598e-06,
+      "loss": 0.8346,
+      "step": 1400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.574932512935613e-06,
+      "loss": 0.7665,
+      "step": 1401
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.570711635753144e-06,
+      "loss": 0.8244,
+      "step": 1402
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.566488266849372e-06,
+      "loss": 0.7639,
+      "step": 1403
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.562262410317893e-06,
+      "loss": 0.8461,
+      "step": 1404
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5580340702547185e-06,
+      "loss": 0.7808,
+      "step": 1405
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.553803250758261e-06,
+      "loss": 0.9139,
+      "step": 1406
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.549569955929341e-06,
+      "loss": 0.7312,
+      "step": 1407
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5453341898711774e-06,
+      "loss": 0.6745,
+      "step": 1408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.541095956689384e-06,
+      "loss": 0.8473,
+      "step": 1409
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.536855260491964e-06,
+      "loss": 0.76,
+      "step": 1410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.532612105389311e-06,
+      "loss": 0.7456,
+      "step": 1411
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5283664954942e-06,
+      "loss": 0.7599,
+      "step": 1412
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.524118434921785e-06,
+      "loss": 0.7564,
+      "step": 1413
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.519867927789597e-06,
+      "loss": 0.752,
+      "step": 1414
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.515614978217538e-06,
+      "loss": 0.804,
+      "step": 1415
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.511359590327878e-06,
+      "loss": 0.7919,
+      "step": 1416
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 1.0367615222930908,
+      "eval_runtime": 268.2832,
+      "eval_samples_per_second": 3.724,
+      "eval_steps_per_second": 0.466,
+      "step": 1416
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.507101768245248e-06,
+      "loss": 0.8669,
+      "step": 1417
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.5028415160966395e-06,
+      "loss": 0.8575,
+      "step": 1418
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.498578838011401e-06,
+      "loss": 0.8592,
+      "step": 1419
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.494313738121231e-06,
+      "loss": 0.8949,
+      "step": 1420
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.490046220560176e-06,
+      "loss": 0.8077,
+      "step": 1421
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.485776289464622e-06,
+      "loss": 0.8486,
+      "step": 1422
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.481503948973303e-06,
+      "loss": 0.8181,
+      "step": 1423
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.47722920322728e-06,
+      "loss": 0.7903,
+      "step": 1424
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.4729520563699476e-06,
+      "loss": 0.6929,
+      "step": 1425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.46867251254703e-06,
+      "loss": 0.8414,
+      "step": 1426
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.464390575906576e-06,
+      "loss": 0.9059,
+      "step": 1427
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.460106250598947e-06,
+      "loss": 0.9224,
+      "step": 1428
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.455819540776825e-06,
+      "loss": 0.6655,
+      "step": 1429
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.451530450595203e-06,
+      "loss": 0.7846,
+      "step": 1430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.447238984211379e-06,
+      "loss": 0.7918,
+      "step": 1431
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.442945145784954e-06,
+      "loss": 0.9527,
+      "step": 1432
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.438648939477835e-06,
+      "loss": 0.7507,
+      "step": 1433
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.4343503694542114e-06,
+      "loss": 0.8926,
+      "step": 1434
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.430049439880577e-06,
+      "loss": 1.0331,
+      "step": 1435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.425746154925702e-06,
+      "loss": 0.8403,
+      "step": 1436
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.421440518760648e-06,
+      "loss": 0.7929,
+      "step": 1437
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.417132535558748e-06,
+      "loss": 0.8996,
+      "step": 1438
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.412822209495617e-06,
+      "loss": 0.7115,
+      "step": 1439
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.4085095447491365e-06,
+      "loss": 0.7024,
+      "step": 1440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.404194545499455e-06,
+      "loss": 0.9465,
+      "step": 1441
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.399877215928985e-06,
+      "loss": 0.8044,
+      "step": 1442
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.395557560222397e-06,
+      "loss": 0.6062,
+      "step": 1443
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.391235582566619e-06,
+      "loss": 0.7368,
+      "step": 1444
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.386911287150824e-06,
+      "loss": 0.7269,
+      "step": 1445
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.382584678166436e-06,
+      "loss": 0.8868,
+      "step": 1446
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.37825575980712e-06,
+      "loss": 0.9653,
+      "step": 1447
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.373924536268779e-06,
+      "loss": 0.7983,
+      "step": 1448
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.3695910117495516e-06,
+      "loss": 0.8581,
+      "step": 1449
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.365255190449806e-06,
+      "loss": 0.86,
+      "step": 1450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.3609170765721375e-06,
+      "loss": 0.9287,
+      "step": 1451
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.35657667432136e-06,
+      "loss": 0.7142,
+      "step": 1452
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.35223398790451e-06,
+      "loss": 0.9649,
+      "step": 1453
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.347889021530837e-06,
+      "loss": 0.8857,
+      "step": 1454
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.343541779411796e-06,
+      "loss": 0.6911,
+      "step": 1455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.339192265761055e-06,
+      "loss": 0.8385,
+      "step": 1456
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.3348404847944775e-06,
+      "loss": 0.7312,
+      "step": 1457
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.330486440730129e-06,
+      "loss": 0.8133,
+      "step": 1458
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.326130137788265e-06,
+      "loss": 0.8751,
+      "step": 1459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.3217715801913335e-06,
+      "loss": 0.6908,
+      "step": 1460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.317410772163966e-06,
+      "loss": 0.6678,
+      "step": 1461
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.313047717932977e-06,
+      "loss": 0.9451,
+      "step": 1462
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.308682421727355e-06,
+      "loss": 0.6952,
+      "step": 1463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.304314887778264e-06,
+      "loss": 0.7807,
+      "step": 1464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.29994512031904e-06,
+      "loss": 0.9211,
+      "step": 1465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.295573123585174e-06,
+      "loss": 0.7447,
+      "step": 1466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.291198901814331e-06,
+      "loss": 0.7386,
+      "step": 1467
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.28682245924632e-06,
+      "loss": 0.6676,
+      "step": 1468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.282443800123113e-06,
+      "loss": 0.8112,
+      "step": 1469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.278062928688825e-06,
+      "loss": 0.7335,
+      "step": 1470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.2736798491897155e-06,
+      "loss": 0.8828,
+      "step": 1471
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.269294565874183e-06,
+      "loss": 0.7157,
+      "step": 1472
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.264907082992768e-06,
+      "loss": 0.7587,
+      "step": 1473
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.2605174047981365e-06,
+      "loss": 0.8258,
+      "step": 1474
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.256125535545087e-06,
+      "loss": 0.8392,
+      "step": 1475
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.251731479490536e-06,
+      "loss": 0.9391,
+      "step": 1476
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.247335240893527e-06,
+      "loss": 0.8991,
+      "step": 1477
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.242936824015212e-06,
+      "loss": 0.7769,
+      "step": 1478
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.238536233118862e-06,
+      "loss": 0.9265,
+      "step": 1479
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.234133472469847e-06,
+      "loss": 0.8697,
+      "step": 1480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.229728546335646e-06,
+      "loss": 0.7461,
+      "step": 1481
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.225321458985833e-06,
+      "loss": 0.7963,
+      "step": 1482
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.2209122146920795e-06,
+      "loss": 0.8294,
+      "step": 1483
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.216500817728146e-06,
+      "loss": 0.7819,
+      "step": 1484
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.2120872723698836e-06,
+      "loss": 0.6435,
+      "step": 1485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.207671582895218e-06,
+      "loss": 0.725,
+      "step": 1486
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.203253753584161e-06,
+      "loss": 0.9022,
+      "step": 1487
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.198833788718793e-06,
+      "loss": 1.0053,
+      "step": 1488
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.19441169258327e-06,
+      "loss": 0.7589,
+      "step": 1489
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.189987469463804e-06,
+      "loss": 0.7821,
+      "step": 1490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.185561123648681e-06,
+      "loss": 0.796,
+      "step": 1491
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.181132659428234e-06,
+      "loss": 0.8024,
+      "step": 1492
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.176702081094857e-06,
+      "loss": 0.7459,
+      "step": 1493
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.172269392942989e-06,
+      "loss": 0.7367,
+      "step": 1494
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.1678345992691134e-06,
+      "loss": 0.7041,
+      "step": 1495
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.163397704371755e-06,
+      "loss": 0.9484,
+      "step": 1496
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.15895871255148e-06,
+      "loss": 0.8342,
+      "step": 1497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.15451762811088e-06,
+      "loss": 0.877,
+      "step": 1498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.1500744553545796e-06,
+      "loss": 0.9226,
+      "step": 1499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.145629198589226e-06,
+      "loss": 0.7541,
+      "step": 1500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.1411818621234855e-06,
+      "loss": 0.7388,
+      "step": 1501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.136732450268039e-06,
+      "loss": 0.7822,
+      "step": 1502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.132280967335587e-06,
+      "loss": 0.7093,
+      "step": 1503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.127827417640827e-06,
+      "loss": 0.8519,
+      "step": 1504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.123371805500466e-06,
+      "loss": 0.7749,
+      "step": 1505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.118914135233208e-06,
+      "loss": 0.7953,
+      "step": 1506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.114454411159752e-06,
+      "loss": 0.9955,
+      "step": 1507
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.109992637602788e-06,
+      "loss": 0.8843,
+      "step": 1508
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.105528818886993e-06,
+      "loss": 0.8694,
+      "step": 1509
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.1010629593390254e-06,
+      "loss": 0.9738,
+      "step": 1510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.096595063287523e-06,
+      "loss": 0.8214,
+      "step": 1511
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.092125135063097e-06,
+      "loss": 0.9725,
+      "step": 1512
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.087653178998326e-06,
+      "loss": 0.8117,
+      "step": 1513
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.0831791994277565e-06,
+      "loss": 0.9358,
+      "step": 1514
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.078703200687899e-06,
+      "loss": 0.8061,
+      "step": 1515
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.074225187117213e-06,
+      "loss": 0.798,
+      "step": 1516
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.069745163056121e-06,
+      "loss": 0.7755,
+      "step": 1517
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.065263132846987e-06,
+      "loss": 0.6874,
+      "step": 1518
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.060779100834121e-06,
+      "loss": 0.7539,
+      "step": 1519
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.056293071363773e-06,
+      "loss": 0.8316,
+      "step": 1520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.051805048784135e-06,
+      "loss": 0.8392,
+      "step": 1521
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.047315037445321e-06,
+      "loss": 0.7544,
+      "step": 1522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.0428230416993805e-06,
+      "loss": 0.6489,
+      "step": 1523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.038329065900282e-06,
+      "loss": 0.7169,
+      "step": 1524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.033833114403916e-06,
+      "loss": 0.8197,
+      "step": 1525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.029335191568084e-06,
+      "loss": 0.7736,
+      "step": 1526
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.024835301752506e-06,
+      "loss": 0.9876,
+      "step": 1527
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.0203334493187995e-06,
+      "loss": 0.9049,
+      "step": 1528
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.015829638630489e-06,
+      "loss": 0.9423,
+      "step": 1529
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.011323874052997e-06,
+      "loss": 0.6799,
+      "step": 1530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.0068161599536425e-06,
+      "loss": 0.8435,
+      "step": 1531
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.002306500701623e-06,
+      "loss": 0.6984,
+      "step": 1532
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.997794900668037e-06,
+      "loss": 0.7299,
+      "step": 1533
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.993281364225852e-06,
+      "loss": 0.709,
+      "step": 1534
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.98876589574992e-06,
+      "loss": 0.8922,
+      "step": 1535
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.984248499616962e-06,
+      "loss": 0.734,
+      "step": 1536
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.979729180205565e-06,
+      "loss": 0.7569,
+      "step": 1537
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.975207941896183e-06,
+      "loss": 0.7865,
+      "step": 1538
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.970684789071136e-06,
+      "loss": 0.7004,
+      "step": 1539
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.9661597261145884e-06,
+      "loss": 0.8083,
+      "step": 1540
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.961632757412564e-06,
+      "loss": 0.828,
+      "step": 1541
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.95710388735293e-06,
+      "loss": 0.7107,
+      "step": 1542
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.952573120325396e-06,
+      "loss": 0.7688,
+      "step": 1543
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9480404607215146e-06,
+      "loss": 0.7668,
+      "step": 1544
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.943505912934671e-06,
+      "loss": 0.86,
+      "step": 1545
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.9389694813600785e-06,
+      "loss": 0.8593,
+      "step": 1546
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.934431170394776e-06,
+      "loss": 0.8509,
+      "step": 1547
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.929890984437629e-06,
+      "loss": 0.7964,
+      "step": 1548
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.925348927889314e-06,
+      "loss": 1.0028,
+      "step": 1549
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.920805005152322e-06,
+      "loss": 0.7815,
+      "step": 1550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.916259220630961e-06,
+      "loss": 0.9377,
+      "step": 1551
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.911711578731333e-06,
+      "loss": 0.7634,
+      "step": 1552
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.907162083861344e-06,
+      "loss": 0.6944,
+      "step": 1553
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.902610740430698e-06,
+      "loss": 0.9568,
+      "step": 1554
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.898057552850887e-06,
+      "loss": 0.9048,
+      "step": 1555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8935025255351926e-06,
+      "loss": 0.8771,
+      "step": 1556
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.888945662898684e-06,
+      "loss": 0.716,
+      "step": 1557
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.884386969358201e-06,
+      "loss": 0.7499,
+      "step": 1558
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.879826449332363e-06,
+      "loss": 0.8372,
+      "step": 1559
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.875264107241558e-06,
+      "loss": 0.6733,
+      "step": 1560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.870699947507944e-06,
+      "loss": 0.685,
+      "step": 1561
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.866133974555431e-06,
+      "loss": 0.7039,
+      "step": 1562
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.861566192809701e-06,
+      "loss": 0.7865,
+      "step": 1563
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.856996606698177e-06,
+      "loss": 0.8474,
+      "step": 1564
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.852425220650037e-06,
+      "loss": 0.9083,
+      "step": 1565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.8478520390962e-06,
+      "loss": 0.6751,
+      "step": 1566
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.84327706646933e-06,
+      "loss": 0.8632,
+      "step": 1567
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.83870030720382e-06,
+      "loss": 0.8064,
+      "step": 1568
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.8341217657358035e-06,
+      "loss": 0.9168,
+      "step": 1569
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.829541446503134e-06,
+      "loss": 0.8816,
+      "step": 1570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.824959353945395e-06,
+      "loss": 0.7874,
+      "step": 1571
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.82037549250388e-06,
+      "loss": 0.8048,
+      "step": 1572
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.815789866621605e-06,
+      "loss": 0.7967,
+      "step": 1573
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.8112024807432915e-06,
+      "loss": 0.7529,
+      "step": 1574
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.806613339315371e-06,
+      "loss": 0.9147,
+      "step": 1575
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.802022446785972e-06,
+      "loss": 0.6106,
+      "step": 1576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.797429807604925e-06,
+      "loss": 0.8163,
+      "step": 1577
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.792835426223748e-06,
+      "loss": 0.7456,
+      "step": 1578
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.788239307095655e-06,
+      "loss": 0.7979,
+      "step": 1579
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.783641454675533e-06,
+      "loss": 0.9002,
+      "step": 1580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.779041873419963e-06,
+      "loss": 0.7064,
+      "step": 1581
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.77444056778719e-06,
+      "loss": 0.7799,
+      "step": 1582
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.769837542237139e-06,
+      "loss": 0.7816,
+      "step": 1583
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.765232801231396e-06,
+      "loss": 0.8002,
+      "step": 1584
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.760626349233213e-06,
+      "loss": 0.7965,
+      "step": 1585
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.7560181907074965e-06,
+      "loss": 0.7012,
+      "step": 1586
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.751408330120815e-06,
+      "loss": 0.8172,
+      "step": 1587
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.746796771941378e-06,
+      "loss": 0.8561,
+      "step": 1588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.7421835206390465e-06,
+      "loss": 0.716,
+      "step": 1589
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.737568580685318e-06,
+      "loss": 0.8691,
+      "step": 1590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.7329519565533306e-06,
+      "loss": 0.7289,
+      "step": 1591
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.728333652717852e-06,
+      "loss": 0.7951,
+      "step": 1592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.723713673655281e-06,
+      "loss": 0.8098,
+      "step": 1593
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.719092023843637e-06,
+      "loss": 0.6995,
+      "step": 1594
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.7144687077625625e-06,
+      "loss": 1.0442,
+      "step": 1595
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.709843729893311e-06,
+      "loss": 0.7996,
+      "step": 1596
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.705217094718747e-06,
+      "loss": 0.8238,
+      "step": 1597
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.700588806723345e-06,
+      "loss": 0.6588,
+      "step": 1598
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.69595887039318e-06,
+      "loss": 0.8916,
+      "step": 1599
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.691327290215922e-06,
+      "loss": 0.9594,
+      "step": 1600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.686694070680839e-06,
+      "loss": 0.7763,
+      "step": 1601
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.6820592162787825e-06,
+      "loss": 0.8663,
+      "step": 1602
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.677422731502194e-06,
+      "loss": 0.7139,
+      "step": 1603
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.672784620845092e-06,
+      "loss": 0.6489,
+      "step": 1604
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.668144888803072e-06,
+      "loss": 0.8217,
+      "step": 1605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.6635035398733e-06,
+      "loss": 0.7694,
+      "step": 1606
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.658860578554513e-06,
+      "loss": 0.7361,
+      "step": 1607
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.654216009347003e-06,
+      "loss": 0.929,
+      "step": 1608
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.6495698367526295e-06,
+      "loss": 0.7519,
+      "step": 1609
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.6449220652748e-06,
+      "loss": 0.7888,
+      "step": 1610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.6402726994184766e-06,
+      "loss": 0.8255,
+      "step": 1611
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.635621743690159e-06,
+      "loss": 0.9062,
+      "step": 1612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.630969202597898e-06,
+      "loss": 0.7494,
+      "step": 1613
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.626315080651271e-06,
+      "loss": 0.7022,
+      "step": 1614
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.621659382361396e-06,
+      "loss": 0.7735,
+      "step": 1615
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.617002112240914e-06,
+      "loss": 0.8225,
+      "step": 1616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.612343274803994e-06,
+      "loss": 0.6249,
+      "step": 1617
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.607682874566317e-06,
+      "loss": 0.8611,
+      "step": 1618
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.603020916045086e-06,
+      "loss": 0.7932,
+      "step": 1619
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.598357403759009e-06,
+      "loss": 0.9676,
+      "step": 1620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.593692342228304e-06,
+      "loss": 0.8292,
+      "step": 1621
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.589025735974689e-06,
+      "loss": 0.8963,
+      "step": 1622
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.5843575895213795e-06,
+      "loss": 0.8881,
+      "step": 1623
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.579687907393082e-06,
+      "loss": 0.7218,
+      "step": 1624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.5750166941159964e-06,
+      "loss": 0.733,
+      "step": 1625
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.5703439542178e-06,
+      "loss": 0.7412,
+      "step": 1626
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.565669692227654e-06,
+      "loss": 0.781,
+      "step": 1627
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.560993912676194e-06,
+      "loss": 0.6802,
+      "step": 1628
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.556316620095531e-06,
+      "loss": 0.7592,
+      "step": 1629
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.551637819019233e-06,
+      "loss": 0.9526,
+      "step": 1630
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.5469575139823385e-06,
+      "loss": 0.9086,
+      "step": 1631
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.542275709521338e-06,
+      "loss": 0.7871,
+      "step": 1632
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.53759241017418e-06,
+      "loss": 0.7811,
+      "step": 1633
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.532907620480259e-06,
+      "loss": 0.7815,
+      "step": 1634
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.528221344980417e-06,
+      "loss": 0.7998,
+      "step": 1635
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.52353358821693e-06,
+      "loss": 0.7669,
+      "step": 1636
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 6.51884435473352e-06,
+      "loss": 1.0003,
+      "step": 1637
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.5141536490753285e-06,
+      "loss": 0.8186,
+      "step": 1638
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.509461475788932e-06,
+      "loss": 0.8063,
+      "step": 1639
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.504767839422328e-06,
+      "loss": 0.897,
+      "step": 1640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.500072744524931e-06,
+      "loss": 0.779,
+      "step": 1641
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.495376195647569e-06,
+      "loss": 0.7304,
+      "step": 1642
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.490678197342482e-06,
+      "loss": 0.7722,
+      "step": 1643
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.485978754163307e-06,
+      "loss": 0.8341,
+      "step": 1644
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.481277870665093e-06,
+      "loss": 0.7353,
+      "step": 1645
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.476575551404277e-06,
+      "loss": 0.8605,
+      "step": 1646
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.471871800938691e-06,
+      "loss": 0.7962,
+      "step": 1647
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.467166623827553e-06,
+      "loss": 0.8686,
+      "step": 1648
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.462460024631464e-06,
+      "loss": 0.9356,
+      "step": 1649
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.457752007912402e-06,
+      "loss": 0.8558,
+      "step": 1650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.453042578233722e-06,
+      "loss": 0.6997,
+      "step": 1651
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.448331740160145e-06,
+      "loss": 0.8701,
+      "step": 1652
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 1.019251823425293,
+      "eval_runtime": 267.9006,
+      "eval_samples_per_second": 3.729,
+      "eval_steps_per_second": 0.467,
+      "step": 1652
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.443619498257761e-06,
+      "loss": 0.7861,
+      "step": 1653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.438905857094018e-06,
+      "loss": 0.7611,
+      "step": 1654
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.434190821237721e-06,
+      "loss": 0.7244,
+      "step": 1655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.4294743952590265e-06,
+      "loss": 0.7189,
+      "step": 1656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.424756583729439e-06,
+      "loss": 0.8293,
+      "step": 1657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.420037391221805e-06,
+      "loss": 0.84,
+      "step": 1658
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.415316822310313e-06,
+      "loss": 0.9294,
+      "step": 1659
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.410594881570478e-06,
+      "loss": 1.1337,
+      "step": 1660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.405871573579155e-06,
+      "loss": 0.7957,
+      "step": 1661
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.401146902914513e-06,
+      "loss": 0.7517,
+      "step": 1662
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.396420874156052e-06,
+      "loss": 0.9707,
+      "step": 1663
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.391693491884583e-06,
+      "loss": 0.7599,
+      "step": 1664
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.386964760682229e-06,
+      "loss": 0.8022,
+      "step": 1665
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.382234685132421e-06,
+      "loss": 0.671,
+      "step": 1666
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.377503269819896e-06,
+      "loss": 0.7363,
+      "step": 1667
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.372770519330683e-06,
+      "loss": 0.9233,
+      "step": 1668
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.368036438252112e-06,
+      "loss": 0.8845,
+      "step": 1669
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.363301031172799e-06,
+      "loss": 0.7487,
+      "step": 1670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.358564302682646e-06,
+      "loss": 0.7732,
+      "step": 1671
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.353826257372836e-06,
+      "loss": 0.789,
+      "step": 1672
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.349086899835829e-06,
+      "loss": 0.7905,
+      "step": 1673
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.344346234665355e-06,
+      "loss": 0.7905,
+      "step": 1674
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.339604266456412e-06,
+      "loss": 0.7234,
+      "step": 1675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.334860999805264e-06,
+      "loss": 0.89,
+      "step": 1676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.330116439309429e-06,
+      "loss": 0.9055,
+      "step": 1677
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.325370589567682e-06,
+      "loss": 0.8932,
+      "step": 1678
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.320623455180048e-06,
+      "loss": 0.8352,
+      "step": 1679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.315875040747795e-06,
+      "loss": 0.7015,
+      "step": 1680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.31112535087343e-06,
+      "loss": 0.8798,
+      "step": 1681
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.306374390160703e-06,
+      "loss": 0.6815,
+      "step": 1682
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.3016221632145915e-06,
+      "loss": 0.7445,
+      "step": 1683
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.296868674641299e-06,
+      "loss": 0.8447,
+      "step": 1684
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.292113929048256e-06,
+      "loss": 0.8892,
+      "step": 1685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.287357931044104e-06,
+      "loss": 0.7975,
+      "step": 1686
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.282600685238708e-06,
+      "loss": 0.7653,
+      "step": 1687
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.2778421962431345e-06,
+      "loss": 0.8072,
+      "step": 1688
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.273082468669662e-06,
+      "loss": 0.6993,
+      "step": 1689
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.268321507131763e-06,
+      "loss": 0.8691,
+      "step": 1690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.26355931624411e-06,
+      "loss": 0.841,
+      "step": 1691
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.258795900622564e-06,
+      "loss": 0.7442,
+      "step": 1692
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.2540312648841775e-06,
+      "loss": 0.8477,
+      "step": 1693
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.249265413647181e-06,
+      "loss": 0.8999,
+      "step": 1694
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.244498351530988e-06,
+      "loss": 0.9991,
+      "step": 1695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.239730083156179e-06,
+      "loss": 0.8485,
+      "step": 1696
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.234960613144512e-06,
+      "loss": 0.8955,
+      "step": 1697
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.230189946118902e-06,
+      "loss": 0.9616,
+      "step": 1698
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.22541808670343e-06,
+      "loss": 0.8274,
+      "step": 1699
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.220645039523327e-06,
+      "loss": 0.781,
+      "step": 1700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.215870809204983e-06,
+      "loss": 0.8835,
+      "step": 1701
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.2110954003759284e-06,
+      "loss": 0.8088,
+      "step": 1702
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.206318817664838e-06,
+      "loss": 0.7366,
+      "step": 1703
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.201541065701522e-06,
+      "loss": 0.8072,
+      "step": 1704
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.196762149116929e-06,
+      "loss": 0.648,
+      "step": 1705
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.191982072543133e-06,
+      "loss": 0.7447,
+      "step": 1706
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.187200840613334e-06,
+      "loss": 0.7933,
+      "step": 1707
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.18241845796185e-06,
+      "loss": 0.8465,
+      "step": 1708
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.177634929224115e-06,
+      "loss": 0.8928,
+      "step": 1709
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.172850259036674e-06,
+      "loss": 0.8205,
+      "step": 1710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.168064452037177e-06,
+      "loss": 0.7432,
+      "step": 1711
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.163277512864381e-06,
+      "loss": 0.7173,
+      "step": 1712
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.158489446158135e-06,
+      "loss": 0.7838,
+      "step": 1713
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.153700256559381e-06,
+      "loss": 0.9283,
+      "step": 1714
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.148909948710153e-06,
+      "loss": 0.9365,
+      "step": 1715
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.144118527253566e-06,
+      "loss": 0.6187,
+      "step": 1716
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.139325996833817e-06,
+      "loss": 0.7246,
+      "step": 1717
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.1345323620961736e-06,
+      "loss": 0.8539,
+      "step": 1718
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.1297376276869795e-06,
+      "loss": 0.8305,
+      "step": 1719
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.124941798253637e-06,
+      "loss": 0.6663,
+      "step": 1720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.120144878444618e-06,
+      "loss": 0.9437,
+      "step": 1721
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.115346872909445e-06,
+      "loss": 0.8885,
+      "step": 1722
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.110547786298695e-06,
+      "loss": 0.7806,
+      "step": 1723
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.105747623263994e-06,
+      "loss": 0.8618,
+      "step": 1724
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.1009463884580115e-06,
+      "loss": 0.8203,
+      "step": 1725
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.09614408653445e-06,
+      "loss": 0.9499,
+      "step": 1726
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.091340722148056e-06,
+      "loss": 0.7276,
+      "step": 1727
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.086536299954597e-06,
+      "loss": 0.7828,
+      "step": 1728
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.081730824610872e-06,
+      "loss": 0.8343,
+      "step": 1729
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.076924300774697e-06,
+      "loss": 0.6948,
+      "step": 1730
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.072116733104908e-06,
+      "loss": 0.8671,
+      "step": 1731
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.0673081262613465e-06,
+      "loss": 0.6016,
+      "step": 1732
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.062498484904868e-06,
+      "loss": 0.7806,
+      "step": 1733
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.0576878136973285e-06,
+      "loss": 0.6163,
+      "step": 1734
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.052876117301578e-06,
+      "loss": 0.7998,
+      "step": 1735
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.048063400381466e-06,
+      "loss": 0.7892,
+      "step": 1736
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.04324966760183e-06,
+      "loss": 0.8627,
+      "step": 1737
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.03843492362849e-06,
+      "loss": 0.8486,
+      "step": 1738
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.033619173128247e-06,
+      "loss": 0.8075,
+      "step": 1739
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.028802420768878e-06,
+      "loss": 0.7822,
+      "step": 1740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.023984671219129e-06,
+      "loss": 0.7715,
+      "step": 1741
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.019165929148717e-06,
+      "loss": 1.0195,
+      "step": 1742
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.014346199228319e-06,
+      "loss": 0.6829,
+      "step": 1743
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.009525486129567e-06,
+      "loss": 0.7233,
+      "step": 1744
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.004703794525049e-06,
+      "loss": 0.851,
+      "step": 1745
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.9998811290882985e-06,
+      "loss": 0.6982,
+      "step": 1746
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.995057494493795e-06,
+      "loss": 0.7574,
+      "step": 1747
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.99023289541696e-06,
+      "loss": 0.7725,
+      "step": 1748
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.985407336534146e-06,
+      "loss": 0.6535,
+      "step": 1749
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.980580822522635e-06,
+      "loss": 0.6778,
+      "step": 1750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.9757533580606374e-06,
+      "loss": 0.8918,
+      "step": 1751
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.970924947827283e-06,
+      "loss": 0.661,
+      "step": 1752
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.966095596502623e-06,
+      "loss": 0.8684,
+      "step": 1753
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.961265308767612e-06,
+      "loss": 0.9413,
+      "step": 1754
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.956434089304122e-06,
+      "loss": 0.735,
+      "step": 1755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.9516019427949205e-06,
+      "loss": 0.7334,
+      "step": 1756
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.946768873923677e-06,
+      "loss": 0.7728,
+      "step": 1757
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.941934887374955e-06,
+      "loss": 0.754,
+      "step": 1758
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9370999878342075e-06,
+      "loss": 0.8462,
+      "step": 1759
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.93226417998777e-06,
+      "loss": 0.6794,
+      "step": 1760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.927427468522864e-06,
+      "loss": 0.8295,
+      "step": 1761
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9225898581275785e-06,
+      "loss": 0.8031,
+      "step": 1762
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.917751353490882e-06,
+      "loss": 0.7535,
+      "step": 1763
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.9129119593026055e-06,
+      "loss": 0.7867,
+      "step": 1764
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.908071680253441e-06,
+      "loss": 0.5099,
+      "step": 1765
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.903230521034942e-06,
+      "loss": 0.7524,
+      "step": 1766
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.898388486339515e-06,
+      "loss": 0.7248,
+      "step": 1767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.89354558086041e-06,
+      "loss": 0.6886,
+      "step": 1768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.888701809291727e-06,
+      "loss": 0.7659,
+      "step": 1769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.883857176328399e-06,
+      "loss": 0.8977,
+      "step": 1770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.879011686666202e-06,
+      "loss": 0.6683,
+      "step": 1771
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.874165345001737e-06,
+      "loss": 1.0129,
+      "step": 1772
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8693181560324304e-06,
+      "loss": 0.8211,
+      "step": 1773
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.864470124456533e-06,
+      "loss": 0.5836,
+      "step": 1774
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.859621254973111e-06,
+      "loss": 0.7451,
+      "step": 1775
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.854771552282041e-06,
+      "loss": 0.907,
+      "step": 1776
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.84992102108401e-06,
+      "loss": 0.8019,
+      "step": 1777
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8450696660805054e-06,
+      "loss": 0.8161,
+      "step": 1778
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8402174919738164e-06,
+      "loss": 0.7245,
+      "step": 1779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.835364503467022e-06,
+      "loss": 0.8087,
+      "step": 1780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.830510705263996e-06,
+      "loss": 0.8771,
+      "step": 1781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.825656102069392e-06,
+      "loss": 0.6287,
+      "step": 1782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.820800698588643e-06,
+      "loss": 0.7347,
+      "step": 1783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.815944499527964e-06,
+      "loss": 0.8863,
+      "step": 1784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.811087509594337e-06,
+      "loss": 0.8002,
+      "step": 1785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.806229733495511e-06,
+      "loss": 0.7222,
+      "step": 1786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.801371175939996e-06,
+      "loss": 0.6501,
+      "step": 1787
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.796511841637061e-06,
+      "loss": 0.9007,
+      "step": 1788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.791651735296726e-06,
+      "loss": 0.7025,
+      "step": 1789
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.786790861629763e-06,
+      "loss": 0.8426,
+      "step": 1790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.781929225347684e-06,
+      "loss": 0.6317,
+      "step": 1791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.777066831162743e-06,
+      "loss": 0.8952,
+      "step": 1792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.772203683787927e-06,
+      "loss": 0.7287,
+      "step": 1793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.767339787936952e-06,
+      "loss": 0.7516,
+      "step": 1794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.76247514832426e-06,
+      "loss": 0.8844,
+      "step": 1795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.757609769665018e-06,
+      "loss": 0.7991,
+      "step": 1796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.752743656675104e-06,
+      "loss": 0.8731,
+      "step": 1797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.74787681407111e-06,
+      "loss": 0.6627,
+      "step": 1798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.743009246570336e-06,
+      "loss": 0.7311,
+      "step": 1799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.738140958890781e-06,
+      "loss": 0.7227,
+      "step": 1800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.733271955751146e-06,
+      "loss": 0.7055,
+      "step": 1801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.728402241870825e-06,
+      "loss": 0.8555,
+      "step": 1802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.7235318219699e-06,
+      "loss": 0.7868,
+      "step": 1803
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.7186607007691366e-06,
+      "loss": 0.8623,
+      "step": 1804
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.713788882989981e-06,
+      "loss": 0.7736,
+      "step": 1805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.708916373354555e-06,
+      "loss": 0.9456,
+      "step": 1806
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.704043176585649e-06,
+      "loss": 0.6887,
+      "step": 1807
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.699169297406723e-06,
+      "loss": 0.7537,
+      "step": 1808
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.694294740541895e-06,
+      "loss": 0.8797,
+      "step": 1809
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.689419510715941e-06,
+      "loss": 0.7586,
+      "step": 1810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.684543612654292e-06,
+      "loss": 0.8328,
+      "step": 1811
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.67966705108302e-06,
+      "loss": 0.7381,
+      "step": 1812
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.674789830728847e-06,
+      "loss": 1.0648,
+      "step": 1813
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.669911956319131e-06,
+      "loss": 0.7295,
+      "step": 1814
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.665033432581864e-06,
+      "loss": 0.7897,
+      "step": 1815
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.6601542642456655e-06,
+      "loss": 0.8105,
+      "step": 1816
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.655274456039783e-06,
+      "loss": 0.7073,
+      "step": 1817
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.650394012694082e-06,
+      "loss": 0.7645,
+      "step": 1818
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.645512938939044e-06,
+      "loss": 0.8155,
+      "step": 1819
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.640631239505762e-06,
+      "loss": 1.0018,
+      "step": 1820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.635748919125936e-06,
+      "loss": 0.7351,
+      "step": 1821
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.630865982531867e-06,
+      "loss": 0.7433,
+      "step": 1822
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.625982434456454e-06,
+      "loss": 0.9961,
+      "step": 1823
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.6210982796331856e-06,
+      "loss": 0.673,
+      "step": 1824
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.616213522796144e-06,
+      "loss": 1.0792,
+      "step": 1825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.61132816867999e-06,
+      "loss": 1.0037,
+      "step": 1826
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.606442222019967e-06,
+      "loss": 0.758,
+      "step": 1827
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.601555687551891e-06,
+      "loss": 0.8124,
+      "step": 1828
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.596668570012146e-06,
+      "loss": 0.7336,
+      "step": 1829
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.591780874137683e-06,
+      "loss": 0.7942,
+      "step": 1830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.586892604666016e-06,
+      "loss": 0.7288,
+      "step": 1831
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.582003766335209e-06,
+      "loss": 0.8289,
+      "step": 1832
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.577114363883884e-06,
+      "loss": 0.7707,
+      "step": 1833
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.572224402051205e-06,
+      "loss": 0.7696,
+      "step": 1834
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.56733388557688e-06,
+      "loss": 0.6736,
+      "step": 1835
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.562442819201155e-06,
+      "loss": 0.6947,
+      "step": 1836
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.557551207664806e-06,
+      "loss": 0.7547,
+      "step": 1837
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.552659055709144e-06,
+      "loss": 0.7358,
+      "step": 1838
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.547766368075997e-06,
+      "loss": 0.7739,
+      "step": 1839
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.542873149507716e-06,
+      "loss": 0.8199,
+      "step": 1840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.537979404747165e-06,
+      "loss": 0.6472,
+      "step": 1841
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.533085138537717e-06,
+      "loss": 0.7943,
+      "step": 1842
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.528190355623253e-06,
+      "loss": 0.7753,
+      "step": 1843
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.523295060748155e-06,
+      "loss": 0.8438,
+      "step": 1844
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.5183992586573e-06,
+      "loss": 0.7841,
+      "step": 1845
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.513502954096055e-06,
+      "loss": 0.8888,
+      "step": 1846
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.508606151810278e-06,
+      "loss": 0.9943,
+      "step": 1847
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.503708856546303e-06,
+      "loss": 0.8496,
+      "step": 1848
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.498811073050948e-06,
+      "loss": 0.8908,
+      "step": 1849
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.493912806071504e-06,
+      "loss": 0.6755,
+      "step": 1850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.4890140603557285e-06,
+      "loss": 0.8549,
+      "step": 1851
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.48411484065184e-06,
+      "loss": 0.6507,
+      "step": 1852
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.479215151708525e-06,
+      "loss": 0.7968,
+      "step": 1853
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.474314998274914e-06,
+      "loss": 0.9051,
+      "step": 1854
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.469414385100596e-06,
+      "loss": 0.7152,
+      "step": 1855
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.4645133169356025e-06,
+      "loss": 0.7948,
+      "step": 1856
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.459611798530408e-06,
+      "loss": 0.697,
+      "step": 1857
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.454709834635919e-06,
+      "loss": 0.7676,
+      "step": 1858
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.449807430003478e-06,
+      "loss": 0.8472,
+      "step": 1859
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.4449045893848526e-06,
+      "loss": 0.7588,
+      "step": 1860
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.440001317532235e-06,
+      "loss": 0.8365,
+      "step": 1861
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.435097619198234e-06,
+      "loss": 0.7487,
+      "step": 1862
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.430193499135872e-06,
+      "loss": 0.7774,
+      "step": 1863
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.425288962098577e-06,
+      "loss": 0.7545,
+      "step": 1864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.420384012840189e-06,
+      "loss": 0.8279,
+      "step": 1865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.415478656114939e-06,
+      "loss": 0.715,
+      "step": 1866
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.410572896677457e-06,
+      "loss": 0.7693,
+      "step": 1867
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.4056667392827645e-06,
+      "loss": 0.8038,
+      "step": 1868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.400760188686267e-06,
+      "loss": 0.8206,
+      "step": 1869
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.395853249643751e-06,
+      "loss": 0.7561,
+      "step": 1870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.390945926911381e-06,
+      "loss": 0.6824,
+      "step": 1871
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.38603822524569e-06,
+      "loss": 0.8195,
+      "step": 1872
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.381130149403581e-06,
+      "loss": 0.7798,
+      "step": 1873
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.3762217041423204e-06,
+      "loss": 0.6798,
+      "step": 1874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.371312894219532e-06,
+      "loss": 0.6628,
+      "step": 1875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.366403724393191e-06,
+      "loss": 0.7256,
+      "step": 1876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.361494199421625e-06,
+      "loss": 0.8278,
+      "step": 1877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.3565843240635e-06,
+      "loss": 0.7383,
+      "step": 1878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.351674103077831e-06,
+      "loss": 0.8555,
+      "step": 1879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.346763541223957e-06,
+      "loss": 0.8369,
+      "step": 1880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.341852643261558e-06,
+      "loss": 0.8539,
+      "step": 1881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.33694141395063e-06,
+      "loss": 0.7739,
+      "step": 1882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.332029858051498e-06,
+      "loss": 0.8637,
+      "step": 1883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.327117980324799e-06,
+      "loss": 0.7614,
+      "step": 1884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.3222057855314824e-06,
+      "loss": 0.7729,
+      "step": 1885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.317293278432808e-06,
+      "loss": 0.8611,
+      "step": 1886
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.312380463790335e-06,
+      "loss": 0.8297,
+      "step": 1887
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.307467346365922e-06,
+      "loss": 0.8851,
+      "step": 1888
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 1.0008699893951416,
+      "eval_runtime": 267.9602,
+      "eval_samples_per_second": 3.728,
+      "eval_steps_per_second": 0.466,
+      "step": 1888
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.302553930921722e-06,
+      "loss": 0.9724,
+      "step": 1889
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.297640222220174e-06,
+      "loss": 0.8591,
+      "step": 1890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.292726225024003e-06,
+      "loss": 0.7449,
+      "step": 1891
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.287811944096214e-06,
+      "loss": 0.8564,
+      "step": 1892
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.282897384200088e-06,
+      "loss": 0.7152,
+      "step": 1893
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.2779825500991746e-06,
+      "loss": 0.6963,
+      "step": 1894
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.2730674465572895e-06,
+      "loss": 0.8989,
+      "step": 1895
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.268152078338507e-06,
+      "loss": 0.7642,
+      "step": 1896
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.263236450207165e-06,
+      "loss": 0.983,
+      "step": 1897
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.258320566927846e-06,
+      "loss": 0.7734,
+      "step": 1898
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.2534044332653854e-06,
+      "loss": 0.7092,
+      "step": 1899
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.248488053984856e-06,
+      "loss": 0.9293,
+      "step": 1900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.243571433851573e-06,
+      "loss": 0.9214,
+      "step": 1901
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.238654577631081e-06,
+      "loss": 0.7068,
+      "step": 1902
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.233737490089159e-06,
+      "loss": 0.7364,
+      "step": 1903
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.228820175991806e-06,
+      "loss": 0.7443,
+      "step": 1904
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.223902640105242e-06,
+      "loss": 0.8321,
+      "step": 1905
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.2189848871958995e-06,
+      "loss": 0.814,
+      "step": 1906
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.214066922030426e-06,
+      "loss": 0.8511,
+      "step": 1907
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.20914874937567e-06,
+      "loss": 0.7629,
+      "step": 1908
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.204230373998684e-06,
+      "loss": 0.784,
+      "step": 1909
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.199311800666717e-06,
+      "loss": 0.9027,
+      "step": 1910
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.1943930341472115e-06,
+      "loss": 0.9017,
+      "step": 1911
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.1894740792077904e-06,
+      "loss": 0.7535,
+      "step": 1912
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.184554940616266e-06,
+      "loss": 0.7154,
+      "step": 1913
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.179635623140627e-06,
+      "loss": 0.9013,
+      "step": 1914
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.174716131549032e-06,
+      "loss": 0.5612,
+      "step": 1915
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.169796470609814e-06,
+      "loss": 0.8373,
+      "step": 1916
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.164876645091468e-06,
+      "loss": 0.8679,
+      "step": 1917
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.159956659762645e-06,
+      "loss": 0.8878,
+      "step": 1918
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.155036519392156e-06,
+      "loss": 0.7452,
+      "step": 1919
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.150116228748959e-06,
+      "loss": 0.9376,
+      "step": 1920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.1451957926021586e-06,
+      "loss": 0.7917,
+      "step": 1921
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.140275215721001e-06,
+      "loss": 0.8638,
+      "step": 1922
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.135354502874869e-06,
+      "loss": 0.7773,
+      "step": 1923
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.130433658833276e-06,
+      "loss": 0.7309,
+      "step": 1924
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.125512688365864e-06,
+      "loss": 0.6998,
+      "step": 1925
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.120591596242394e-06,
+      "loss": 0.7275,
+      "step": 1926
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.11567038723275e-06,
+      "loss": 0.7195,
+      "step": 1927
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.110749066106926e-06,
+      "loss": 0.9193,
+      "step": 1928
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.105827637635025e-06,
+      "loss": 0.7274,
+      "step": 1929
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.100906106587257e-06,
+      "loss": 0.76,
+      "step": 1930
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.095984477733927e-06,
+      "loss": 0.7759,
+      "step": 1931
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.091062755845436e-06,
+      "loss": 0.7123,
+      "step": 1932
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.086140945692278e-06,
+      "loss": 0.7224,
+      "step": 1933
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.08121905204503e-06,
+      "loss": 0.8518,
+      "step": 1934
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.07629707967435e-06,
+      "loss": 0.6891,
+      "step": 1935
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.071375033350973e-06,
+      "loss": 0.9734,
+      "step": 1936
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.0664529178457055e-06,
+      "loss": 0.9348,
+      "step": 1937
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.06153073792942e-06,
+      "loss": 0.7675,
+      "step": 1938
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.056608498373054e-06,
+      "loss": 0.7238,
+      "step": 1939
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.0516862039476e-06,
+      "loss": 0.6863,
+      "step": 1940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.046763859424106e-06,
+      "loss": 0.8304,
+      "step": 1941
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.041841469573668e-06,
+      "loss": 0.8076,
+      "step": 1942
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.036919039167424e-06,
+      "loss": 0.9063,
+      "step": 1943
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.031996572976553e-06,
+      "loss": 0.7661,
+      "step": 1944
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.027074075772268e-06,
+      "loss": 0.8697,
+      "step": 1945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.022151552325813e-06,
+      "loss": 0.9263,
+      "step": 1946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.0172290074084595e-06,
+      "loss": 0.6941,
+      "step": 1947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.012306445791493e-06,
+      "loss": 0.7933,
+      "step": 1948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.007383872246222e-06,
+      "loss": 0.7688,
+      "step": 1949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.002461291543962e-06,
+      "loss": 0.6791,
+      "step": 1950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9975387084560385e-06,
+      "loss": 0.8612,
+      "step": 1951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9926161277537806e-06,
+      "loss": 0.871,
+      "step": 1952
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.987693554208507e-06,
+      "loss": 0.9419,
+      "step": 1953
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.982770992591542e-06,
+      "loss": 0.7266,
+      "step": 1954
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9778484476741875e-06,
+      "loss": 0.7125,
+      "step": 1955
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.972925924227733e-06,
+      "loss": 0.7455,
+      "step": 1956
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.968003427023449e-06,
+      "loss": 0.9797,
+      "step": 1957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.963080960832577e-06,
+      "loss": 0.8143,
+      "step": 1958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.958158530426333e-06,
+      "loss": 0.8645,
+      "step": 1959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.953236140575896e-06,
+      "loss": 0.7242,
+      "step": 1960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9483137960524e-06,
+      "loss": 0.6796,
+      "step": 1961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.943391501626948e-06,
+      "loss": 0.9908,
+      "step": 1962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9384692620705815e-06,
+      "loss": 0.7056,
+      "step": 1963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.933547082154298e-06,
+      "loss": 0.7153,
+      "step": 1964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.928624966649028e-06,
+      "loss": 0.8229,
+      "step": 1965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9237029203256515e-06,
+      "loss": 0.8079,
+      "step": 1966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.918780947954972e-06,
+      "loss": 0.8274,
+      "step": 1967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.913859054307723e-06,
+      "loss": 0.8775,
+      "step": 1968
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.908937244154565e-06,
+      "loss": 0.918,
+      "step": 1969
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.904015522266074e-06,
+      "loss": 0.7394,
+      "step": 1970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.899093893412745e-06,
+      "loss": 0.8645,
+      "step": 1971
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.894172362364976e-06,
+      "loss": 0.6384,
+      "step": 1972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.889250933893075e-06,
+      "loss": 0.8157,
+      "step": 1973
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.884329612767252e-06,
+      "loss": 0.8573,
+      "step": 1974
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.879408403757607e-06,
+      "loss": 1.074,
+      "step": 1975
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.874487311634139e-06,
+      "loss": 0.8624,
+      "step": 1976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.869566341166725e-06,
+      "loss": 0.8923,
+      "step": 1977
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8646454971251315e-06,
+      "loss": 0.7685,
+      "step": 1978
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.859724784279e-06,
+      "loss": 0.7983,
+      "step": 1979
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.854804207397843e-06,
+      "loss": 0.7117,
+      "step": 1980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.849883771251042e-06,
+      "loss": 0.878,
+      "step": 1981
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.844963480607845e-06,
+      "loss": 0.7576,
+      "step": 1982
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.840043340237357e-06,
+      "loss": 0.7217,
+      "step": 1983
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.835123354908535e-06,
+      "loss": 0.8769,
+      "step": 1984
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.830203529390186e-06,
+      "loss": 0.7457,
+      "step": 1985
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.825283868450969e-06,
+      "loss": 0.7339,
+      "step": 1986
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.820364376859375e-06,
+      "loss": 0.7613,
+      "step": 1987
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.815445059383736e-06,
+      "loss": 0.7609,
+      "step": 1988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.81052592079221e-06,
+      "loss": 0.7996,
+      "step": 1989
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.80560696585279e-06,
+      "loss": 0.8303,
+      "step": 1990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8006881993332834e-06,
+      "loss": 0.7902,
+      "step": 1991
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.795769626001317e-06,
+      "loss": 0.6385,
+      "step": 1992
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.790851250624331e-06,
+      "loss": 0.657,
+      "step": 1993
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7859330779695756e-06,
+      "loss": 0.8015,
+      "step": 1994
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.781015112804102e-06,
+      "loss": 0.8435,
+      "step": 1995
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.776097359894762e-06,
+      "loss": 0.8412,
+      "step": 1996
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.771179824008195e-06,
+      "loss": 0.7059,
+      "step": 1997
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7662625099108415e-06,
+      "loss": 0.6819,
+      "step": 1998
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7613454223689196e-06,
+      "loss": 0.6898,
+      "step": 1999
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.756428566148429e-06,
+      "loss": 0.7895,
+      "step": 2000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.751511946015145e-06,
+      "loss": 0.8334,
+      "step": 2001
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.746595566734616e-06,
+      "loss": 0.8486,
+      "step": 2002
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.741679433072155e-06,
+      "loss": 0.7667,
+      "step": 2003
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.736763549792836e-06,
+      "loss": 0.7099,
+      "step": 2004
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.731847921661494e-06,
+      "loss": 0.6972,
+      "step": 2005
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.726932553442712e-06,
+      "loss": 0.6979,
+      "step": 2006
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.722017449900827e-06,
+      "loss": 0.7417,
+      "step": 2007
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.717102615799913e-06,
+      "loss": 0.8005,
+      "step": 2008
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.712188055903786e-06,
+      "loss": 0.696,
+      "step": 2009
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7072737749759985e-06,
+      "loss": 1.0669,
+      "step": 2010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.702359777779828e-06,
+      "loss": 0.8315,
+      "step": 2011
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.6974460690782805e-06,
+      "loss": 0.7544,
+      "step": 2012
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.692532653634078e-06,
+      "loss": 0.805,
+      "step": 2013
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.687619536209666e-06,
+      "loss": 0.7368,
+      "step": 2014
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.682706721567194e-06,
+      "loss": 0.8294,
+      "step": 2015
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.677794214468519e-06,
+      "loss": 0.8,
+      "step": 2016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.672882019675203e-06,
+      "loss": 0.7209,
+      "step": 2017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.667970141948503e-06,
+      "loss": 0.875,
+      "step": 2018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.663058586049372e-06,
+      "loss": 0.8397,
+      "step": 2019
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.658147356738445e-06,
+      "loss": 0.9221,
+      "step": 2020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.653236458776043e-06,
+      "loss": 0.9142,
+      "step": 2021
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.6483258969221715e-06,
+      "loss": 0.6518,
+      "step": 2022
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.6434156759365005e-06,
+      "loss": 0.7807,
+      "step": 2023
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.638505800578377e-06,
+      "loss": 0.6669,
+      "step": 2024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.6335962756068095e-06,
+      "loss": 0.7816,
+      "step": 2025
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.628687105780469e-06,
+      "loss": 0.8542,
+      "step": 2026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.623778295857681e-06,
+      "loss": 0.7266,
+      "step": 2027
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.61886985059642e-06,
+      "loss": 0.7574,
+      "step": 2028
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.613961774754312e-06,
+      "loss": 0.6649,
+      "step": 2029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.609054073088621e-06,
+      "loss": 0.9381,
+      "step": 2030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.60414675035625e-06,
+      "loss": 0.7051,
+      "step": 2031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.599239811313735e-06,
+      "loss": 0.9148,
+      "step": 2032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.5943332607172355e-06,
+      "loss": 0.8232,
+      "step": 2033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.5894271033225444e-06,
+      "loss": 0.6826,
+      "step": 2034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.584521343885063e-06,
+      "loss": 0.9008,
+      "step": 2035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.579615987159814e-06,
+      "loss": 0.8844,
+      "step": 2036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.574711037901422e-06,
+      "loss": 0.8457,
+      "step": 2037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.56980650086413e-06,
+      "loss": 0.7442,
+      "step": 2038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.564902380801767e-06,
+      "loss": 0.6913,
+      "step": 2039
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.559998682467766e-06,
+      "loss": 0.8372,
+      "step": 2040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.555095410615148e-06,
+      "loss": 0.8098,
+      "step": 2041
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.550192569996523e-06,
+      "loss": 0.9364,
+      "step": 2042
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.545290165364084e-06,
+      "loss": 0.7616,
+      "step": 2043
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5403882014695955e-06,
+      "loss": 0.798,
+      "step": 2044
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.535486683064398e-06,
+      "loss": 0.7986,
+      "step": 2045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.530585614899406e-06,
+      "loss": 0.8262,
+      "step": 2046
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.525685001725088e-06,
+      "loss": 0.7886,
+      "step": 2047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5207848482914785e-06,
+      "loss": 0.6813,
+      "step": 2048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5158851593481605e-06,
+      "loss": 0.6928,
+      "step": 2049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.510985939644273e-06,
+      "loss": 0.6996,
+      "step": 2050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.5060871939284976e-06,
+      "loss": 0.8519,
+      "step": 2051
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.501188926949053e-06,
+      "loss": 0.8304,
+      "step": 2052
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.496291143453699e-06,
+      "loss": 0.7084,
+      "step": 2053
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.491393848189725e-06,
+      "loss": 0.6239,
+      "step": 2054
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.486497045903947e-06,
+      "loss": 0.5937,
+      "step": 2055
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.481600741342702e-06,
+      "loss": 0.8288,
+      "step": 2056
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.476704939251845e-06,
+      "loss": 0.7977,
+      "step": 2057
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4718096443767475e-06,
+      "loss": 0.7286,
+      "step": 2058
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.466914861462285e-06,
+      "loss": 0.7138,
+      "step": 2059
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.462020595252838e-06,
+      "loss": 0.9023,
+      "step": 2060
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.457126850492285e-06,
+      "loss": 0.7984,
+      "step": 2061
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.452233631924004e-06,
+      "loss": 0.8656,
+      "step": 2062
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.447340944290858e-06,
+      "loss": 0.8643,
+      "step": 2063
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.442448792335195e-06,
+      "loss": 0.8157,
+      "step": 2064
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.437557180798847e-06,
+      "loss": 0.7795,
+      "step": 2065
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4326661144231205e-06,
+      "loss": 0.7586,
+      "step": 2066
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.427775597948797e-06,
+      "loss": 0.7164,
+      "step": 2067
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4228856361161185e-06,
+      "loss": 0.8624,
+      "step": 2068
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.417996233664792e-06,
+      "loss": 0.7873,
+      "step": 2069
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.413107395333987e-06,
+      "loss": 0.6868,
+      "step": 2070
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.408219125862318e-06,
+      "loss": 0.9004,
+      "step": 2071
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.403331429987857e-06,
+      "loss": 0.7187,
+      "step": 2072
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.39844431244811e-06,
+      "loss": 0.672,
+      "step": 2073
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.3935577779800334e-06,
+      "loss": 0.7668,
+      "step": 2074
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.388671831320011e-06,
+      "loss": 0.753,
+      "step": 2075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.383786477203857e-06,
+      "loss": 0.782,
+      "step": 2076
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.378901720366815e-06,
+      "loss": 0.8197,
+      "step": 2077
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.374017565543547e-06,
+      "loss": 0.902,
+      "step": 2078
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.369134017468134e-06,
+      "loss": 0.8218,
+      "step": 2079
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3642510808740654e-06,
+      "loss": 0.7275,
+      "step": 2080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.359368760494239e-06,
+      "loss": 0.8606,
+      "step": 2081
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.354487061060958e-06,
+      "loss": 0.6429,
+      "step": 2082
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.34960598730592e-06,
+      "loss": 0.7135,
+      "step": 2083
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.3447255439602195e-06,
+      "loss": 0.7356,
+      "step": 2084
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.339845735754335e-06,
+      "loss": 0.8192,
+      "step": 2085
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.334966567418137e-06,
+      "loss": 0.903,
+      "step": 2086
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.33008804368087e-06,
+      "loss": 0.7743,
+      "step": 2087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.325210169271154e-06,
+      "loss": 0.809,
+      "step": 2088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3203329489169805e-06,
+      "loss": 0.7532,
+      "step": 2089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.31545638734571e-06,
+      "loss": 1.1286,
+      "step": 2090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.310580489284061e-06,
+      "loss": 0.6514,
+      "step": 2091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.3057052594581074e-06,
+      "loss": 0.9059,
+      "step": 2092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.300830702593278e-06,
+      "loss": 0.7783,
+      "step": 2093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.295956823414352e-06,
+      "loss": 0.9092,
+      "step": 2094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.291083626645447e-06,
+      "loss": 0.7881,
+      "step": 2095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.286211117010021e-06,
+      "loss": 0.8844,
+      "step": 2096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.281339299230864e-06,
+      "loss": 0.7465,
+      "step": 2097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.276468178030101e-06,
+      "loss": 0.7446,
+      "step": 2098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.271597758129176e-06,
+      "loss": 0.6915,
+      "step": 2099
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.2667280442488556e-06,
+      "loss": 0.8258,
+      "step": 2100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.261859041109221e-06,
+      "loss": 0.735,
+      "step": 2101
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.256990753429666e-06,
+      "loss": 0.8007,
+      "step": 2102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.252123185928891e-06,
+      "loss": 0.8014,
+      "step": 2103
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.247256343324898e-06,
+      "loss": 0.9036,
+      "step": 2104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.242390230334983e-06,
+      "loss": 0.808,
+      "step": 2105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.237524851675741e-06,
+      "loss": 0.7131,
+      "step": 2106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.23266021206305e-06,
+      "loss": 0.731,
+      "step": 2107
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.227796316212075e-06,
+      "loss": 0.7918,
+      "step": 2108
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.222933168837257e-06,
+      "loss": 0.7814,
+      "step": 2109
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.2180707746523166e-06,
+      "loss": 0.6984,
+      "step": 2110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.213209138370239e-06,
+      "loss": 0.7249,
+      "step": 2111
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.208348264703276e-06,
+      "loss": 0.8297,
+      "step": 2112
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.203488158362941e-06,
+      "loss": 0.8099,
+      "step": 2113
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.198628824060006e-06,
+      "loss": 0.8136,
+      "step": 2114
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.19377026650449e-06,
+      "loss": 0.8095,
+      "step": 2115
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1889124904056644e-06,
+      "loss": 0.7331,
+      "step": 2116
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.184055500472036e-06,
+      "loss": 0.8271,
+      "step": 2117
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.179199301411358e-06,
+      "loss": 0.678,
+      "step": 2118
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.17434389793061e-06,
+      "loss": 0.7908,
+      "step": 2119
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1694892947360055e-06,
+      "loss": 0.7468,
+      "step": 2120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.164635496532977e-06,
+      "loss": 0.7441,
+      "step": 2121
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.159782508026184e-06,
+      "loss": 0.7966,
+      "step": 2122
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.154930333919496e-06,
+      "loss": 0.8246,
+      "step": 2123
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1500789789159925e-06,
+      "loss": 0.7562,
+      "step": 2124
+    },
+    {
+      "epoch": 1.8,
+      "eval_loss": 0.9790877103805542,
+      "eval_runtime": 267.8424,
+      "eval_samples_per_second": 3.73,
+      "eval_steps_per_second": 0.467,
+      "step": 2124
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1452284477179605e-06,
+      "loss": 0.8464,
+      "step": 2125
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.14037874502689e-06,
+      "loss": 0.939,
+      "step": 2126
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.135529875543469e-06,
+      "loss": 0.8583,
+      "step": 2127
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.130681843967572e-06,
+      "loss": 0.9246,
+      "step": 2128
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.125834654998264e-06,
+      "loss": 0.9086,
+      "step": 2129
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.120988313333799e-06,
+      "loss": 0.837,
+      "step": 2130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.116142823671602e-06,
+      "loss": 0.9611,
+      "step": 2131
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.111298190708276e-06,
+      "loss": 0.7007,
+      "step": 2132
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1064544191395905e-06,
+      "loss": 0.6329,
+      "step": 2133
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.101611513660486e-06,
+      "loss": 0.8608,
+      "step": 2134
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0967694789650584e-06,
+      "loss": 0.6047,
+      "step": 2135
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.09192831974656e-06,
+      "loss": 0.744,
+      "step": 2136
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.087088040697396e-06,
+      "loss": 0.8451,
+      "step": 2137
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.082248646509119e-06,
+      "loss": 0.7718,
+      "step": 2138
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.077410141872422e-06,
+      "loss": 0.7497,
+      "step": 2139
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.072572531477139e-06,
+      "loss": 0.7965,
+      "step": 2140
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.06773582001223e-06,
+      "loss": 0.7959,
+      "step": 2141
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.062900012165794e-06,
+      "loss": 0.8721,
+      "step": 2142
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.058065112625046e-06,
+      "loss": 0.8349,
+      "step": 2143
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.053231126076325e-06,
+      "loss": 0.6482,
+      "step": 2144
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.04839805720508e-06,
+      "loss": 0.8474,
+      "step": 2145
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.04356591069588e-06,
+      "loss": 0.6902,
+      "step": 2146
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.038734691232389e-06,
+      "loss": 0.6934,
+      "step": 2147
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.03390440349738e-06,
+      "loss": 0.8091,
+      "step": 2148
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.029075052172717e-06,
+      "loss": 0.6923,
+      "step": 2149
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.024246641939364e-06,
+      "loss": 0.9406,
+      "step": 2150
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.019419177477367e-06,
+      "loss": 0.9114,
+      "step": 2151
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.014592663465857e-06,
+      "loss": 0.7233,
+      "step": 2152
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.00976710458304e-06,
+      "loss": 0.7368,
+      "step": 2153
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.004942505506205e-06,
+      "loss": 0.8715,
+      "step": 2154
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.000118870911703e-06,
+      "loss": 0.7537,
+      "step": 2155
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.995296205474954e-06,
+      "loss": 0.8859,
+      "step": 2156
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9904745138704334e-06,
+      "loss": 0.7829,
+      "step": 2157
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9856538007716825e-06,
+      "loss": 1.0376,
+      "step": 2158
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9808340708512836e-06,
+      "loss": 0.7795,
+      "step": 2159
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.976015328780872e-06,
+      "loss": 0.798,
+      "step": 2160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.971197579231123e-06,
+      "loss": 0.8752,
+      "step": 2161
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.966380826871754e-06,
+      "loss": 0.6738,
+      "step": 2162
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.961565076371512e-06,
+      "loss": 0.8686,
+      "step": 2163
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.956750332398172e-06,
+      "loss": 0.8698,
+      "step": 2164
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9519365996185345e-06,
+      "loss": 0.8231,
+      "step": 2165
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9471238826984234e-06,
+      "loss": 0.8075,
+      "step": 2166
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.942312186302674e-06,
+      "loss": 0.8384,
+      "step": 2167
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.937501515095133e-06,
+      "loss": 0.7256,
+      "step": 2168
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9326918737386535e-06,
+      "loss": 0.8185,
+      "step": 2169
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.927883266895094e-06,
+      "loss": 0.8062,
+      "step": 2170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.923075699225305e-06,
+      "loss": 0.7412,
+      "step": 2171
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.918269175389129e-06,
+      "loss": 0.7885,
+      "step": 2172
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.913463700045403e-06,
+      "loss": 0.7449,
+      "step": 2173
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.908659277851946e-06,
+      "loss": 0.7471,
+      "step": 2174
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.903855913465551e-06,
+      "loss": 0.7621,
+      "step": 2175
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.899053611541992e-06,
+      "loss": 0.9667,
+      "step": 2176
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.894252376736007e-06,
+      "loss": 0.8331,
+      "step": 2177
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.889452213701306e-06,
+      "loss": 0.6438,
+      "step": 2178
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8846531270905565e-06,
+      "loss": 0.6317,
+      "step": 2179
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8798551215553835e-06,
+      "loss": 0.6799,
+      "step": 2180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.875058201746363e-06,
+      "loss": 0.7799,
+      "step": 2181
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.870262372313022e-06,
+      "loss": 0.7409,
+      "step": 2182
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.865467637903828e-06,
+      "loss": 0.6716,
+      "step": 2183
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.860674003166185e-06,
+      "loss": 0.572,
+      "step": 2184
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.855881472746434e-06,
+      "loss": 0.6429,
+      "step": 2185
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.851090051289848e-06,
+      "loss": 0.6452,
+      "step": 2186
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846299743440621e-06,
+      "loss": 0.7806,
+      "step": 2187
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.841510553841869e-06,
+      "loss": 0.9336,
+      "step": 2188
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.83672248713562e-06,
+      "loss": 0.7547,
+      "step": 2189
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.831935547962824e-06,
+      "loss": 0.7651,
+      "step": 2190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.827149740963329e-06,
+      "loss": 0.6874,
+      "step": 2191
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.822365070775888e-06,
+      "loss": 0.797,
+      "step": 2192
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.817581542038151e-06,
+      "loss": 0.873,
+      "step": 2193
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8127991593866665e-06,
+      "loss": 0.9031,
+      "step": 2194
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.808017927456868e-06,
+      "loss": 0.7952,
+      "step": 2195
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.803237850883072e-06,
+      "loss": 0.8635,
+      "step": 2196
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.798458934298478e-06,
+      "loss": 0.7873,
+      "step": 2197
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7936811823351637e-06,
+      "loss": 0.7987,
+      "step": 2198
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7889045996240732e-06,
+      "loss": 0.6353,
+      "step": 2199
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7841291907950185e-06,
+      "loss": 0.7678,
+      "step": 2200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7793549604766727e-06,
+      "loss": 1.0851,
+      "step": 2201
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7745819132965722e-06,
+      "loss": 0.8878,
+      "step": 2202
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7698100538810987e-06,
+      "loss": 0.6991,
+      "step": 2203
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.76503938685549e-06,
+      "loss": 0.8541,
+      "step": 2204
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.760269916843821e-06,
+      "loss": 0.735,
+      "step": 2205
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.755501648469014e-06,
+      "loss": 0.7022,
+      "step": 2206
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7507345863528204e-06,
+      "loss": 0.7472,
+      "step": 2207
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7459687351158245e-06,
+      "loss": 0.7403,
+      "step": 2208
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7412040993774358e-06,
+      "loss": 0.886,
+      "step": 2209
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7364406837558916e-06,
+      "loss": 0.7492,
+      "step": 2210
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.731678492868238e-06,
+      "loss": 0.7361,
+      "step": 2211
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7269175313303403e-06,
+      "loss": 0.6478,
+      "step": 2212
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.722157803756865e-06,
+      "loss": 0.8438,
+      "step": 2213
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.717399314761294e-06,
+      "loss": 0.9706,
+      "step": 2214
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7126420689558974e-06,
+      "loss": 0.662,
+      "step": 2215
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.707886070951747e-06,
+      "loss": 0.6631,
+      "step": 2216
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.703131325358701e-06,
+      "loss": 0.8389,
+      "step": 2217
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6983778367854097e-06,
+      "loss": 0.9226,
+      "step": 2218
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.693625609839299e-06,
+      "loss": 0.6656,
+      "step": 2219
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6888746491265715e-06,
+      "loss": 0.6147,
+      "step": 2220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6841249592522066e-06,
+      "loss": 0.9657,
+      "step": 2221
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6793765448199536e-06,
+      "loss": 0.6533,
+      "step": 2222
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6746294104323194e-06,
+      "loss": 0.8813,
+      "step": 2223
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6698835606905736e-06,
+      "loss": 0.812,
+      "step": 2224
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6651390001947374e-06,
+      "loss": 0.7552,
+      "step": 2225
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.660395733543589e-06,
+      "loss": 0.7119,
+      "step": 2226
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.655653765334647e-06,
+      "loss": 0.7788,
+      "step": 2227
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.650913100164173e-06,
+      "loss": 0.8114,
+      "step": 2228
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.6461737426271637e-06,
+      "loss": 0.8645,
+      "step": 2229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6414356973173546e-06,
+      "loss": 0.8545,
+      "step": 2230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6366989688272025e-06,
+      "loss": 0.7357,
+      "step": 2231
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6319635617478894e-06,
+      "loss": 0.7665,
+      "step": 2232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.627229480669317e-06,
+      "loss": 0.7787,
+      "step": 2233
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6224967301801057e-06,
+      "loss": 0.7685,
+      "step": 2234
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.61776531486758e-06,
+      "loss": 0.6205,
+      "step": 2235
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6130352393177737e-06,
+      "loss": 0.8765,
+      "step": 2236
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6083065081154177e-06,
+      "loss": 0.6735,
+      "step": 2237
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.603579125843949e-06,
+      "loss": 0.658,
+      "step": 2238
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.598853097085488e-06,
+      "loss": 0.8122,
+      "step": 2239
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5941284264208483e-06,
+      "loss": 0.6748,
+      "step": 2240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.589405118429522e-06,
+      "loss": 0.6891,
+      "step": 2241
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.584683177689689e-06,
+      "loss": 0.841,
+      "step": 2242
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.579962608778196e-06,
+      "loss": 0.8261,
+      "step": 2243
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.575243416270563e-06,
+      "loss": 0.8378,
+      "step": 2244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5705256047409735e-06,
+      "loss": 0.7375,
+      "step": 2245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5658091787622795e-06,
+      "loss": 0.73,
+      "step": 2246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5610941429059833e-06,
+      "loss": 0.7639,
+      "step": 2247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5563805017422417e-06,
+      "loss": 0.6742,
+      "step": 2248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5516682598398556e-06,
+      "loss": 0.8046,
+      "step": 2249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.54695742176628e-06,
+      "loss": 0.7474,
+      "step": 2250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5422479920875995e-06,
+      "loss": 0.7192,
+      "step": 2251
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.537539975368538e-06,
+      "loss": 0.9549,
+      "step": 2252
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.532833376172447e-06,
+      "loss": 0.709,
+      "step": 2253
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.528128199061309e-06,
+      "loss": 0.6875,
+      "step": 2254
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5234244485957235e-06,
+      "loss": 0.8864,
+      "step": 2255
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5187221293349083e-06,
+      "loss": 0.7799,
+      "step": 2256
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.514021245836693e-06,
+      "loss": 0.7097,
+      "step": 2257
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5093218026575204e-06,
+      "loss": 0.7867,
+      "step": 2258
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.504623804352432e-06,
+      "loss": 0.8661,
+      "step": 2259
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.4999272554750707e-06,
+      "loss": 0.731,
+      "step": 2260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.4952321605776724e-06,
+      "loss": 0.721,
+      "step": 2261
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.490538524211069e-06,
+      "loss": 0.7176,
+      "step": 2262
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.485846350924673e-06,
+      "loss": 0.8828,
+      "step": 2263
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.481155645266483e-06,
+      "loss": 0.6256,
+      "step": 2264
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.476466411783069e-06,
+      "loss": 0.637,
+      "step": 2265
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4717786550195852e-06,
+      "loss": 0.9842,
+      "step": 2266
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.467092379519743e-06,
+      "loss": 0.7177,
+      "step": 2267
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.462407589825822e-06,
+      "loss": 0.9481,
+      "step": 2268
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4577242904786623e-06,
+      "loss": 0.6375,
+      "step": 2269
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.453042486017663e-06,
+      "loss": 0.6376,
+      "step": 2270
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4483621809807688e-06,
+      "loss": 0.7718,
+      "step": 2271
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4436833799044712e-06,
+      "loss": 0.9631,
+      "step": 2272
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4390060873238052e-06,
+      "loss": 0.6721,
+      "step": 2273
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4343303077723477e-06,
+      "loss": 0.7612,
+      "step": 2274
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.429656045782202e-06,
+      "loss": 0.7757,
+      "step": 2275
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4249833058840065e-06,
+      "loss": 0.7255,
+      "step": 2276
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.4203120926069177e-06,
+      "loss": 0.7935,
+      "step": 2277
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.4156424104786217e-06,
+      "loss": 0.9109,
+      "step": 2278
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.4109742640253126e-06,
+      "loss": 0.8761,
+      "step": 2279
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.406307657771697e-06,
+      "loss": 0.6887,
+      "step": 2280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.401642596240991e-06,
+      "loss": 0.6664,
+      "step": 2281
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3969790839549156e-06,
+      "loss": 0.7386,
+      "step": 2282
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.392317125433684e-06,
+      "loss": 0.7188,
+      "step": 2283
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3876567251960083e-06,
+      "loss": 1.0118,
+      "step": 2284
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3829978877590856e-06,
+      "loss": 0.8451,
+      "step": 2285
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.378340617638605e-06,
+      "loss": 0.7262,
+      "step": 2286
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.37368491934873e-06,
+      "loss": 0.6939,
+      "step": 2287
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3690307974021052e-06,
+      "loss": 0.7017,
+      "step": 2288
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3643782563098415e-06,
+      "loss": 0.7523,
+      "step": 2289
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3597273005815255e-06,
+      "loss": 0.7468,
+      "step": 2290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3550779347252005e-06,
+      "loss": 0.9166,
+      "step": 2291
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3504301632473713e-06,
+      "loss": 0.7485,
+      "step": 2292
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3457839906529967e-06,
+      "loss": 0.8642,
+      "step": 2293
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3411394214454883e-06,
+      "loss": 0.7097,
+      "step": 2294
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.336496460126701e-06,
+      "loss": 0.7435,
+      "step": 2295
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.33185511119693e-06,
+      "loss": 0.8514,
+      "step": 2296
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.327215379154909e-06,
+      "loss": 0.9581,
+      "step": 2297
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3225772684978062e-06,
+      "loss": 0.6315,
+      "step": 2298
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.3179407837212183e-06,
+      "loss": 0.6965,
+      "step": 2299
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.313305929319164e-06,
+      "loss": 0.7661,
+      "step": 2300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3086727097840787e-06,
+      "loss": 0.7951,
+      "step": 2301
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.304041129606822e-06,
+      "loss": 0.8233,
+      "step": 2302
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.299411193276657e-06,
+      "loss": 0.782,
+      "step": 2303
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.294782905281255e-06,
+      "loss": 0.6883,
+      "step": 2304
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2901562701066904e-06,
+      "loss": 1.0414,
+      "step": 2305
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2855312922374383e-06,
+      "loss": 0.8138,
+      "step": 2306
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2809079761563635e-06,
+      "loss": 0.7987,
+      "step": 2307
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.276286326344721e-06,
+      "loss": 0.8027,
+      "step": 2308
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2716663472821487e-06,
+      "loss": 0.7035,
+      "step": 2309
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2670480434466702e-06,
+      "loss": 0.6987,
+      "step": 2310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2624314193146834e-06,
+      "loss": 0.717,
+      "step": 2311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2578164793609556e-06,
+      "loss": 0.6978,
+      "step": 2312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.253203228058622e-06,
+      "loss": 0.8655,
+      "step": 2313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2485916698791866e-06,
+      "loss": 0.7224,
+      "step": 2314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.243981809292505e-06,
+      "loss": 0.7275,
+      "step": 2315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2393736507667895e-06,
+      "loss": 0.6831,
+      "step": 2316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2347671987686046e-06,
+      "loss": 0.7916,
+      "step": 2317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.230162457762862e-06,
+      "loss": 0.8631,
+      "step": 2318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2255594322128114e-06,
+      "loss": 0.7407,
+      "step": 2319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2209581265800406e-06,
+      "loss": 0.7528,
+      "step": 2320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2163585453244684e-06,
+      "loss": 0.929,
+      "step": 2321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.2117606929043475e-06,
+      "loss": 0.6439,
+      "step": 2322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.207164573776253e-06,
+      "loss": 0.832,
+      "step": 2323
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.202570192395077e-06,
+      "loss": 0.6961,
+      "step": 2324
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1979775532140278e-06,
+      "loss": 0.81,
+      "step": 2325
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1933866606846297e-06,
+      "loss": 0.7244,
+      "step": 2326
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.18879751925671e-06,
+      "loss": 0.7857,
+      "step": 2327
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.184210133378397e-06,
+      "loss": 0.8415,
+      "step": 2328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.179624507496121e-06,
+      "loss": 0.9238,
+      "step": 2329
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.175040646054607e-06,
+      "loss": 0.7531,
+      "step": 2330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.170458553496867e-06,
+      "loss": 0.7738,
+      "step": 2331
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.165878234264199e-06,
+      "loss": 0.8382,
+      "step": 2332
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1612996927961803e-06,
+      "loss": 0.6521,
+      "step": 2333
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1567229335306715e-06,
+      "loss": 0.5927,
+      "step": 2334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.1521479609038013e-06,
+      "loss": 0.777,
+      "step": 2335
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1475747793499655e-06,
+      "loss": 0.7751,
+      "step": 2336
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1430033933018234e-06,
+      "loss": 0.6752,
+      "step": 2337
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1384338071902997e-06,
+      "loss": 0.7721,
+      "step": 2338
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1338660254445696e-06,
+      "loss": 0.6486,
+      "step": 2339
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1293000524920593e-06,
+      "loss": 0.8591,
+      "step": 2340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.124735892758442e-06,
+      "loss": 0.954,
+      "step": 2341
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1201735506676383e-06,
+      "loss": 0.8641,
+      "step": 2342
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.115613030641801e-06,
+      "loss": 0.8058,
+      "step": 2343
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.111054337101318e-06,
+      "loss": 0.8496,
+      "step": 2344
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.106497474464807e-06,
+      "loss": 0.7711,
+      "step": 2345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1019424471491145e-06,
+      "loss": 0.7619,
+      "step": 2346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0973892595693043e-06,
+      "loss": 0.8569,
+      "step": 2347
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0928379161386578e-06,
+      "loss": 0.7531,
+      "step": 2348
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0882884212686682e-06,
+      "loss": 0.8016,
+      "step": 2349
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.08374077936904e-06,
+      "loss": 0.8162,
+      "step": 2350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0791949948476783e-06,
+      "loss": 0.6572,
+      "step": 2351
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0746510721106894e-06,
+      "loss": 0.777,
+      "step": 2352
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0701090155623725e-06,
+      "loss": 0.8364,
+      "step": 2353
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0655688296052248e-06,
+      "loss": 0.7943,
+      "step": 2354
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0610305186399236e-06,
+      "loss": 0.7925,
+      "step": 2355
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.056494087065331e-06,
+      "loss": 0.7168,
+      "step": 2356
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.051959539278485e-06,
+      "loss": 0.8624,
+      "step": 2357
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.047426879674604e-06,
+      "loss": 0.6828,
+      "step": 2358
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0428961126470726e-06,
+      "loss": 0.9179,
+      "step": 2359
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0383672425874385e-06,
+      "loss": 0.6838,
+      "step": 2360
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.982319712638855,
+      "eval_runtime": 268.2202,
+      "eval_samples_per_second": 3.725,
+      "eval_steps_per_second": 0.466,
+      "step": 2360
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0338402738854115e-06,
+      "loss": 0.6073,
+      "step": 2361
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0293152109288647e-06,
+      "loss": 0.852,
+      "step": 2362
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.024792058103817e-06,
+      "loss": 0.7345,
+      "step": 2363
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0202708197944375e-06,
+      "loss": 0.6796,
+      "step": 2364
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0157515003830396e-06,
+      "loss": 0.7259,
+      "step": 2365
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0112341042500802e-06,
+      "loss": 0.4942,
+      "step": 2366
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0067186357741486e-06,
+      "loss": 0.4838,
+      "step": 2367
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.002205099331965e-06,
+      "loss": 0.5235,
+      "step": 2368
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9976934992983775e-06,
+      "loss": 0.4722,
+      "step": 2369
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9931838400463596e-06,
+      "loss": 0.4696,
+      "step": 2370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.988676125947003e-06,
+      "loss": 0.4266,
+      "step": 2371
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9841703613695127e-06,
+      "loss": 0.5952,
+      "step": 2372
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9796665506812017e-06,
+      "loss": 0.5111,
+      "step": 2373
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.975164698247496e-06,
+      "loss": 0.4388,
+      "step": 2374
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.970664808431917e-06,
+      "loss": 0.4789,
+      "step": 2375
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9661668855960864e-06,
+      "loss": 0.4541,
+      "step": 2376
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9616709340997184e-06,
+      "loss": 0.4971,
+      "step": 2377
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9571769583006203e-06,
+      "loss": 0.445,
+      "step": 2378
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9526849625546804e-06,
+      "loss": 0.5203,
+      "step": 2379
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.948194951215868e-06,
+      "loss": 0.4306,
+      "step": 2380
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9437069286362272e-06,
+      "loss": 0.3698,
+      "step": 2381
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9392208991658807e-06,
+      "loss": 0.5655,
+      "step": 2382
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9347368671530153e-06,
+      "loss": 0.5727,
+      "step": 2383
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.930254836943881e-06,
+      "loss": 0.6206,
+      "step": 2384
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9257748128827867e-06,
+      "loss": 0.5136,
+      "step": 2385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9212967993121035e-06,
+      "loss": 0.4342,
+      "step": 2386
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.916820800572244e-06,
+      "loss": 0.5279,
+      "step": 2387
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.912346821001677e-06,
+      "loss": 0.4527,
+      "step": 2388
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9078748649369036e-06,
+      "loss": 0.5649,
+      "step": 2389
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9034049367124785e-06,
+      "loss": 0.6081,
+      "step": 2390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.898937040660975e-06,
+      "loss": 0.5159,
+      "step": 2391
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8944711811130077e-06,
+      "loss": 0.5095,
+      "step": 2392
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8900073623972132e-06,
+      "loss": 0.4498,
+      "step": 2393
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.8855455888402494e-06,
+      "loss": 0.3825,
+      "step": 2394
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.881085864766795e-06,
+      "loss": 0.5049,
+      "step": 2395
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8766281944995365e-06,
+      "loss": 0.3998,
+      "step": 2396
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.872172582359175e-06,
+      "loss": 0.4946,
+      "step": 2397
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8677190326644144e-06,
+      "loss": 0.4572,
+      "step": 2398
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8632675497319607e-06,
+      "loss": 0.5926,
+      "step": 2399
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.858818137876518e-06,
+      "loss": 0.5429,
+      "step": 2400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.854370801410775e-06,
+      "loss": 0.629,
+      "step": 2401
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.849925544645422e-06,
+      "loss": 0.4686,
+      "step": 2402
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.845482371889121e-06,
+      "loss": 0.5333,
+      "step": 2403
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.841041287448521e-06,
+      "loss": 0.6221,
+      "step": 2404
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.8366022956282452e-06,
+      "loss": 0.4829,
+      "step": 2405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.832165400730888e-06,
+      "loss": 0.5005,
+      "step": 2406
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.8277306070570144e-06,
+      "loss": 0.3978,
+      "step": 2407
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.8232979189051445e-06,
+      "loss": 0.4625,
+      "step": 2408
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.818867340571767e-06,
+      "loss": 0.4416,
+      "step": 2409
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.814438876351321e-06,
+      "loss": 0.488,
+      "step": 2410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.810012530536197e-06,
+      "loss": 0.5092,
+      "step": 2411
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.8055883074167346e-06,
+      "loss": 0.3946,
+      "step": 2412
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.8011662112812068e-06,
+      "loss": 0.4355,
+      "step": 2413
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.796746246415841e-06,
+      "loss": 0.5807,
+      "step": 2414
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.792328417104784e-06,
+      "loss": 0.467,
+      "step": 2415
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.787912727630118e-06,
+      "loss": 0.4458,
+      "step": 2416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7834991822718543e-06,
+      "loss": 0.494,
+      "step": 2417
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7790877853079213e-06,
+      "loss": 0.471,
+      "step": 2418
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.77467854101417e-06,
+      "loss": 0.4008,
+      "step": 2419
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7702714536643564e-06,
+      "loss": 0.4639,
+      "step": 2420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7658665275301544e-06,
+      "loss": 0.6486,
+      "step": 2421
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7614637668811394e-06,
+      "loss": 0.814,
+      "step": 2422
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.757063175984788e-06,
+      "loss": 0.4795,
+      "step": 2423
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.752664759106476e-06,
+      "loss": 0.4972,
+      "step": 2424
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7482685205094647e-06,
+      "loss": 0.5306,
+      "step": 2425
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7438744644549166e-06,
+      "loss": 0.5976,
+      "step": 2426
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7394825952018656e-06,
+      "loss": 0.4318,
+      "step": 2427
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.735092917007234e-06,
+      "loss": 0.5917,
+      "step": 2428
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.730705434125818e-06,
+      "loss": 0.4612,
+      "step": 2429
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7263201508102866e-06,
+      "loss": 0.4996,
+      "step": 2430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.721937071311178e-06,
+      "loss": 0.5347,
+      "step": 2431
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7175561998768884e-06,
+      "loss": 0.7044,
+      "step": 2432
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.713177540753681e-06,
+      "loss": 0.5226,
+      "step": 2433
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7088010981856715e-06,
+      "loss": 0.4318,
+      "step": 2434
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7044268764148267e-06,
+      "loss": 0.4626,
+      "step": 2435
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7000548796809638e-06,
+      "loss": 0.4323,
+      "step": 2436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6956851122217355e-06,
+      "loss": 0.4467,
+      "step": 2437
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6913175782726468e-06,
+      "loss": 0.5866,
+      "step": 2438
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6869522820670246e-06,
+      "loss": 0.4379,
+      "step": 2439
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.682589227836034e-06,
+      "loss": 0.4856,
+      "step": 2440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6782284198086673e-06,
+      "loss": 0.4413,
+      "step": 2441
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6738698622117354e-06,
+      "loss": 0.4613,
+      "step": 2442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6695135592698734e-06,
+      "loss": 0.5584,
+      "step": 2443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.665159515205524e-06,
+      "loss": 0.4887,
+      "step": 2444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.660807734238947e-06,
+      "loss": 0.6187,
+      "step": 2445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.656458220588205e-06,
+      "loss": 0.4535,
+      "step": 2446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.652110978469165e-06,
+      "loss": 0.4547,
+      "step": 2447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.647766012095493e-06,
+      "loss": 0.4685,
+      "step": 2448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.643423325678641e-06,
+      "loss": 0.5158,
+      "step": 2449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.639082923427866e-06,
+      "loss": 0.4374,
+      "step": 2450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.634744809550196e-06,
+      "loss": 0.401,
+      "step": 2451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.63040898825045e-06,
+      "loss": 0.4226,
+      "step": 2452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6260754637312226e-06,
+      "loss": 0.5496,
+      "step": 2453
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6217442401928815e-06,
+      "loss": 0.5075,
+      "step": 2454
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6174153218335674e-06,
+      "loss": 0.4495,
+      "step": 2455
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.613088712849179e-06,
+      "loss": 0.4887,
+      "step": 2456
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.608764417433384e-06,
+      "loss": 0.5973,
+      "step": 2457
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6044424397776037e-06,
+      "loss": 0.3799,
+      "step": 2458
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6001227840710165e-06,
+      "loss": 0.5126,
+      "step": 2459
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.5958054545005485e-06,
+      "loss": 0.4279,
+      "step": 2460
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.5914904552508648e-06,
+      "loss": 0.3824,
+      "step": 2461
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.587177790504385e-06,
+      "loss": 0.5059,
+      "step": 2462
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.5828674644412533e-06,
+      "loss": 0.4747,
+      "step": 2463
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.578559481239354e-06,
+      "loss": 0.6325,
+      "step": 2464
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.5742538450742986e-06,
+      "loss": 0.4965,
+      "step": 2465
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.569950560119424e-06,
+      "loss": 0.4164,
+      "step": 2466
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.5656496305457902e-06,
+      "loss": 0.4713,
+      "step": 2467
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.561351060522168e-06,
+      "loss": 0.4797,
+      "step": 2468
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.5570548542150464e-06,
+      "loss": 0.5426,
+      "step": 2469
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.552761015788623e-06,
+      "loss": 0.5095,
+      "step": 2470
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.5484695494047985e-06,
+      "loss": 0.4645,
+      "step": 2471
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.544180459223177e-06,
+      "loss": 0.4892,
+      "step": 2472
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.5398937494010533e-06,
+      "loss": 0.4177,
+      "step": 2473
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.535609424093426e-06,
+      "loss": 0.4393,
+      "step": 2474
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.5313274874529704e-06,
+      "loss": 0.6141,
+      "step": 2475
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.5270479436300532e-06,
+      "loss": 0.5426,
+      "step": 2476
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.522770796772722e-06,
+      "loss": 0.5648,
+      "step": 2477
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.5184960510266983e-06,
+      "loss": 0.4309,
+      "step": 2478
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.5142237105353797e-06,
+      "loss": 0.4397,
+      "step": 2479
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.5099537794398264e-06,
+      "loss": 0.5545,
+      "step": 2480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.5056862618787705e-06,
+      "loss": 0.4954,
+      "step": 2481
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.5014211619885996e-06,
+      "loss": 0.5728,
+      "step": 2482
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4971584839033613e-06,
+      "loss": 0.4436,
+      "step": 2483
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4928982317547546e-06,
+      "loss": 0.4415,
+      "step": 2484
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4886404096721225e-06,
+      "loss": 0.5367,
+      "step": 2485
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4843850217824633e-06,
+      "loss": 0.5283,
+      "step": 2486
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.4801320722104043e-06,
+      "loss": 0.5674,
+      "step": 2487
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.475881565078217e-06,
+      "loss": 0.5421,
+      "step": 2488
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.471633504505802e-06,
+      "loss": 0.4952,
+      "step": 2489
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4673878946106906e-06,
+      "loss": 0.4871,
+      "step": 2490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.463144739508038e-06,
+      "loss": 0.4724,
+      "step": 2491
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4589040433106182e-06,
+      "loss": 0.3908,
+      "step": 2492
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.454665810128824e-06,
+      "loss": 0.5553,
+      "step": 2493
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4504300440706603e-06,
+      "loss": 0.5109,
+      "step": 2494
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4461967492417406e-06,
+      "loss": 0.3925,
+      "step": 2495
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4419659297452853e-06,
+      "loss": 0.5052,
+      "step": 2496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.437737589682107e-06,
+      "loss": 0.4253,
+      "step": 2497
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.43351173315063e-06,
+      "loss": 0.4464,
+      "step": 2498
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4292883642468575e-06,
+      "loss": 0.4823,
+      "step": 2499
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.4250674870643874e-06,
+      "loss": 0.5384,
+      "step": 2500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.420849105694402e-06,
+      "loss": 0.491,
+      "step": 2501
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4166332242256658e-06,
+      "loss": 0.4867,
+      "step": 2502
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4124198467445193e-06,
+      "loss": 0.5531,
+      "step": 2503
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.4082089773348714e-06,
+      "loss": 0.5537,
+      "step": 2504
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.404000620078207e-06,
+      "loss": 0.509,
+      "step": 2505
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.399794779053572e-06,
+      "loss": 0.4688,
+      "step": 2506
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.395591458337575e-06,
+      "loss": 0.4536,
+      "step": 2507
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3913906620043833e-06,
+      "loss": 0.3644,
+      "step": 2508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.387192394125709e-06,
+      "loss": 0.4744,
+      "step": 2509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3829966587708273e-06,
+      "loss": 0.3505,
+      "step": 2510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3788034600065457e-06,
+      "loss": 0.4387,
+      "step": 2511
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.3746128018972203e-06,
+      "loss": 0.5378,
+      "step": 2512
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3704246885047423e-06,
+      "loss": 0.5172,
+      "step": 2513
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.366239123888537e-06,
+      "loss": 0.5131,
+      "step": 2514
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3620561121055613e-06,
+      "loss": 0.5423,
+      "step": 2515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.357875657210291e-06,
+      "loss": 0.4632,
+      "step": 2516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3536977632547312e-06,
+      "loss": 0.5615,
+      "step": 2517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3495224342884003e-06,
+      "loss": 0.4846,
+      "step": 2518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.345349674358332e-06,
+      "loss": 0.4661,
+      "step": 2519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3411794875090722e-06,
+      "loss": 0.4971,
+      "step": 2520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.337011877782664e-06,
+      "loss": 0.3766,
+      "step": 2521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3328468492186666e-06,
+      "loss": 0.6597,
+      "step": 2522
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.328684405854124e-06,
+      "loss": 0.5612,
+      "step": 2523
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.3245245517235827e-06,
+      "loss": 0.5318,
+      "step": 2524
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.320367290859075e-06,
+      "loss": 0.4497,
+      "step": 2525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.316212627290123e-06,
+      "loss": 0.4711,
+      "step": 2526
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3120605650437317e-06,
+      "loss": 0.5136,
+      "step": 2527
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3079111081443785e-06,
+      "loss": 0.5506,
+      "step": 2528
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3037642606140216e-06,
+      "loss": 0.4575,
+      "step": 2529
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.299620026472088e-06,
+      "loss": 0.4118,
+      "step": 2530
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.295478409735473e-06,
+      "loss": 0.5302,
+      "step": 2531
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2913394144185347e-06,
+      "loss": 0.555,
+      "step": 2532
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.287203044533084e-06,
+      "loss": 0.6137,
+      "step": 2533
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2830693040884e-06,
+      "loss": 0.4526,
+      "step": 2534
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2789381970911995e-06,
+      "loss": 0.5187,
+      "step": 2535
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2748097275456553e-06,
+      "loss": 0.4654,
+      "step": 2536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.270683899453381e-06,
+      "loss": 0.4182,
+      "step": 2537
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2665607168134306e-06,
+      "loss": 0.4668,
+      "step": 2538
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2624401836222947e-06,
+      "loss": 0.5426,
+      "step": 2539
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.258322303873892e-06,
+      "loss": 0.5663,
+      "step": 2540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2542070815595734e-06,
+      "loss": 0.4508,
+      "step": 2541
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2500945206681128e-06,
+      "loss": 0.5503,
+      "step": 2542
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2459846251857037e-06,
+      "loss": 0.6079,
+      "step": 2543
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.24187739909596e-06,
+      "loss": 0.5764,
+      "step": 2544
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2377728463798976e-06,
+      "loss": 0.5334,
+      "step": 2545
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2336709710159566e-06,
+      "loss": 0.5115,
+      "step": 2546
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.229571776979969e-06,
+      "loss": 0.5829,
+      "step": 2547
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2254752682451743e-06,
+      "loss": 0.7378,
+      "step": 2548
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.2213814487822066e-06,
+      "loss": 0.548,
+      "step": 2549
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.2172903225590966e-06,
+      "loss": 0.61,
+      "step": 2550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.213201893541262e-06,
+      "loss": 0.4331,
+      "step": 2551
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.209116165691504e-06,
+      "loss": 0.7781,
+      "step": 2552
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.20503314297001e-06,
+      "loss": 0.4386,
+      "step": 2553
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.2009528293343435e-06,
+      "loss": 0.5265,
+      "step": 2554
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.196875228739442e-06,
+      "loss": 0.459,
+      "step": 2555
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1928003451376156e-06,
+      "loss": 0.5079,
+      "step": 2556
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.188728182478533e-06,
+      "loss": 0.4822,
+      "step": 2557
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.184658744709239e-06,
+      "loss": 0.5858,
+      "step": 2558
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1805920357741245e-06,
+      "loss": 0.506,
+      "step": 2559
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1765280596149413e-06,
+      "loss": 0.5382,
+      "step": 2560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1724668201707926e-06,
+      "loss": 0.4956,
+      "step": 2561
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.168408321378128e-06,
+      "loss": 0.4113,
+      "step": 2562
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1643525671707423e-06,
+      "loss": 0.4186,
+      "step": 2563
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1602995614797644e-06,
+      "loss": 0.4831,
+      "step": 2564
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1562493082336633e-06,
+      "loss": 0.5451,
+      "step": 2565
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.152201811358246e-06,
+      "loss": 0.6445,
+      "step": 2566
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.148157074776636e-06,
+      "loss": 0.4024,
+      "step": 2567
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1441151024092904e-06,
+      "loss": 0.4202,
+      "step": 2568
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1400758981739794e-06,
+      "loss": 0.4033,
+      "step": 2569
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1360394659858e-06,
+      "loss": 0.5208,
+      "step": 2570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1320058097571533e-06,
+      "loss": 0.4386,
+      "step": 2571
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1279749333977538e-06,
+      "loss": 0.501,
+      "step": 2572
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.123946840814622e-06,
+      "loss": 0.4767,
+      "step": 2573
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.119921535912078e-06,
+      "loss": 0.4513,
+      "step": 2574
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.115899022591743e-06,
+      "loss": 0.5412,
+      "step": 2575
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1118793047525277e-06,
+      "loss": 0.4973,
+      "step": 2576
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1078623862906355e-06,
+      "loss": 0.4849,
+      "step": 2577
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1038482710995623e-06,
+      "loss": 0.4936,
+      "step": 2578
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.099836963070076e-06,
+      "loss": 0.52,
+      "step": 2579
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.0958284660902323e-06,
+      "loss": 0.5333,
+      "step": 2580
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.091822784045354e-06,
+      "loss": 0.5848,
+      "step": 2581
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.087819920818048e-06,
+      "loss": 0.3763,
+      "step": 2582
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.0838198802881754e-06,
+      "loss": 0.617,
+      "step": 2583
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.079822666332869e-06,
+      "loss": 0.4089,
+      "step": 2584
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.0758282828265208e-06,
+      "loss": 0.4547,
+      "step": 2585
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.071836733640779e-06,
+      "loss": 0.5396,
+      "step": 2586
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.0678480226445453e-06,
+      "loss": 0.5523,
+      "step": 2587
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.063862153703966e-06,
+      "loss": 0.3981,
+      "step": 2588
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.0598791306824374e-06,
+      "loss": 0.5207,
+      "step": 2589
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.0558989574405994e-06,
+      "loss": 0.506,
+      "step": 2590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.051921637836322e-06,
+      "loss": 0.5745,
+      "step": 2591
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.0479471757247176e-06,
+      "loss": 0.4736,
+      "step": 2592
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.043975574958119e-06,
+      "loss": 0.4437,
+      "step": 2593
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.040006839386099e-06,
+      "loss": 0.3698,
+      "step": 2594
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.03604097285544e-06,
+      "loss": 0.5037,
+      "step": 2595
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.032077979210151e-06,
+      "loss": 0.5011,
+      "step": 2596
+    },
+    {
+      "epoch": 2.2,
+      "eval_loss": 1.0270994901657104,
+      "eval_runtime": 268.019,
+      "eval_samples_per_second": 3.727,
+      "eval_steps_per_second": 0.466,
+      "step": 2596
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.028117862291454e-06,
+      "loss": 0.5455,
+      "step": 2597
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.024160625937785e-06,
+      "loss": 0.4116,
+      "step": 2598
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.0202062739847867e-06,
+      "loss": 0.6043,
+      "step": 2599
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.016254810265302e-06,
+      "loss": 0.4459,
+      "step": 2600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.0123062386093785e-06,
+      "loss": 0.5848,
+      "step": 2601
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.008360562844265e-06,
+      "loss": 0.4134,
+      "step": 2602
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.004417786794393e-06,
+      "loss": 0.6051,
+      "step": 2603
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.0004779142813923e-06,
+      "loss": 0.5003,
+      "step": 2604
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9965409491240696e-06,
+      "loss": 0.4323,
+      "step": 2605
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.992606895138426e-06,
+      "loss": 0.4925,
+      "step": 2606
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.988675756137628e-06,
+      "loss": 0.5165,
+      "step": 2607
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.984747535932024e-06,
+      "loss": 0.6392,
+      "step": 2608
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9808222383291326e-06,
+      "loss": 0.5188,
+      "step": 2609
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.976899867133638e-06,
+      "loss": 0.4339,
+      "step": 2610
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9729804261473904e-06,
+      "loss": 0.4901,
+      "step": 2611
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.969063919169395e-06,
+      "loss": 0.4585,
+      "step": 2612
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.965150349995815e-06,
+      "loss": 0.4234,
+      "step": 2613
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9612397224199736e-06,
+      "loss": 0.5003,
+      "step": 2614
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.957332040232331e-06,
+      "loss": 0.4726,
+      "step": 2615
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9534273072205007e-06,
+      "loss": 0.522,
+      "step": 2616
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9495255271692297e-06,
+      "loss": 0.4795,
+      "step": 2617
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.945626703860416e-06,
+      "loss": 0.4984,
+      "step": 2618
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9417308410730777e-06,
+      "loss": 0.4967,
+      "step": 2619
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9378379425833703e-06,
+      "loss": 0.4878,
+      "step": 2620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9339480121645763e-06,
+      "loss": 0.4796,
+      "step": 2621
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9300610535871e-06,
+      "loss": 0.5676,
+      "step": 2622
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.926177070618466e-06,
+      "loss": 0.5209,
+      "step": 2623
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.922296067023312e-06,
+      "loss": 0.6343,
+      "step": 2624
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.918418046563389e-06,
+      "loss": 0.4953,
+      "step": 2625
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9145430129975624e-06,
+      "loss": 0.4585,
+      "step": 2626
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.910670970081792e-06,
+      "loss": 0.4557,
+      "step": 2627
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9068019215691492e-06,
+      "loss": 0.5131,
+      "step": 2628
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9029358712097917e-06,
+      "loss": 0.4415,
+      "step": 2629
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8990728227509852e-06,
+      "loss": 0.4748,
+      "step": 2630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.895212779937073e-06,
+      "loss": 0.5467,
+      "step": 2631
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8913557465094912e-06,
+      "loss": 0.537,
+      "step": 2632
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.887501726206759e-06,
+      "loss": 0.5802,
+      "step": 2633
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8836507227644735e-06,
+      "loss": 0.5048,
+      "step": 2634
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8798027399153102e-06,
+      "loss": 0.3967,
+      "step": 2635
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8759577813890113e-06,
+      "loss": 0.3919,
+      "step": 2636
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8721158509123916e-06,
+      "loss": 0.4544,
+      "step": 2637
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8682769522093347e-06,
+      "loss": 0.4811,
+      "step": 2638
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.864441089000777e-06,
+      "loss": 0.4775,
+      "step": 2639
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8606082650047203e-06,
+      "loss": 0.4733,
+      "step": 2640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8567784839362123e-06,
+      "loss": 0.5984,
+      "step": 2641
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8529517495073629e-06,
+      "loss": 0.5656,
+      "step": 2642
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.849128065427317e-06,
+      "loss": 0.4466,
+      "step": 2643
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8453074354022698e-06,
+      "loss": 0.4834,
+      "step": 2644
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8414898631354554e-06,
+      "loss": 0.4821,
+      "step": 2645
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8376753523271433e-06,
+      "loss": 0.5275,
+      "step": 2646
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8338639066746373e-06,
+      "loss": 0.407,
+      "step": 2647
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.830055529872265e-06,
+      "loss": 0.5167,
+      "step": 2648
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8262502256113828e-06,
+      "loss": 0.5089,
+      "step": 2649
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8224479975803756e-06,
+      "loss": 0.4791,
+      "step": 2650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8186488494646337e-06,
+      "loss": 0.5255,
+      "step": 2651
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.814852784946573e-06,
+      "loss": 0.3961,
+      "step": 2652
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8110598077056107e-06,
+      "loss": 0.5068,
+      "step": 2653
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8072699214181843e-06,
+      "loss": 0.5606,
+      "step": 2654
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8034831297577227e-06,
+      "loss": 0.6338,
+      "step": 2655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.799699436394663e-06,
+      "loss": 0.4759,
+      "step": 2656
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.795918844996436e-06,
+      "loss": 0.4798,
+      "step": 2657
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7921413592274673e-06,
+      "loss": 0.4813,
+      "step": 2658
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7883669827491729e-06,
+      "loss": 0.6781,
+      "step": 2659
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7845957192199514e-06,
+      "loss": 0.4132,
+      "step": 2660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7808275722951861e-06,
+      "loss": 0.4535,
+      "step": 2661
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.777062545627245e-06,
+      "loss": 0.4206,
+      "step": 2662
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.773300642865462e-06,
+      "loss": 0.4688,
+      "step": 2663
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7695418676561504e-06,
+      "loss": 0.525,
+      "step": 2664
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7657862236425854e-06,
+      "loss": 0.4324,
+      "step": 2665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.7620337144650168e-06,
+      "loss": 0.4285,
+      "step": 2666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7582843437606462e-06,
+      "loss": 0.5373,
+      "step": 2667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.754538115163638e-06,
+      "loss": 0.4698,
+      "step": 2668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7507950323051114e-06,
+      "loss": 0.5167,
+      "step": 2669
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7470550988131353e-06,
+      "loss": 0.4664,
+      "step": 2670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7433183183127267e-06,
+      "loss": 0.4707,
+      "step": 2671
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7395846944258437e-06,
+      "loss": 0.48,
+      "step": 2672
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7358542307713865e-06,
+      "loss": 0.5,
+      "step": 2673
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7321269309651982e-06,
+      "loss": 0.5394,
+      "step": 2674
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.728402798620044e-06,
+      "loss": 0.5959,
+      "step": 2675
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7246818373456293e-06,
+      "loss": 0.55,
+      "step": 2676
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7209640507485748e-06,
+      "loss": 0.4411,
+      "step": 2677
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.7172494424324377e-06,
+      "loss": 0.5653,
+      "step": 2678
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7135380159976834e-06,
+      "loss": 0.4998,
+      "step": 2679
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7098297750416976e-06,
+      "loss": 0.4561,
+      "step": 2680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7061247231587785e-06,
+      "loss": 0.4662,
+      "step": 2681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.7024228639401325e-06,
+      "loss": 0.4884,
+      "step": 2682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.698724200973874e-06,
+      "loss": 0.4311,
+      "step": 2683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6950287378450136e-06,
+      "loss": 0.4764,
+      "step": 2684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.691336478135463e-06,
+      "loss": 0.4103,
+      "step": 2685
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6876474254240354e-06,
+      "loss": 0.5208,
+      "step": 2686
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6839615832864254e-06,
+      "loss": 0.4753,
+      "step": 2687
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.680278955295222e-06,
+      "loss": 0.4602,
+      "step": 2688
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6765995450198935e-06,
+      "loss": 0.4467,
+      "step": 2689
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.6729233560267981e-06,
+      "loss": 0.3838,
+      "step": 2690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.669250391879163e-06,
+      "loss": 0.4428,
+      "step": 2691
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6655806561370935e-06,
+      "loss": 0.3837,
+      "step": 2692
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6619141523575654e-06,
+      "loss": 0.4626,
+      "step": 2693
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6582508840944224e-06,
+      "loss": 0.3841,
+      "step": 2694
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6545908548983725e-06,
+      "loss": 0.4558,
+      "step": 2695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.65093406831698e-06,
+      "loss": 0.4391,
+      "step": 2696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6472805278946685e-06,
+      "loss": 0.5041,
+      "step": 2697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6436302371727224e-06,
+      "loss": 0.5067,
+      "step": 2698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.639983199689264e-06,
+      "loss": 0.5265,
+      "step": 2699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6363394189792715e-06,
+      "loss": 0.5467,
+      "step": 2700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6326988985745578e-06,
+      "loss": 0.4251,
+      "step": 2701
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.6290616420037881e-06,
+      "loss": 0.5143,
+      "step": 2702
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6254276527924517e-06,
+      "loss": 0.6457,
+      "step": 2703
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6217969344628776e-06,
+      "loss": 0.5216,
+      "step": 2704
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6181694905342231e-06,
+      "loss": 0.5597,
+      "step": 2705
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6145453245224718e-06,
+      "loss": 0.4517,
+      "step": 2706
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6109244399404312e-06,
+      "loss": 0.5914,
+      "step": 2707
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6073068402977237e-06,
+      "loss": 0.4173,
+      "step": 2708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6036925291007909e-06,
+      "loss": 0.3838,
+      "step": 2709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.6000815098528915e-06,
+      "loss": 0.5566,
+      "step": 2710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5964737860540842e-06,
+      "loss": 0.4966,
+      "step": 2711
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5928693612012414e-06,
+      "loss": 0.5177,
+      "step": 2712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.5892682387880299e-06,
+      "loss": 0.5023,
+      "step": 2713
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5856704223049253e-06,
+      "loss": 0.4969,
+      "step": 2714
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5820759152391928e-06,
+      "loss": 0.4203,
+      "step": 2715
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5784847210748893e-06,
+      "loss": 0.4611,
+      "step": 2716
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5748968432928636e-06,
+      "loss": 0.5258,
+      "step": 2717
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5713122853707475e-06,
+      "loss": 0.5163,
+      "step": 2718
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5677310507829585e-06,
+      "loss": 0.5372,
+      "step": 2719
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5641531430006873e-06,
+      "loss": 0.5156,
+      "step": 2720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5605785654919025e-06,
+      "loss": 0.5376,
+      "step": 2721
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5570073217213515e-06,
+      "loss": 0.5534,
+      "step": 2722
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.5534394151505389e-06,
+      "loss": 0.5853,
+      "step": 2723
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.549874849237743e-06,
+      "loss": 0.4533,
+      "step": 2724
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.546313627437997e-06,
+      "loss": 0.4475,
+      "step": 2725
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5427557532031013e-06,
+      "loss": 0.4804,
+      "step": 2726
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5392012299816073e-06,
+      "loss": 0.45,
+      "step": 2727
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5356500612188153e-06,
+      "loss": 0.4757,
+      "step": 2728
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.532102250356778e-06,
+      "loss": 0.5136,
+      "step": 2729
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5285578008342939e-06,
+      "loss": 0.4169,
+      "step": 2730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5250167160869023e-06,
+      "loss": 0.5715,
+      "step": 2731
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5214789995468793e-06,
+      "loss": 0.5457,
+      "step": 2732
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5179446546432363e-06,
+      "loss": 0.4027,
+      "step": 2733
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5144136848017242e-06,
+      "loss": 0.4765,
+      "step": 2734
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5108860934448116e-06,
+      "loss": 0.4638,
+      "step": 2735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.5073618839917003e-06,
+      "loss": 0.4114,
+      "step": 2736
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.503841059858307e-06,
+      "loss": 0.4571,
+      "step": 2737
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.5003236244572756e-06,
+      "loss": 0.6017,
+      "step": 2738
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.496809581197961e-06,
+      "loss": 0.6599,
+      "step": 2739
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4932989334864284e-06,
+      "loss": 0.5497,
+      "step": 2740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.489791684725454e-06,
+      "loss": 0.5808,
+      "step": 2741
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4862878383145196e-06,
+      "loss": 0.4497,
+      "step": 2742
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4827873976498108e-06,
+      "loss": 0.5409,
+      "step": 2743
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4792903661242063e-06,
+      "loss": 0.553,
+      "step": 2744
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.475796747127285e-06,
+      "loss": 0.506,
+      "step": 2745
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4723065440453217e-06,
+      "loss": 0.4943,
+      "step": 2746
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4688197602612708e-06,
+      "loss": 0.5119,
+      "step": 2747
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4653363991547815e-06,
+      "loss": 0.5962,
+      "step": 2748
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.4618564641021765e-06,
+      "loss": 0.5343,
+      "step": 2749
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4583799584764675e-06,
+      "loss": 0.4549,
+      "step": 2750
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.454906885647337e-06,
+      "loss": 0.5684,
+      "step": 2751
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.451437248981138e-06,
+      "loss": 0.5121,
+      "step": 2752
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4479710518408973e-06,
+      "loss": 0.5567,
+      "step": 2753
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4445082975863056e-06,
+      "loss": 0.5292,
+      "step": 2754
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4410489895737185e-06,
+      "loss": 0.5,
+      "step": 2755
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.437593131156147e-06,
+      "loss": 0.5247,
+      "step": 2756
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4341407256832612e-06,
+      "loss": 0.446,
+      "step": 2757
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4306917765013888e-06,
+      "loss": 0.5978,
+      "step": 2758
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4272462869534987e-06,
+      "loss": 0.5082,
+      "step": 2759
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.4238042603792141e-06,
+      "loss": 0.5019,
+      "step": 2760
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.420365700114793e-06,
+      "loss": 0.4585,
+      "step": 2761
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.416930609493144e-06,
+      "loss": 0.5135,
+      "step": 2762
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.413498991843807e-06,
+      "loss": 0.4928,
+      "step": 2763
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4100708504929539e-06,
+      "loss": 0.5531,
+      "step": 2764
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4066461887633908e-06,
+      "loss": 0.4867,
+      "step": 2765
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.4032250099745499e-06,
+      "loss": 0.4733,
+      "step": 2766
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3998073174424882e-06,
+      "loss": 0.3983,
+      "step": 2767
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3963931144798803e-06,
+      "loss": 0.5736,
+      "step": 2768
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3929824043960205e-06,
+      "loss": 0.6455,
+      "step": 2769
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3895751904968225e-06,
+      "loss": 0.4519,
+      "step": 2770
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3861714760848022e-06,
+      "loss": 0.5112,
+      "step": 2771
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.3827712644590907e-06,
+      "loss": 0.4237,
+      "step": 2772
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3793745589154172e-06,
+      "loss": 0.5095,
+      "step": 2773
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3759813627461204e-06,
+      "loss": 0.5485,
+      "step": 2774
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.372591679240135e-06,
+      "loss": 0.494,
+      "step": 2775
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.369205511682985e-06,
+      "loss": 0.4339,
+      "step": 2776
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3658228633567933e-06,
+      "loss": 0.5119,
+      "step": 2777
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3624437375402694e-06,
+      "loss": 0.4342,
+      "step": 2778
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.35906813750871e-06,
+      "loss": 0.3786,
+      "step": 2779
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3556960665339909e-06,
+      "loss": 0.6057,
+      "step": 2780
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3523275278845677e-06,
+      "loss": 0.5371,
+      "step": 2781
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3489625248254795e-06,
+      "loss": 0.4878,
+      "step": 2782
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3456010606183278e-06,
+      "loss": 0.4771,
+      "step": 2783
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.3422431385212925e-06,
+      "loss": 0.4919,
+      "step": 2784
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3388887617891122e-06,
+      "loss": 0.5919,
+      "step": 2785
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3355379336730977e-06,
+      "loss": 0.4853,
+      "step": 2786
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3321906574211162e-06,
+      "loss": 0.4468,
+      "step": 2787
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.328846936277589e-06,
+      "loss": 0.4835,
+      "step": 2788
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3255067734834965e-06,
+      "loss": 0.6741,
+      "step": 2789
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.322170172276368e-06,
+      "loss": 0.449,
+      "step": 2790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3188371358902823e-06,
+      "loss": 0.5206,
+      "step": 2791
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3155076675558587e-06,
+      "loss": 0.6016,
+      "step": 2792
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3121817705002605e-06,
+      "loss": 0.5916,
+      "step": 2793
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.308859447947195e-06,
+      "loss": 0.5172,
+      "step": 2794
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3055407031168943e-06,
+      "loss": 0.6026,
+      "step": 2795
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.3022255392261313e-06,
+      "loss": 0.5061,
+      "step": 2796
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2989139594882005e-06,
+      "loss": 0.444,
+      "step": 2797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2956059671129312e-06,
+      "loss": 0.4673,
+      "step": 2798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2923015653066706e-06,
+      "loss": 0.4122,
+      "step": 2799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2890007572722836e-06,
+      "loss": 0.547,
+      "step": 2800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2857035462091549e-06,
+      "loss": 0.4249,
+      "step": 2801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.282409935313183e-06,
+      "loss": 0.4346,
+      "step": 2802
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2791199277767768e-06,
+      "loss": 0.4564,
+      "step": 2803
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2758335267888488e-06,
+      "loss": 0.4384,
+      "step": 2804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2725507355348187e-06,
+      "loss": 0.3833,
+      "step": 2805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.269271557196612e-06,
+      "loss": 0.4234,
+      "step": 2806
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2659959949526428e-06,
+      "loss": 0.5163,
+      "step": 2807
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.2627240519778279e-06,
+      "loss": 0.4464,
+      "step": 2808
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2594557314435691e-06,
+      "loss": 0.5316,
+      "step": 2809
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.256191036517766e-06,
+      "loss": 0.5139,
+      "step": 2810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.252929970364798e-06,
+      "loss": 0.4755,
+      "step": 2811
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2496725361455264e-06,
+      "loss": 0.4526,
+      "step": 2812
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2464187370172953e-06,
+      "loss": 0.4114,
+      "step": 2813
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2431685761339241e-06,
+      "loss": 0.4698,
+      "step": 2814
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2399220566457076e-06,
+      "loss": 0.6921,
+      "step": 2815
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2366791816994062e-06,
+      "loss": 0.4334,
+      "step": 2816
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2334399544382515e-06,
+      "loss": 0.4648,
+      "step": 2817
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2302043780019423e-06,
+      "loss": 0.5115,
+      "step": 2818
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2269724555266321e-06,
+      "loss": 0.4484,
+      "step": 2819
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.2237441901449382e-06,
+      "loss": 0.6987,
+      "step": 2820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.220519584985927e-06,
+      "loss": 0.6858,
+      "step": 2821
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.217298643175125e-06,
+      "loss": 0.4339,
+      "step": 2822
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2140813678345043e-06,
+      "loss": 0.5328,
+      "step": 2823
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2108677620824805e-06,
+      "loss": 0.5154,
+      "step": 2824
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2076578290339153e-06,
+      "loss": 0.3181,
+      "step": 2825
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.204451571800111e-06,
+      "loss": 0.4701,
+      "step": 2826
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.2012489934888066e-06,
+      "loss": 0.3529,
+      "step": 2827
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1980500972041726e-06,
+      "loss": 0.5508,
+      "step": 2828
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1948548860468124e-06,
+      "loss": 0.4869,
+      "step": 2829
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.191663363113762e-06,
+      "loss": 0.512,
+      "step": 2830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1884755314984742e-06,
+      "loss": 0.4805,
+      "step": 2831
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1852913942908307e-06,
+      "loss": 0.4495,
+      "step": 2832
+    },
+    {
+      "epoch": 2.39,
+      "eval_loss": 1.0266658067703247,
+      "eval_runtime": 267.9157,
+      "eval_samples_per_second": 3.729,
+      "eval_steps_per_second": 0.467,
+      "step": 2832
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1821109545771253e-06,
+      "loss": 0.4929,
+      "step": 2833
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1789342154400768e-06,
+      "loss": 0.4721,
+      "step": 2834
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1757611799588126e-06,
+      "loss": 0.5839,
+      "step": 2835
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1725918512088669e-06,
+      "loss": 0.4117,
+      "step": 2836
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1694262322621864e-06,
+      "loss": 0.3522,
+      "step": 2837
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1662643261871193e-06,
+      "loss": 0.5141,
+      "step": 2838
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1631061360484174e-06,
+      "loss": 0.575,
+      "step": 2839
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1599516649072268e-06,
+      "loss": 0.4757,
+      "step": 2840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1568009158210908e-06,
+      "loss": 0.3892,
+      "step": 2841
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1536538918439493e-06,
+      "loss": 0.5594,
+      "step": 2842
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.150510596026123e-06,
+      "loss": 0.5335,
+      "step": 2843
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.147371031414327e-06,
+      "loss": 0.3519,
+      "step": 2844
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1442352010516522e-06,
+      "loss": 0.532,
+      "step": 2845
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1411031079775786e-06,
+      "loss": 0.4932,
+      "step": 2846
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1379747552279585e-06,
+      "loss": 0.6044,
+      "step": 2847
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1348501458350176e-06,
+      "loss": 0.4159,
+      "step": 2848
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1317292828273562e-06,
+      "loss": 0.3627,
+      "step": 2849
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1286121692299428e-06,
+      "loss": 0.5109,
+      "step": 2850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1254988080641132e-06,
+      "loss": 0.4783,
+      "step": 2851
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1223892023475608e-06,
+      "loss": 0.666,
+      "step": 2852
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1192833550943422e-06,
+      "loss": 0.4672,
+      "step": 2853
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1161812693148755e-06,
+      "loss": 0.4669,
+      "step": 2854
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.113082948015925e-06,
+      "loss": 0.417,
+      "step": 2855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1099883942006123e-06,
+      "loss": 0.6225,
+      "step": 2856
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1068976108684004e-06,
+      "loss": 0.5571,
+      "step": 2857
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1038106010151061e-06,
+      "loss": 0.5528,
+      "step": 2858
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1007273676328846e-06,
+      "loss": 0.4779,
+      "step": 2859
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0976479137102286e-06,
+      "loss": 0.4254,
+      "step": 2860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0945722422319699e-06,
+      "loss": 0.43,
+      "step": 2861
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0915003561792748e-06,
+      "loss": 0.4565,
+      "step": 2862
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0884322585296386e-06,
+      "loss": 0.4393,
+      "step": 2863
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0853679522568871e-06,
+      "loss": 0.5064,
+      "step": 2864
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0823074403311656e-06,
+      "loss": 0.3918,
+      "step": 2865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0792507257189504e-06,
+      "loss": 0.4356,
+      "step": 2866
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0761978113830291e-06,
+      "loss": 0.429,
+      "step": 2867
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0731487002825113e-06,
+      "loss": 0.4628,
+      "step": 2868
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.070103395372814e-06,
+      "loss": 0.4325,
+      "step": 2869
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0670618996056731e-06,
+      "loss": 0.4283,
+      "step": 2870
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.064024215929128e-06,
+      "loss": 0.5337,
+      "step": 2871
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.060990347287521e-06,
+      "loss": 0.5147,
+      "step": 2872
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0579602966214997e-06,
+      "loss": 0.3685,
+      "step": 2873
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0549340668680107e-06,
+      "loss": 0.5334,
+      "step": 2874
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0519116609602975e-06,
+      "loss": 0.4525,
+      "step": 2875
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0488930818278964e-06,
+      "loss": 0.361,
+      "step": 2876
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0458783323966314e-06,
+      "loss": 0.4911,
+      "step": 2877
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0428674155886225e-06,
+      "loss": 0.5827,
+      "step": 2878
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0398603343222663e-06,
+      "loss": 0.4665,
+      "step": 2879
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0368570915122483e-06,
+      "loss": 0.5715,
+      "step": 2880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0338576900695257e-06,
+      "loss": 0.4744,
+      "step": 2881
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0308621329013424e-06,
+      "loss": 0.37,
+      "step": 2882
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.02787042291121e-06,
+      "loss": 0.4672,
+      "step": 2883
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0248825629989102e-06,
+      "loss": 0.4781,
+      "step": 2884
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0218985560604961e-06,
+      "loss": 0.5254,
+      "step": 2885
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.018918404988284e-06,
+      "loss": 0.4973,
+      "step": 2886
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0159421126708547e-06,
+      "loss": 0.445,
+      "step": 2887
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0129696819930495e-06,
+      "loss": 0.4005,
+      "step": 2888
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.01000111583596e-06,
+      "loss": 0.4696,
+      "step": 2889
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0070364170769438e-06,
+      "loss": 0.3887,
+      "step": 2890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.004075588589598e-06,
+      "loss": 0.4491,
+      "step": 2891
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0011186332437773e-06,
+      "loss": 0.498,
+      "step": 2892
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.981655539055734e-07,
+      "loss": 0.4109,
+      "step": 2893
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.952163534373322e-07,
+      "loss": 0.4583,
+      "step": 2894
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.922710346976322e-07,
+      "loss": 0.5681,
+      "step": 2895
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.89329600541289e-07,
+      "loss": 0.4346,
+      "step": 2896
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.863920538193573e-07,
+      "loss": 0.5689,
+      "step": 2897
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.834583973791207e-07,
+      "loss": 0.4587,
+      "step": 2898
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.80528634064094e-07,
+      "loss": 0.5379,
+      "step": 2899
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.776027667140181e-07,
+      "loss": 0.55,
+      "step": 2900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.746807981648542e-07,
+      "loss": 0.5936,
+      "step": 2901
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.717627312487915e-07,
+      "loss": 0.5487,
+      "step": 2902
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.688485687942305e-07,
+      "loss": 0.519,
+      "step": 2903
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.65938313625792e-07,
+      "loss": 0.5321,
+      "step": 2904
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.630319685643042e-07,
+      "loss": 0.4994,
+      "step": 2905
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.601295364268126e-07,
+      "loss": 0.5523,
+      "step": 2906
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.572310200265661e-07,
+      "loss": 0.5768,
+      "step": 2907
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.543364221730157e-07,
+      "loss": 0.5441,
+      "step": 2908
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.514457456718179e-07,
+      "loss": 0.6147,
+      "step": 2909
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.485589933248279e-07,
+      "loss": 0.3534,
+      "step": 2910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.456761679300963e-07,
+      "loss": 0.4372,
+      "step": 2911
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.427972722818695e-07,
+      "loss": 0.525,
+      "step": 2912
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.399223091705789e-07,
+      "loss": 0.4995,
+      "step": 2913
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.370512813828542e-07,
+      "loss": 0.4369,
+      "step": 2914
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.341841917015004e-07,
+      "loss": 0.4055,
+      "step": 2915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.313210429055137e-07,
+      "loss": 0.4834,
+      "step": 2916
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.284618377700621e-07,
+      "loss": 0.5456,
+      "step": 2917
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.256065790664997e-07,
+      "loss": 0.3755,
+      "step": 2918
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.227552695623521e-07,
+      "loss": 0.3776,
+      "step": 2919
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.199079120213139e-07,
+      "loss": 0.4959,
+      "step": 2920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.170645092032526e-07,
+      "loss": 0.3691,
+      "step": 2921
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.142250638642026e-07,
+      "loss": 0.434,
+      "step": 2922
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.113895787563604e-07,
+      "loss": 0.5779,
+      "step": 2923
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.085580566280866e-07,
+      "loss": 0.4821,
+      "step": 2924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.057305002238959e-07,
+      "loss": 0.4677,
+      "step": 2925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.02906912284467e-07,
+      "loss": 0.508,
+      "step": 2926
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.000872955466233e-07,
+      "loss": 0.4239,
+      "step": 2927
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.972716527433461e-07,
+      "loss": 0.4842,
+      "step": 2928
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.944599866037579e-07,
+      "loss": 0.508,
+      "step": 2929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.916522998531346e-07,
+      "loss": 0.4341,
+      "step": 2930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.888485952128906e-07,
+      "loss": 0.4474,
+      "step": 2931
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.860488754005792e-07,
+      "loss": 0.5189,
+      "step": 2932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.832531431298935e-07,
+      "loss": 0.585,
+      "step": 2933
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.804614011106621e-07,
+      "loss": 0.5052,
+      "step": 2934
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.776736520488444e-07,
+      "loss": 0.5346,
+      "step": 2935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.748898986465321e-07,
+      "loss": 0.5681,
+      "step": 2936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.721101436019381e-07,
+      "loss": 0.495,
+      "step": 2937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.693343896094087e-07,
+      "loss": 0.5598,
+      "step": 2938
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.665626393594028e-07,
+      "loss": 0.4076,
+      "step": 2939
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.637948955385072e-07,
+      "loss": 0.4472,
+      "step": 2940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.610311608294164e-07,
+      "loss": 0.518,
+      "step": 2941
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.582714379109469e-07,
+      "loss": 0.5128,
+      "step": 2942
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.555157294580246e-07,
+      "loss": 0.5009,
+      "step": 2943
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.527640381416808e-07,
+      "loss": 0.4787,
+      "step": 2944
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.500163666290557e-07,
+      "loss": 0.5741,
+      "step": 2945
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.472727175833934e-07,
+      "loss": 0.5224,
+      "step": 2946
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.445330936640394e-07,
+      "loss": 0.4446,
+      "step": 2947
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.417974975264381e-07,
+      "loss": 0.5719,
+      "step": 2948
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.390659318221245e-07,
+      "loss": 0.5401,
+      "step": 2949
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.363383991987372e-07,
+      "loss": 0.4613,
+      "step": 2950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.336149022999957e-07,
+      "loss": 0.4539,
+      "step": 2951
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.308954437657147e-07,
+      "loss": 0.4598,
+      "step": 2952
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.281800262317874e-07,
+      "loss": 0.4626,
+      "step": 2953
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.254686523301997e-07,
+      "loss": 0.5129,
+      "step": 2954
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.227613246890115e-07,
+      "loss": 0.5089,
+      "step": 2955
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.20058045932361e-07,
+      "loss": 0.4206,
+      "step": 2956
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.173588186804637e-07,
+      "loss": 0.486,
+      "step": 2957
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.146636455496077e-07,
+      "loss": 0.3949,
+      "step": 2958
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.119725291521519e-07,
+      "loss": 0.5366,
+      "step": 2959
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.092854720965238e-07,
+      "loss": 0.5605,
+      "step": 2960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.066024769872116e-07,
+      "loss": 0.4434,
+      "step": 2961
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.039235464247741e-07,
+      "loss": 0.477,
+      "step": 2962
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.012486830058236e-07,
+      "loss": 0.5438,
+      "step": 2963
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.985778893230345e-07,
+      "loss": 0.4148,
+      "step": 2964
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.959111679651321e-07,
+      "loss": 0.421,
+      "step": 2965
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.932485215169e-07,
+      "loss": 0.4862,
+      "step": 2966
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.905899525591704e-07,
+      "loss": 0.5497,
+      "step": 2967
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.879354636688197e-07,
+      "loss": 0.6,
+      "step": 2968
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.852850574187726e-07,
+      "loss": 0.5699,
+      "step": 2969
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.826387363779975e-07,
+      "loss": 0.5077,
+      "step": 2970
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.799965031115008e-07,
+      "loss": 0.3533,
+      "step": 2971
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.773583601803292e-07,
+      "loss": 0.5884,
+      "step": 2972
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 7.747243101415602e-07,
+      "loss": 0.4711,
+      "step": 2973
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.720943555483118e-07,
+      "loss": 0.6051,
+      "step": 2974
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.694684989497248e-07,
+      "loss": 0.4417,
+      "step": 2975
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.668467428909726e-07,
+      "loss": 0.4013,
+      "step": 2976
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.642290899132493e-07,
+      "loss": 0.5886,
+      "step": 2977
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.616155425537786e-07,
+      "loss": 0.5195,
+      "step": 2978
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.590061033458013e-07,
+      "loss": 0.4136,
+      "step": 2979
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.564007748185743e-07,
+      "loss": 0.4877,
+      "step": 2980
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.53799559497373e-07,
+      "loss": 0.3974,
+      "step": 2981
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.512024599034851e-07,
+      "loss": 0.4531,
+      "step": 2982
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.486094785542092e-07,
+      "loss": 0.5951,
+      "step": 2983
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.460206179628532e-07,
+      "loss": 0.4214,
+      "step": 2984
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.434358806387259e-07,
+      "loss": 0.5514,
+      "step": 2985
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.408552690871485e-07,
+      "loss": 0.5195,
+      "step": 2986
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.382787858094337e-07,
+      "loss": 0.4648,
+      "step": 2987
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.357064333029002e-07,
+      "loss": 0.4391,
+      "step": 2988
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.331382140608562e-07,
+      "loss": 0.4638,
+      "step": 2989
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.3057413057261e-07,
+      "loss": 0.4762,
+      "step": 2990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.280141853234591e-07,
+      "loss": 0.4508,
+      "step": 2991
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.254583807946869e-07,
+      "loss": 0.414,
+      "step": 2992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.229067194635664e-07,
+      "loss": 0.6263,
+      "step": 2993
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.203592038033542e-07,
+      "loss": 0.5056,
+      "step": 2994
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.178158362832883e-07,
+      "loss": 0.5181,
+      "step": 2995
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.152766193685873e-07,
+      "loss": 0.4312,
+      "step": 2996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.127415555204426e-07,
+      "loss": 0.3681,
+      "step": 2997
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.102106471960263e-07,
+      "loss": 0.5442,
+      "step": 2998
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.076838968484773e-07,
+      "loss": 0.4879,
+      "step": 2999
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.051613069269081e-07,
+      "loss": 0.5118,
+      "step": 3000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.026428798763935e-07,
+      "loss": 0.4771,
+      "step": 3001
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.001286181379802e-07,
+      "loss": 0.4804,
+      "step": 3002
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.976185241486749e-07,
+      "loss": 0.5163,
+      "step": 3003
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.951126003414405e-07,
+      "loss": 0.5162,
+      "step": 3004
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.926108491452027e-07,
+      "loss": 0.4108,
+      "step": 3005
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.901132729848414e-07,
+      "loss": 0.5467,
+      "step": 3006
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.876198742811896e-07,
+      "loss": 0.4628,
+      "step": 3007
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.851306554510323e-07,
+      "loss": 0.5777,
+      "step": 3008
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.82645618907099e-07,
+      "loss": 0.441,
+      "step": 3009
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.80164767058073e-07,
+      "loss": 0.5689,
+      "step": 3010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.77688102308573e-07,
+      "loss": 0.4586,
+      "step": 3011
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.752156270591648e-07,
+      "loss": 0.4808,
+      "step": 3012
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.727473437063514e-07,
+      "loss": 0.4975,
+      "step": 3013
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.702832546425736e-07,
+      "loss": 0.4256,
+      "step": 3014
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.678233622562075e-07,
+      "loss": 0.4978,
+      "step": 3015
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.653676689315575e-07,
+      "loss": 0.5041,
+      "step": 3016
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.629161770488623e-07,
+      "loss": 0.5047,
+      "step": 3017
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.604688889842865e-07,
+      "loss": 0.4767,
+      "step": 3018
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.580258071099204e-07,
+      "loss": 0.5524,
+      "step": 3019
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.555869337937781e-07,
+      "loss": 0.4548,
+      "step": 3020
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.531522713997912e-07,
+      "loss": 0.533,
+      "step": 3021
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.507218222878154e-07,
+      "loss": 0.6037,
+      "step": 3022
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.482955888136161e-07,
+      "loss": 0.5221,
+      "step": 3023
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.458735733288779e-07,
+      "loss": 0.485,
+      "step": 3024
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.43455778181194e-07,
+      "loss": 0.4808,
+      "step": 3025
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.410422057140687e-07,
+      "loss": 0.4851,
+      "step": 3026
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.386328582669127e-07,
+      "loss": 0.398,
+      "step": 3027
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.362277381750398e-07,
+      "loss": 0.4792,
+      "step": 3028
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.338268477696685e-07,
+      "loss": 0.5749,
+      "step": 3029
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.314301893779179e-07,
+      "loss": 0.4299,
+      "step": 3030
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.29037765322803e-07,
+      "loss": 0.4758,
+      "step": 3031
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.266495779232379e-07,
+      "loss": 0.3755,
+      "step": 3032
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.242656294940242e-07,
+      "loss": 0.5404,
+      "step": 3033
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.218859223458635e-07,
+      "loss": 0.5032,
+      "step": 3034
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.195104587853379e-07,
+      "loss": 0.401,
+      "step": 3035
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.171392411149207e-07,
+      "loss": 0.4786,
+      "step": 3036
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.147722716329707e-07,
+      "loss": 0.519,
+      "step": 3037
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.124095526337259e-07,
+      "loss": 0.5022,
+      "step": 3038
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.100510864073072e-07,
+      "loss": 0.5102,
+      "step": 3039
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.076968752397111e-07,
+      "loss": 0.5082,
+      "step": 3040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.053469214128105e-07,
+      "loss": 0.4839,
+      "step": 3041
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.030012272043534e-07,
+      "loss": 0.4645,
+      "step": 3042
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.006597948879572e-07,
+      "loss": 0.5165,
+      "step": 3043
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.983226267331094e-07,
+      "loss": 0.4771,
+      "step": 3044
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.959897250051621e-07,
+      "loss": 0.4156,
+      "step": 3045
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.936610919653368e-07,
+      "loss": 0.66,
+      "step": 3046
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.913367298707124e-07,
+      "loss": 0.3347,
+      "step": 3047
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.890166409742309e-07,
+      "loss": 0.4058,
+      "step": 3048
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.867008275246916e-07,
+      "loss": 0.5432,
+      "step": 3049
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.843892917667499e-07,
+      "loss": 0.5159,
+      "step": 3050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.820820359409162e-07,
+      "loss": 0.4773,
+      "step": 3051
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.797790622835492e-07,
+      "loss": 0.5584,
+      "step": 3052
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.774803730268597e-07,
+      "loss": 0.5329,
+      "step": 3053
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.751859703989049e-07,
+      "loss": 0.5791,
+      "step": 3054
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.728958566235881e-07,
+      "loss": 0.5195,
+      "step": 3055
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.70610033920655e-07,
+      "loss": 0.5207,
+      "step": 3056
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.683285045056891e-07,
+      "loss": 0.4488,
+      "step": 3057
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.660512705901189e-07,
+      "loss": 0.5296,
+      "step": 3058
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.637783343812031e-07,
+      "loss": 0.4785,
+      "step": 3059
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.615096980820379e-07,
+      "loss": 0.5391,
+      "step": 3060
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.592453638915507e-07,
+      "loss": 0.457,
+      "step": 3061
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.569853340045006e-07,
+      "loss": 0.5382,
+      "step": 3062
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.547296106114741e-07,
+      "loss": 0.4864,
+      "step": 3063
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.5247819589888e-07,
+      "loss": 0.5142,
+      "step": 3064
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.502310920489562e-07,
+      "loss": 0.4448,
+      "step": 3065
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.479883012397585e-07,
+      "loss": 0.5081,
+      "step": 3066
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.457498256451637e-07,
+      "loss": 0.5884,
+      "step": 3067
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.435156674348674e-07,
+      "loss": 0.5625,
+      "step": 3068
+    },
+    {
+      "epoch": 2.59,
+      "eval_loss": 1.0249528884887695,
+      "eval_runtime": 267.9778,
+      "eval_samples_per_second": 3.728,
+      "eval_steps_per_second": 0.466,
+      "step": 3068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.412858287743749e-07,
+      "loss": 0.518,
+      "step": 3069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.390603118250126e-07,
+      "loss": 0.5198,
+      "step": 3070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.36839118743912e-07,
+      "loss": 0.4423,
+      "step": 3071
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.34622251684015e-07,
+      "loss": 0.472,
+      "step": 3072
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.32409712794072e-07,
+      "loss": 0.3859,
+      "step": 3073
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.302015042186364e-07,
+      "loss": 0.4499,
+      "step": 3074
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.279976280980664e-07,
+      "loss": 0.5935,
+      "step": 3075
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.257980865685181e-07,
+      "loss": 0.5091,
+      "step": 3076
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.236028817619477e-07,
+      "loss": 0.5317,
+      "step": 3077
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.214120158061087e-07,
+      "loss": 0.472,
+      "step": 3078
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.192254908245476e-07,
+      "loss": 0.4844,
+      "step": 3079
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.170433089366056e-07,
+      "loss": 0.4945,
+      "step": 3080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.148654722574087e-07,
+      "loss": 0.4894,
+      "step": 3081
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.126919828978799e-07,
+      "loss": 0.4149,
+      "step": 3082
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.105228429647207e-07,
+      "loss": 0.4765,
+      "step": 3083
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.083580545604206e-07,
+      "loss": 0.524,
+      "step": 3084
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.061976197832508e-07,
+      "loss": 0.4888,
+      "step": 3085
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.040415407272614e-07,
+      "loss": 0.4611,
+      "step": 3086
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.018898194822835e-07,
+      "loss": 0.5275,
+      "step": 3087
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.9974245813392e-07,
+      "loss": 0.4446,
+      "step": 3088
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.975994587635507e-07,
+      "loss": 0.634,
+      "step": 3089
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.954608234483282e-07,
+      "loss": 0.4069,
+      "step": 3090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.933265542611726e-07,
+      "loss": 0.3313,
+      "step": 3091
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.911966532707751e-07,
+      "loss": 0.4865,
+      "step": 3092
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.890711225415884e-07,
+      "loss": 0.4666,
+      "step": 3093
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.869499641338348e-07,
+      "loss": 0.4486,
+      "step": 3094
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.848331801034945e-07,
+      "loss": 0.6061,
+      "step": 3095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.827207725023087e-07,
+      "loss": 0.6323,
+      "step": 3096
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.806127433777774e-07,
+      "loss": 0.4563,
+      "step": 3097
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.785090947731558e-07,
+      "loss": 0.4612,
+      "step": 3098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.7640982872745543e-07,
+      "loss": 0.525,
+      "step": 3099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.74314947275436e-07,
+      "loss": 0.4474,
+      "step": 3100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.722244524476105e-07,
+      "loss": 0.4691,
+      "step": 3101
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.701383462702386e-07,
+      "loss": 0.4899,
+      "step": 3102
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.680566307653267e-07,
+      "loss": 0.5195,
+      "step": 3103
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.659793079506264e-07,
+      "loss": 0.4919,
+      "step": 3104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.639063798396276e-07,
+      "loss": 0.4809,
+      "step": 3105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.618378484415681e-07,
+      "loss": 0.4408,
+      "step": 3106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5977371576141573e-07,
+      "loss": 0.4554,
+      "step": 3107
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.577139837998795e-07,
+      "loss": 0.4295,
+      "step": 3108
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.5565865455340255e-07,
+      "loss": 0.4696,
+      "step": 3109
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.536077300141589e-07,
+      "loss": 0.4247,
+      "step": 3110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.515612121700563e-07,
+      "loss": 0.4655,
+      "step": 3111
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.495191030047269e-07,
+      "loss": 0.4188,
+      "step": 3112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.4748140449753285e-07,
+      "loss": 0.5216,
+      "step": 3113
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.454481186235604e-07,
+      "loss": 0.4272,
+      "step": 3114
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.434192473536181e-07,
+      "loss": 0.4673,
+      "step": 3115
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.413947926542372e-07,
+      "loss": 0.5403,
+      "step": 3116
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.393747564876644e-07,
+      "loss": 0.4662,
+      "step": 3117
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.373591408118694e-07,
+      "loss": 0.4922,
+      "step": 3118
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.3534794758053167e-07,
+      "loss": 0.4341,
+      "step": 3119
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.333411787430469e-07,
+      "loss": 0.4805,
+      "step": 3120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.3133883624452254e-07,
+      "loss": 0.4191,
+      "step": 3121
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.293409220257744e-07,
+      "loss": 0.5396,
+      "step": 3122
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.2734743802332834e-07,
+      "loss": 0.4228,
+      "step": 3123
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.2535838616941163e-07,
+      "loss": 0.5475,
+      "step": 3124
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.2337376839196034e-07,
+      "loss": 0.4685,
+      "step": 3125
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.2139358661461073e-07,
+      "loss": 0.4812,
+      "step": 3126
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1941784275669915e-07,
+      "loss": 0.5514,
+      "step": 3127
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1744653873326156e-07,
+      "loss": 0.3934,
+      "step": 3128
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.154796764550273e-07,
+      "loss": 0.6156,
+      "step": 3129
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1351725782842577e-07,
+      "loss": 0.5224,
+      "step": 3130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1155928475557436e-07,
+      "loss": 0.5489,
+      "step": 3131
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.0960575913428447e-07,
+      "loss": 0.5414,
+      "step": 3132
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.076566828580547e-07,
+      "loss": 0.4597,
+      "step": 3133
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.0571205781607283e-07,
+      "loss": 0.4267,
+      "step": 3134
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.0377188589321215e-07,
+      "loss": 0.4889,
+      "step": 3135
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.0183616897002676e-07,
+      "loss": 0.441,
+      "step": 3136
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.999049089227558e-07,
+      "loss": 0.4472,
+      "step": 3137
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9797810762331745e-07,
+      "loss": 0.4953,
+      "step": 3138
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.960557669393078e-07,
+      "loss": 0.5263,
+      "step": 3139
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.941378887340014e-07,
+      "loss": 0.5414,
+      "step": 3140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.922244748663423e-07,
+      "loss": 0.4737,
+      "step": 3141
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.903155271909548e-07,
+      "loss": 0.4612,
+      "step": 3142
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8841104755812706e-07,
+      "loss": 0.4818,
+      "step": 3143
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8651103781382204e-07,
+      "loss": 0.5213,
+      "step": 3144
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8461549979966596e-07,
+      "loss": 0.5151,
+      "step": 3145
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.827244353529541e-07,
+      "loss": 0.5094,
+      "step": 3146
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.8083784630664364e-07,
+      "loss": 0.4404,
+      "step": 3147
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7895573448935405e-07,
+      "loss": 0.4559,
+      "step": 3148
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7707810172536496e-07,
+      "loss": 0.5652,
+      "step": 3149
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7520494983461606e-07,
+      "loss": 0.5491,
+      "step": 3150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.733362806327023e-07,
+      "loss": 0.5043,
+      "step": 3151
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.714720959308754e-07,
+      "loss": 0.4743,
+      "step": 3152
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.696123975360366e-07,
+      "loss": 0.4356,
+      "step": 3153
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6775718725074406e-07,
+      "loss": 0.516,
+      "step": 3154
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.659064668732015e-07,
+      "loss": 0.5003,
+      "step": 3155
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6406023819726244e-07,
+      "loss": 0.4415,
+      "step": 3156
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.622185030124259e-07,
+      "loss": 0.4535,
+      "step": 3157
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.6038126310383727e-07,
+      "loss": 0.5294,
+      "step": 3158
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.585485202522832e-07,
+      "loss": 0.5562,
+      "step": 3159
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.5672027623419035e-07,
+      "loss": 0.3979,
+      "step": 3160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.548965328216275e-07,
+      "loss": 0.5731,
+      "step": 3161
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.530772917823e-07,
+      "loss": 0.5566,
+      "step": 3162
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.512625548795479e-07,
+      "loss": 0.5175,
+      "step": 3163
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4945232387234907e-07,
+      "loss": 0.4166,
+      "step": 3164
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.476466005153073e-07,
+      "loss": 0.491,
+      "step": 3165
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.458453865586664e-07,
+      "loss": 0.4235,
+      "step": 3166
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4404868374829035e-07,
+      "loss": 0.5613,
+      "step": 3167
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4225649382567674e-07,
+      "loss": 0.5031,
+      "step": 3168
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4046881852794587e-07,
+      "loss": 0.5704,
+      "step": 3169
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.386856595878435e-07,
+      "loss": 0.4863,
+      "step": 3170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.3690701873373864e-07,
+      "loss": 0.5022,
+      "step": 3171
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.351328976896173e-07,
+      "loss": 0.4712,
+      "step": 3172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.333632981750884e-07,
+      "loss": 0.3747,
+      "step": 3173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.315982219053765e-07,
+      "loss": 0.5137,
+      "step": 3174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.298376705913231e-07,
+      "loss": 0.6019,
+      "step": 3175
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.280816459393832e-07,
+      "loss": 0.4282,
+      "step": 3176
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2633014965162133e-07,
+      "loss": 0.4757,
+      "step": 3177
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2458318342571957e-07,
+      "loss": 0.537,
+      "step": 3178
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2284074895496174e-07,
+      "loss": 0.4391,
+      "step": 3179
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.2110284792824407e-07,
+      "loss": 0.5089,
+      "step": 3180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.193694820300669e-07,
+      "loss": 0.5245,
+      "step": 3181
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.17640652940534e-07,
+      "loss": 0.5312,
+      "step": 3182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1591636233535503e-07,
+      "loss": 0.4589,
+      "step": 3183
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.141966118858353e-07,
+      "loss": 0.4787,
+      "step": 3184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.124814032588841e-07,
+      "loss": 0.4984,
+      "step": 3185
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1077073811700664e-07,
+      "loss": 0.5489,
+      "step": 3186
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0906461811830436e-07,
+      "loss": 0.5222,
+      "step": 3187
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.073630449164733e-07,
+      "loss": 0.4779,
+      "step": 3188
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.056660201608014e-07,
+      "loss": 0.5144,
+      "step": 3189
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0397354549617066e-07,
+      "loss": 0.4309,
+      "step": 3190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0228562256304995e-07,
+      "loss": 0.3949,
+      "step": 3191
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0060225299749723e-07,
+      "loss": 0.4347,
+      "step": 3192
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9892343843115736e-07,
+      "loss": 0.4841,
+      "step": 3193
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.972491804912597e-07,
+      "loss": 0.4626,
+      "step": 3194
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9557948080061795e-07,
+      "loss": 0.5827,
+      "step": 3195
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.939143409776263e-07,
+      "loss": 0.4702,
+      "step": 3196
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9225376263625984e-07,
+      "loss": 0.4397,
+      "step": 3197
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.905977473860716e-07,
+      "loss": 0.6023,
+      "step": 3198
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.889462968321938e-07,
+      "loss": 0.5398,
+      "step": 3199
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8729941257533254e-07,
+      "loss": 0.521,
+      "step": 3200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8565709621176605e-07,
+      "loss": 0.4824,
+      "step": 3201
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.84019349333351e-07,
+      "loss": 0.4417,
+      "step": 3202
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8238617352750794e-07,
+      "loss": 0.5616,
+      "step": 3203
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8075757037723137e-07,
+      "loss": 0.5212,
+      "step": 3204
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.791335414610818e-07,
+      "loss": 0.3568,
+      "step": 3205
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.775140883531874e-07,
+      "loss": 0.5487,
+      "step": 3206
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7589921262324015e-07,
+      "loss": 0.6421,
+      "step": 3207
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7428891583649473e-07,
+      "loss": 0.5625,
+      "step": 3208
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.726831995537682e-07,
+      "loss": 0.5468,
+      "step": 3209
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.710820653314389e-07,
+      "loss": 0.5361,
+      "step": 3210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.6948551472144204e-07,
+      "loss": 0.42,
+      "step": 3211
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.67893549271272e-07,
+      "loss": 0.5926,
+      "step": 3212
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.663061705239761e-07,
+      "loss": 0.4011,
+      "step": 3213
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.647233800181598e-07,
+      "loss": 0.3419,
+      "step": 3214
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.63145179287978e-07,
+      "loss": 0.5101,
+      "step": 3215
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.615715698631377e-07,
+      "loss": 0.3145,
+      "step": 3216
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.600025532688966e-07,
+      "loss": 0.5107,
+      "step": 3217
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.5843813102605974e-07,
+      "loss": 0.6414,
+      "step": 3218
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.568783046509804e-07,
+      "loss": 0.4938,
+      "step": 3219
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.553230756555547e-07,
+      "loss": 0.5579,
+      "step": 3220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.53772445547224e-07,
+      "loss": 0.4585,
+      "step": 3221
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.5222641582897335e-07,
+      "loss": 0.4857,
+      "step": 3222
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.506849879993267e-07,
+      "loss": 0.4459,
+      "step": 3223
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4914816355234906e-07,
+      "loss": 0.3833,
+      "step": 3224
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4761594397764155e-07,
+      "loss": 0.4267,
+      "step": 3225
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4608833076034456e-07,
+      "loss": 0.7007,
+      "step": 3226
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4456532538113196e-07,
+      "loss": 0.4774,
+      "step": 3227
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4304692931621077e-07,
+      "loss": 0.5667,
+      "step": 3228
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4153314403732254e-07,
+      "loss": 0.5708,
+      "step": 3229
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.4002397101173745e-07,
+      "loss": 0.5564,
+      "step": 3230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.385194117022577e-07,
+      "loss": 0.5287,
+      "step": 3231
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3701946756720996e-07,
+      "loss": 0.4597,
+      "step": 3232
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3552414006045022e-07,
+      "loss": 0.5251,
+      "step": 3233
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.3403343063135908e-07,
+      "loss": 0.4928,
+      "step": 3234
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.3254734072484076e-07,
+      "loss": 0.4772,
+      "step": 3235
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.3106587178132357e-07,
+      "loss": 0.468,
+      "step": 3236
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2958902523675218e-07,
+      "loss": 0.5419,
+      "step": 3237
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2811680252259715e-07,
+      "loss": 0.4605,
+      "step": 3238
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2664920506584253e-07,
+      "loss": 0.4136,
+      "step": 3239
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.25186234288991e-07,
+      "loss": 0.5544,
+      "step": 3240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2372789161006103e-07,
+      "loss": 0.547,
+      "step": 3241
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.222741784425847e-07,
+      "loss": 0.5703,
+      "step": 3242
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2082509619560656e-07,
+      "loss": 0.4559,
+      "step": 3243
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.1938064627368305e-07,
+      "loss": 0.3977,
+      "step": 3244
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.179408300768804e-07,
+      "loss": 0.4803,
+      "step": 3245
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.165056490007733e-07,
+      "loss": 0.4679,
+      "step": 3246
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.150751044364441e-07,
+      "loss": 0.4447,
+      "step": 3247
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1364919777048188e-07,
+      "loss": 0.3979,
+      "step": 3248
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.1222793038497724e-07,
+      "loss": 0.455,
+      "step": 3249
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.108113036575288e-07,
+      "loss": 0.5195,
+      "step": 3250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.093993189612331e-07,
+      "loss": 0.4396,
+      "step": 3251
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.0799197766468883e-07,
+      "loss": 0.4476,
+      "step": 3252
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.0658928113199416e-07,
+      "loss": 0.4522,
+      "step": 3253
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.0519123072274493e-07,
+      "loss": 0.5543,
+      "step": 3254
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.0379782779203382e-07,
+      "loss": 0.4663,
+      "step": 3255
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.024090736904477e-07,
+      "loss": 0.496,
+      "step": 3256
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.0102496976406925e-07,
+      "loss": 0.4332,
+      "step": 3257
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9964551735447313e-07,
+      "loss": 0.4956,
+      "step": 3258
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.982707177987242e-07,
+      "loss": 0.5957,
+      "step": 3259
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9690057242937988e-07,
+      "loss": 0.3777,
+      "step": 3260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9553508257448338e-07,
+      "loss": 0.6038,
+      "step": 3261
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9417424955756937e-07,
+      "loss": 0.4965,
+      "step": 3262
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.928180746976549e-07,
+      "loss": 0.6183,
+      "step": 3263
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9146655930924352e-07,
+      "loss": 0.4696,
+      "step": 3264
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.9011970470232287e-07,
+      "loss": 0.4889,
+      "step": 3265
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8877751218236318e-07,
+      "loss": 0.4513,
+      "step": 3266
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.874399830503154e-07,
+      "loss": 0.5849,
+      "step": 3267
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8610711860260922e-07,
+      "loss": 0.3558,
+      "step": 3268
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.847789201311556e-07,
+      "loss": 0.4263,
+      "step": 3269
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8345538892333982e-07,
+      "loss": 0.4627,
+      "step": 3270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8213652626202672e-07,
+      "loss": 0.4043,
+      "step": 3271
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.808223334255532e-07,
+      "loss": 0.4974,
+      "step": 3272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7951281168772982e-07,
+      "loss": 0.482,
+      "step": 3273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7820796231784342e-07,
+      "loss": 0.4626,
+      "step": 3274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7690778658064622e-07,
+      "loss": 0.574,
+      "step": 3275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7561228573636568e-07,
+      "loss": 0.5424,
+      "step": 3276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7432146104069402e-07,
+      "loss": 0.4338,
+      "step": 3277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7303531374479375e-07,
+      "loss": 0.549,
+      "step": 3278
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7175384509529326e-07,
+      "loss": 0.5065,
+      "step": 3279
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.704770563342839e-07,
+      "loss": 0.405,
+      "step": 3280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.692049486993236e-07,
+      "loss": 0.5236,
+      "step": 3281
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.679375234234315e-07,
+      "loss": 0.5345,
+      "step": 3282
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6667478173508833e-07,
+      "loss": 0.5593,
+      "step": 3283
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.654167248582367e-07,
+      "loss": 0.4539,
+      "step": 3284
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6416335401227502e-07,
+      "loss": 0.528,
+      "step": 3285
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6291467041206487e-07,
+      "loss": 0.5747,
+      "step": 3286
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.616706752679187e-07,
+      "loss": 0.671,
+      "step": 3287
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6043136978560803e-07,
+      "loss": 0.4463,
+      "step": 3288
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5919675516635923e-07,
+      "loss": 0.6078,
+      "step": 3289
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5796683260684953e-07,
+      "loss": 0.4604,
+      "step": 3290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.567416032992114e-07,
+      "loss": 0.4087,
+      "step": 3291
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5552106843102432e-07,
+      "loss": 0.4983,
+      "step": 3292
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5430522918532143e-07,
+      "loss": 0.4353,
+      "step": 3293
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5309408674058169e-07,
+      "loss": 0.4532,
+      "step": 3294
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5188764227073326e-07,
+      "loss": 0.4532,
+      "step": 3295
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.5068589694515078e-07,
+      "loss": 0.4205,
+      "step": 3296
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4948885192865192e-07,
+      "loss": 0.4349,
+      "step": 3297
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4829650838150246e-07,
+      "loss": 0.5518,
+      "step": 3298
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4710886745940623e-07,
+      "loss": 0.511,
+      "step": 3299
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4592593031351298e-07,
+      "loss": 0.4652,
+      "step": 3300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4474769809041112e-07,
+      "loss": 0.525,
+      "step": 3301
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4357417193212932e-07,
+      "loss": 0.4208,
+      "step": 3302
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4240535297613545e-07,
+      "loss": 0.5259,
+      "step": 3303
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4124124235533333e-07,
+      "loss": 0.4486,
+      "step": 3304
+    },
+    {
+      "epoch": 2.79,
+      "eval_loss": 1.0262025594711304,
+      "eval_runtime": 267.6698,
+      "eval_samples_per_second": 3.732,
+      "eval_steps_per_second": 0.467,
+      "step": 3304
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4008184119806423e-07,
+      "loss": 0.4948,
+      "step": 3305
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3892715062810426e-07,
+      "loss": 0.4924,
+      "step": 3306
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3777717176466364e-07,
+      "loss": 0.4513,
+      "step": 3307
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3663190572238639e-07,
+      "loss": 0.4594,
+      "step": 3308
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3549135361134724e-07,
+      "loss": 0.5938,
+      "step": 3309
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3435551653705304e-07,
+      "loss": 0.4476,
+      "step": 3310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3322439560043977e-07,
+      "loss": 0.5206,
+      "step": 3311
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.320979918978721e-07,
+      "loss": 0.437,
+      "step": 3312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3097630652114334e-07,
+      "loss": 0.5283,
+      "step": 3313
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.2985934055747218e-07,
+      "loss": 0.3727,
+      "step": 3314
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.2874709508950423e-07,
+      "loss": 0.4486,
+      "step": 3315
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.2763957119530767e-07,
+      "loss": 0.5639,
+      "step": 3316
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.265367699483766e-07,
+      "loss": 0.4592,
+      "step": 3317
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.254386924176265e-07,
+      "loss": 0.3462,
+      "step": 3318
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.243453396673938e-07,
+      "loss": 0.478,
+      "step": 3319
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2325671275743578e-07,
+      "loss": 0.5261,
+      "step": 3320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2217281274292892e-07,
+      "loss": 0.5214,
+      "step": 3321
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2109364067446894e-07,
+      "loss": 0.4817,
+      "step": 3322
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2001919759806745e-07,
+      "loss": 0.5907,
+      "step": 3323
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.189494845551531e-07,
+      "loss": 0.4028,
+      "step": 3324
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1788450258257034e-07,
+      "loss": 0.5612,
+      "step": 3325
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.168242527125768e-07,
+      "loss": 0.4725,
+      "step": 3326
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1576873597284544e-07,
+      "loss": 0.5152,
+      "step": 3327
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.1471795338645785e-07,
+      "loss": 0.438,
+      "step": 3328
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1367190597191047e-07,
+      "loss": 0.5735,
+      "step": 3329
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.126305947431089e-07,
+      "loss": 0.461,
+      "step": 3330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1159402070936743e-07,
+      "loss": 0.4622,
+      "step": 3331
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.1056218487541015e-07,
+      "loss": 0.4246,
+      "step": 3332
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0953508824136594e-07,
+      "loss": 0.4665,
+      "step": 3333
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.08512731802774e-07,
+      "loss": 0.3827,
+      "step": 3334
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0749511655057498e-07,
+      "loss": 0.613,
+      "step": 3335
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0648224347111713e-07,
+      "loss": 0.4738,
+      "step": 3336
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0547411354614955e-07,
+      "loss": 0.5107,
+      "step": 3337
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0447072775282674e-07,
+      "loss": 0.5234,
+      "step": 3338
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.0347208706370348e-07,
+      "loss": 0.5767,
+      "step": 3339
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.024781924467333e-07,
+      "loss": 0.5835,
+      "step": 3340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0148904486527333e-07,
+      "loss": 0.3699,
+      "step": 3341
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.0050464527807669e-07,
+      "loss": 0.5869,
+      "step": 3342
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.95249946392951e-08,
+      "loss": 0.3781,
+      "step": 3343
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.855009389847903e-08,
+      "loss": 0.4064,
+      "step": 3344
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.757994400057091e-08,
+      "loss": 0.48,
+      "step": 3345
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.661454588591302e-08,
+      "loss": 0.467,
+      "step": 3346
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.565390049023793e-08,
+      "loss": 0.536,
+      "step": 3347
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.469800874467471e-08,
+      "loss": 0.4559,
+      "step": 3348
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.374687157574281e-08,
+      "loss": 0.5745,
+      "step": 3349
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.280048990535307e-08,
+      "loss": 0.4428,
+      "step": 3350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.18588646508095e-08,
+      "loss": 0.4924,
+      "step": 3351
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.092199672480207e-08,
+      "loss": 0.5587,
+      "step": 3352
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.998988703541212e-08,
+      "loss": 0.5237,
+      "step": 3353
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.906253648610918e-08,
+      "loss": 0.4808,
+      "step": 3354
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.81399459757476e-08,
+      "loss": 0.4539,
+      "step": 3355
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.722211639857148e-08,
+      "loss": 0.5635,
+      "step": 3356
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.630904864420586e-08,
+      "loss": 0.5079,
+      "step": 3357
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.540074359766448e-08,
+      "loss": 0.5764,
+      "step": 3358
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.449720213934199e-08,
+      "loss": 0.5438,
+      "step": 3359
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.35984251450167e-08,
+      "loss": 0.5626,
+      "step": 3360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.270441348584956e-08,
+      "loss": 0.3988,
+      "step": 3361
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.181516802838075e-08,
+      "loss": 0.4588,
+      "step": 3362
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.093068963453354e-08,
+      "loss": 0.588,
+      "step": 3363
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.00509791616072e-08,
+      "loss": 0.4485,
+      "step": 3364
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.917603746228186e-08,
+      "loss": 0.5431,
+      "step": 3365
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.830586538461527e-08,
+      "loss": 0.5895,
+      "step": 3366
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.744046377204107e-08,
+      "loss": 0.432,
+      "step": 3367
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.65798334633705e-08,
+      "loss": 0.5295,
+      "step": 3368
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.572397529278852e-08,
+      "loss": 0.5776,
+      "step": 3369
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.487289008985599e-08,
+      "loss": 0.425,
+      "step": 3370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.40265786795058e-08,
+      "loss": 0.554,
+      "step": 3371
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.318504188204567e-08,
+      "loss": 0.5892,
+      "step": 3372
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.234828051315423e-08,
+      "loss": 0.4664,
+      "step": 3373
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.151629538388105e-08,
+      "loss": 0.3965,
+      "step": 3374
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.068908730064772e-08,
+      "loss": 0.4093,
+      "step": 3375
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.986665706524454e-08,
+      "loss": 0.5605,
+      "step": 3376
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.904900547482996e-08,
+      "loss": 0.38,
+      "step": 3377
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.82361333219328e-08,
+      "loss": 0.5629,
+      "step": 3378
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.742804139444725e-08,
+      "loss": 0.4505,
+      "step": 3379
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.662473047563511e-08,
+      "loss": 0.6066,
+      "step": 3380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.58262013441241e-08,
+      "loss": 0.5168,
+      "step": 3381
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.50324547739073e-08,
+      "loss": 0.4895,
+      "step": 3382
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.424349153434095e-08,
+      "loss": 0.5239,
+      "step": 3383
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.345931239014613e-08,
+      "loss": 0.4465,
+      "step": 3384
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.267991810140706e-08,
+      "loss": 0.4633,
+      "step": 3385
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.190530942356888e-08,
+      "loss": 0.5272,
+      "step": 3386
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.113548710743933e-08,
+      "loss": 0.5345,
+      "step": 3387
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.0370451899186e-08,
+      "loss": 0.5397,
+      "step": 3388
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.961020454033628e-08,
+      "loss": 0.5274,
+      "step": 3389
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.88547457677785e-08,
+      "loss": 0.484,
+      "step": 3390
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.810407631375803e-08,
+      "loss": 0.3965,
+      "step": 3391
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.735819690587896e-08,
+      "loss": 0.4793,
+      "step": 3392
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.661710826710076e-08,
+      "loss": 0.5868,
+      "step": 3393
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.588081111574162e-08,
+      "loss": 0.452,
+      "step": 3394
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.514930616547398e-08,
+      "loss": 0.5533,
+      "step": 3395
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.442259412532625e-08,
+      "loss": 0.3947,
+      "step": 3396
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.370067569968051e-08,
+      "loss": 0.5192,
+      "step": 3397
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.298355158827262e-08,
+      "loss": 0.4752,
+      "step": 3398
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.2271222486192653e-08,
+      "loss": 0.5408,
+      "step": 3399
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.156368908388054e-08,
+      "loss": 0.4265,
+      "step": 3400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.086095206712938e-08,
+      "loss": 0.4466,
+      "step": 3401
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.01630121170843e-08,
+      "loss": 0.3721,
+      "step": 3402
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.946986991023861e-08,
+      "loss": 0.5702,
+      "step": 3403
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.878152611843712e-08,
+      "loss": 0.3736,
+      "step": 3404
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.809798140887167e-08,
+      "loss": 0.4847,
+      "step": 3405
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.741923644408508e-08,
+      "loss": 0.4626,
+      "step": 3406
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.674529188196553e-08,
+      "loss": 0.4299,
+      "step": 3407
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.6076148375749363e-08,
+      "loss": 0.3681,
+      "step": 3408
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.541180657402e-08,
+      "loss": 0.4494,
+      "step": 3409
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.475226712070513e-08,
+      "loss": 0.5246,
+      "step": 3410
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.4097530655078934e-08,
+      "loss": 0.4811,
+      "step": 3411
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.344759781175989e-08,
+      "loss": 0.4149,
+      "step": 3412
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2802469220710186e-08,
+      "loss": 0.5133,
+      "step": 3413
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2162145507235186e-08,
+      "loss": 0.5152,
+      "step": 3414
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.152662729198398e-08,
+      "loss": 0.5636,
+      "step": 3415
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.08959151909466e-08,
+      "loss": 0.382,
+      "step": 3416
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.02700098154557e-08,
+      "loss": 0.413,
+      "step": 3417
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9648911772184326e-08,
+      "loss": 0.4134,
+      "step": 3418
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.903262166314592e-08,
+      "loss": 0.5376,
+      "step": 3419
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.8421140085693756e-08,
+      "loss": 0.3832,
+      "step": 3420
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7814467632520944e-08,
+      "loss": 0.5522,
+      "step": 3421
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.721260489165768e-08,
+      "loss": 0.4662,
+      "step": 3422
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6615552446474525e-08,
+      "loss": 0.5096,
+      "step": 3423
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.60233108756769e-08,
+      "loss": 0.5126,
+      "step": 3424
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.54358807533095e-08,
+      "loss": 0.3947,
+      "step": 3425
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.485326264875244e-08,
+      "loss": 0.5099,
+      "step": 3426
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.42754571267212e-08,
+      "loss": 0.5096,
+      "step": 3427
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3702464747267794e-08,
+      "loss": 0.4603,
+      "step": 3428
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.313428606577684e-08,
+      "loss": 0.4226,
+      "step": 3429
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2570921632970045e-08,
+      "loss": 0.5243,
+      "step": 3430
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.201237199490115e-08,
+      "loss": 0.4785,
+      "step": 3431
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.145863769295654e-08,
+      "loss": 0.5336,
+      "step": 3432
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.0909719263855776e-08,
+      "loss": 0.4394,
+      "step": 3433
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.036561723965159e-08,
+      "loss": 0.5654,
+      "step": 3434
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.9826332147727124e-08,
+      "loss": 0.4153,
+      "step": 3435
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9291864510796464e-08,
+      "loss": 0.3756,
+      "step": 3436
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.8762214846904667e-08,
+      "loss": 0.4411,
+      "step": 3437
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.8237383669426632e-08,
+      "loss": 0.3529,
+      "step": 3438
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.7717371487067657e-08,
+      "loss": 0.442,
+      "step": 3439
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.7202178803861225e-08,
+      "loss": 0.5288,
+      "step": 3440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.669180611916955e-08,
+      "loss": 0.5963,
+      "step": 3441
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.618625392768359e-08,
+      "loss": 0.6037,
+      "step": 3442
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.5685522719420263e-08,
+      "loss": 0.4245,
+      "step": 3443
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.5189612979725775e-08,
+      "loss": 0.455,
+      "step": 3444
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.4698525189272294e-08,
+      "loss": 0.3708,
+      "step": 3445
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.421225982405684e-08,
+      "loss": 0.5112,
+      "step": 3446
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.3730817355404613e-08,
+      "loss": 0.5424,
+      "step": 3447
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.3254198249963444e-08,
+      "loss": 0.5155,
+      "step": 3448
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.2782402969707684e-08,
+      "loss": 0.416,
+      "step": 3449
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.2315431971936528e-08,
+      "loss": 0.4492,
+      "step": 3450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.185328570927181e-08,
+      "loss": 0.5769,
+      "step": 3451
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.139596462965965e-08,
+      "loss": 0.4917,
+      "step": 3452
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.094346917636825e-08,
+      "loss": 0.4141,
+      "step": 3453
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.04957997879901e-08,
+      "loss": 0.5499,
+      "step": 3454
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.005295689843867e-08,
+      "loss": 0.5391,
+      "step": 3455
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9614940936950045e-08,
+      "loss": 0.4776,
+      "step": 3456
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9181752328080173e-08,
+      "loss": 0.3767,
+      "step": 3457
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8753391491708183e-08,
+      "loss": 0.4876,
+      "step": 3458
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8329858843031954e-08,
+      "loss": 0.4125,
+      "step": 3459
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.7911154792570883e-08,
+      "loss": 0.5023,
+      "step": 3460
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.7497279746162554e-08,
+      "loss": 0.4518,
+      "step": 3461
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.7088234104965516e-08,
+      "loss": 0.4985,
+      "step": 3462
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.668401826545707e-08,
+      "loss": 0.5189,
+      "step": 3463
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.628463261943214e-08,
+      "loss": 0.5176,
+      "step": 3464
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5890077554003848e-08,
+      "loss": 0.5005,
+      "step": 3465
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5500353451604612e-08,
+      "loss": 0.5135,
+      "step": 3466
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5115460689983375e-08,
+      "loss": 0.4636,
+      "step": 3467
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.473539964220616e-08,
+      "loss": 0.4004,
+      "step": 3468
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.436017067665607e-08,
+      "loss": 0.5472,
+      "step": 3469
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.398977415703162e-08,
+      "loss": 0.6167,
+      "step": 3470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3624210442349518e-08,
+      "loss": 0.4049,
+      "step": 3471
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.3263479886939657e-08,
+      "loss": 0.429,
+      "step": 3472
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.2907582840448463e-08,
+      "loss": 0.4993,
+      "step": 3473
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.2556519647837772e-08,
+      "loss": 0.5494,
+      "step": 3474
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.2210290649383172e-08,
+      "loss": 0.6189,
+      "step": 3475
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1868896180676214e-08,
+      "loss": 0.3963,
+      "step": 3476
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1532336572619984e-08,
+      "loss": 0.4054,
+      "step": 3477
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1200612151432976e-08,
+      "loss": 0.5222,
+      "step": 3478
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0873723238646882e-08,
+      "loss": 0.4002,
+      "step": 3479
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0551670151106586e-08,
+      "loss": 0.5037,
+      "step": 3480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0234453200969052e-08,
+      "loss": 0.4106,
+      "step": 3481
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.922072695703888e-09,
+      "loss": 0.5389,
+      "step": 3482
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.614528938093891e-09,
+      "loss": 0.3591,
+      "step": 3483
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.311822226231726e-09,
+      "loss": 0.4545,
+      "step": 3484
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.013952853523799e-09,
+      "loss": 0.5415,
+      "step": 3485
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.720921108686386e-09,
+      "loss": 0.418,
+      "step": 3486
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.432727275747287e-09,
+      "loss": 0.4303,
+      "step": 3487
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.149371634045833e-09,
+      "loss": 0.4454,
+      "step": 3488
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.87085445822955e-09,
+      "loss": 0.4912,
+      "step": 3489
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.597176018258045e-09,
+      "loss": 0.3872,
+      "step": 3490
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.328336579401352e-09,
+      "loss": 0.4905,
+      "step": 3491
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.064336402237137e-09,
+      "loss": 0.4156,
+      "step": 3492
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.8051757426540465e-09,
+      "loss": 0.3909,
+      "step": 3493
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.55085485185003e-09,
+      "loss": 0.5846,
+      "step": 3494
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.301373976330682e-09,
+      "loss": 0.583,
+      "step": 3495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.056733357910904e-09,
+      "loss": 0.4829,
+      "step": 3496
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.81693323371546e-09,
+      "loss": 0.5801,
+      "step": 3497
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.581973836175092e-09,
+      "loss": 0.5683,
+      "step": 3498
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.351855393029848e-09,
+      "loss": 0.5476,
+      "step": 3499
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.126578127327419e-09,
+      "loss": 0.4826,
+      "step": 3500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.906142257422586e-09,
+      "loss": 0.6303,
+      "step": 3501
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.690547996978323e-09,
+      "loss": 0.5322,
+      "step": 3502
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4797955549641396e-09,
+      "loss": 0.4914,
+      "step": 3503
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.27388513565663e-09,
+      "loss": 0.4952,
+      "step": 3504
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.072816938639479e-09,
+      "loss": 0.4595,
+      "step": 3505
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.876591158802345e-09,
+      "loss": 0.5671,
+      "step": 3506
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6852079863419764e-09,
+      "loss": 0.5055,
+      "step": 3507
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4986676067605417e-09,
+      "loss": 0.4723,
+      "step": 3508
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3169702008661876e-09,
+      "loss": 0.4893,
+      "step": 3509
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.1401159447741467e-09,
+      "loss": 0.4444,
+      "step": 3510
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9681050099039655e-09,
+      "loss": 0.4662,
+      "step": 3511
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.8009375629817205e-09,
+      "loss": 0.5104,
+      "step": 3512
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.6386137660372456e-09,
+      "loss": 0.4664,
+      "step": 3513
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.4811337764074628e-09,
+      "loss": 0.5505,
+      "step": 3514
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3284977467336045e-09,
+      "loss": 0.4008,
+      "step": 3515
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1807058249612155e-09,
+      "loss": 0.4267,
+      "step": 3516
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0377581543407076e-09,
+      "loss": 0.4036,
+      "step": 3517
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8996548734284694e-09,
+      "loss": 0.4208,
+      "step": 3518
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7663961160829802e-09,
+      "loss": 0.4496,
+      "step": 3519
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6379820114692525e-09,
+      "loss": 0.4497,
+      "step": 3520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5144126840554995e-09,
+      "loss": 0.605,
+      "step": 3521
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3956882536136918e-09,
+      "loss": 0.5382,
+      "step": 3522
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.281808835220666e-09,
+      "loss": 0.5168,
+      "step": 3523
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1727745392564604e-09,
+      "loss": 0.4064,
+      "step": 3524
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0685854714048705e-09,
+      "loss": 0.6107,
+      "step": 3525
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.692417326540027e-10,
+      "loss": 0.5718,
+      "step": 3526
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.747434192951654e-10,
+      "loss": 0.4866,
+      "step": 3527
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.850906229223132e-10,
+      "loss": 0.4478,
+      "step": 3528
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.002834304337125e-10,
+      "loss": 0.4479,
+      "step": 3529
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.203219240308311e-10,
+      "loss": 0.4205,
+      "step": 3530
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.452061812183385e-10,
+      "loss": 0.3997,
+      "step": 3531
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.749362748035502e-10,
+      "loss": 0.663,
+      "step": 3532
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.095122728975387e-10,
+      "loss": 0.5305,
+      "step": 3533
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.4893423891402267e-10,
+      "loss": 0.5667,
+      "step": 3534
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9320223156881213e-10,
+      "loss": 0.4295,
+      "step": 3535
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4231630488258386e-10,
+      "loss": 0.5766,
+      "step": 3536
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9627650817755085e-10,
+      "loss": 0.4847,
+      "step": 3537
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5508288607801737e-10,
+      "loss": 0.5777,
+      "step": 3538
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1873547851204425e-10,
+      "loss": 0.4278,
+      "step": 3539
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.723432071089389e-11,
+      "loss": 0.5706,
+      "step": 3540
+    },
+    {
+      "epoch": 2.99,
+      "eval_loss": 1.0262978076934814,
+      "eval_runtime": 267.8777,
+      "eval_samples_per_second": 3.729,
+      "eval_steps_per_second": 0.467,
+      "step": 3540
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.057944320736475e-11,
+      "loss": 0.5025,
+      "step": 3541
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.877087183745687e-11,
+      "loss": 0.4336,
+      "step": 3542
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1808627739261513e-11,
+      "loss": 0.503,
+      "step": 3543
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.692727354071451e-12,
+      "loss": 0.4442,
+      "step": 3544
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.423182425825843e-12,
+      "loss": 0.4248,
+      "step": 3545
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.4569,
+      "step": 3546
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3546,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 4.302766657351434e+18,
+  "trial_name": null,
+  "trial_params": null
+}