diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,59483 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.9667239809851322,
+  "eval_steps": 495,
+  "global_step": 9884,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.7885,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.5697730779647827,
+      "eval_runtime": 504.4845,
+      "eval_samples_per_second": 51.34,
+      "eval_steps_per_second": 25.67,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.75,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.7066,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.7431,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.787,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 1.7067,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014,
+      "loss": 1.6767,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 1.6348,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018,
+      "loss": 1.6035,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.5463,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999949384504,
+      "loss": 1.4953,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997975380195,
+      "loss": 1.6229,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999995444605634,
+      "loss": 1.5145,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999991901521605,
+      "loss": 1.5292,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999987346128469,
+      "loss": 1.4627,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999981778426684,
+      "loss": 1.5632,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999975198416818,
+      "loss": 1.4691,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999967606099532,
+      "loss": 1.4795,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199999590014756,
+      "loss": 1.4336,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999994938454589,
+      "loss": 1.4746,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999938755311378,
+      "loss": 1.5054,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999927113773137,
+      "loss": 1.4902,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999914459932345,
+      "loss": 1.464,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999900793790288,
+      "loss": 1.4368,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999886115348346,
+      "loss": 1.3998,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999870424608002,
+      "loss": 1.4172,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999985372157085,
+      "loss": 1.5098,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999836006238576,
+      "loss": 1.4316,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999817278612976,
+      "loss": 1.4072,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999797538695945,
+      "loss": 1.4374,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999776786489484,
+      "loss": 1.389,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999975502199569,
+      "loss": 1.4446,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999732245216766,
+      "loss": 1.4333,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999708456155024,
+      "loss": 1.4435,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999683654812868,
+      "loss": 1.446,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999657841192803,
+      "loss": 1.4288,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999631015297448,
+      "loss": 1.4139,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999603177129523,
+      "loss": 1.3743,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999574326691838,
+      "loss": 1.494,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999544463987316,
+      "loss": 1.445,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999513589018982,
+      "loss": 1.3324,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999481701789957,
+      "loss": 1.3983,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999448802303474,
+      "loss": 1.4288,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999414890562866,
+      "loss": 1.384,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999379966571557,
+      "loss": 1.4052,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999344030333086,
+      "loss": 1.4187,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999307081851098,
+      "loss": 1.3987,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999269121129322,
+      "loss": 1.317,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999230148171608,
+      "loss": 1.2973,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999190162981899,
+      "loss": 1.3961,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999149165564242,
+      "loss": 1.4202,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999107155922787,
+      "loss": 1.4244,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999906413406179,
+      "loss": 1.3501,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999020099985604,
+      "loss": 1.3944,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998975053698686,
+      "loss": 1.353,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998928995205594,
+      "loss": 1.3952,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998881924510997,
+      "loss": 1.3971,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998833841619655,
+      "loss": 1.4175,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998784746536437,
+      "loss": 1.3614,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999873463926631,
+      "loss": 1.3847,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998683519814353,
+      "loss": 1.3593,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998631388185736,
+      "loss": 1.3817,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998578244385736,
+      "loss": 1.4027,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998524088419737,
+      "loss": 1.3428,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998468920293217,
+      "loss": 1.3928,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999841274001176,
+      "loss": 1.3539,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998355547581057,
+      "loss": 1.3881,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998297343006894,
+      "loss": 1.3505,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998238126295168,
+      "loss": 1.3368,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999817789745187,
+      "loss": 1.3847,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998116656483097,
+      "loss": 1.3177,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998054403395048,
+      "loss": 1.3585,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997991138194027,
+      "loss": 1.3254,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997926860886438,
+      "loss": 1.3391,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997861571478787,
+      "loss": 1.3718,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997795269977684,
+      "loss": 1.4024,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997727956389837,
+      "loss": 1.389,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997659630722068,
+      "loss": 1.3831,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997590292981286,
+      "loss": 1.2967,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997519943174517,
+      "loss": 1.3612,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997448581308874,
+      "loss": 1.4101,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997376207391593,
+      "loss": 1.3403,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999730282142999,
+      "loss": 1.3805,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997228423431493,
+      "loss": 1.3983,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997153013403646,
+      "loss": 1.4454,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997076591354068,
+      "loss": 1.3416,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996999157290504,
+      "loss": 1.3257,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999692071122079,
+      "loss": 1.245,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996841253152868,
+      "loss": 1.3804,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996760783094782,
+      "loss": 1.339,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996679301054679,
+      "loss": 1.3592,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996596807040807,
+      "loss": 1.3102,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996513301061514,
+      "loss": 1.3006,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996428783125252,
+      "loss": 1.3577,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996343253240583,
+      "loss": 1.2859,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996256711416163,
+      "loss": 1.3205,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996169157660752,
+      "loss": 1.3778,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996080591983212,
+      "loss": 1.3072,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999599101439251,
+      "loss": 1.3892,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999590042489772,
+      "loss": 1.3417,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995808823508,
+      "loss": 1.4003,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999571621023263,
+      "loss": 1.3333,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995622585080986,
+      "loss": 1.3216,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995527948062544,
+      "loss": 1.3791,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995432299186884,
+      "loss": 1.3148,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999533563846369,
+      "loss": 1.2994,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995237965902745,
+      "loss": 1.354,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995139281513938,
+      "loss": 1.3351,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995039585307263,
+      "loss": 1.2914,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994938877292804,
+      "loss": 1.3807,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999483715748076,
+      "loss": 1.2714,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999473442588143,
+      "loss": 1.3096,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999463068250521,
+      "loss": 1.3022,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994525927362604,
+      "loss": 1.3576,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994420160464216,
+      "loss": 1.3719,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994313381820753,
+      "loss": 1.3284,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994205591443022,
+      "loss": 1.395,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994096789341942,
+      "loss": 1.3145,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993986975528518,
+      "loss": 1.3061,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993876150013875,
+      "loss": 1.2914,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993764312809225,
+      "loss": 1.3358,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993651463925892,
+      "loss": 1.353,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199935376033753,
+      "loss": 1.3418,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993422731168976,
+      "loss": 1.3239,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999330684731855,
+      "loss": 1.301,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993189951835746,
+      "loss": 1.3802,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993072044732405,
+      "loss": 1.3523,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992953126020461,
+      "loss": 1.3085,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999283319571195,
+      "loss": 1.3252,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992712253819015,
+      "loss": 1.3367,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992590300353896,
+      "loss": 1.3118,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992467335328945,
+      "loss": 1.3104,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992343358756601,
+      "loss": 1.2633,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999221837064942,
+      "loss": 1.3229,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992092371020056,
+      "loss": 1.3575,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991965359881259,
+      "loss": 1.3611,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991837337245887,
+      "loss": 1.2963,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991708303126907,
+      "loss": 1.3061,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991578257537372,
+      "loss": 1.2882,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991447200490452,
+      "loss": 1.3105,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999131513199941,
+      "loss": 1.2965,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999118205207762,
+      "loss": 1.2846,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991047960738555,
+      "loss": 1.3109,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990912857995784,
+      "loss": 1.3433,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990776743862988,
+      "loss": 1.3178,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990639618353942,
+      "loss": 1.321,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990501481482526,
+      "loss": 1.3938,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999036233326273,
+      "loss": 1.3238,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990222173708636,
+      "loss": 1.3275,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990081002834432,
+      "loss": 1.2627,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989938820654416,
+      "loss": 1.3621,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998979562718297,
+      "loss": 1.387,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989651422434596,
+      "loss": 1.3632,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989506206423893,
+      "loss": 1.3478,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989359979165558,
+      "loss": 1.346,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989212740674397,
+      "loss": 1.3437,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998906449096531,
+      "loss": 1.3129,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998891523005331,
+      "loss": 1.3109,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988764957953506,
+      "loss": 1.2669,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988613674681106,
+      "loss": 1.3085,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998846138025143,
+      "loss": 1.3639,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988308074679888,
+      "loss": 1.3465,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019988153757982007,
+      "loss": 1.4004,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987998430173406,
+      "loss": 1.2404,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987842091269807,
+      "loss": 1.3121,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987684741287038,
+      "loss": 1.2232,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987526380241028,
+      "loss": 1.3145,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987367008147806,
+      "loss": 1.3015,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987206625023503,
+      "loss": 1.3321,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987045230884364,
+      "loss": 1.3397,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998688282574672,
+      "loss": 1.2894,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986719409627012,
+      "loss": 1.24,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986554982541783,
+      "loss": 1.3674,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986389544507683,
+      "loss": 1.3213,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986223095541452,
+      "loss": 1.3072,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986055635659944,
+      "loss": 1.3609,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998588716488011,
+      "loss": 1.2811,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985717683219,
+      "loss": 1.243,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985547190693782,
+      "loss": 1.291,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985375687321707,
+      "loss": 1.3221,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985203173120136,
+      "loss": 1.2725,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985029648106537,
+      "loss": 1.211,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984855112298473,
+      "loss": 1.3202,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984679565713615,
+      "loss": 1.3774,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998450300836973,
+      "loss": 1.3143,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019984325440284694,
+      "loss": 1.2864,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998414686147648,
+      "loss": 1.361,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998396727196317,
+      "loss": 1.3023,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998378667176294,
+      "loss": 1.4114,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983605060894073,
+      "loss": 1.288,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983422439374956,
+      "loss": 1.3485,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983238807224073,
+      "loss": 1.3469,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983054164460015,
+      "loss": 1.3098,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982868511101473,
+      "loss": 1.2906,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982681847167242,
+      "loss": 1.2556,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982494172676217,
+      "loss": 1.231,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982305487647396,
+      "loss": 1.3123,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998211579209988,
+      "loss": 1.3257,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981925086052873,
+      "loss": 1.3817,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981733369525679,
+      "loss": 1.3837,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981540642537707,
+      "loss": 1.374,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981346905108467,
+      "loss": 1.2949,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981152157257573,
+      "loss": 1.2903,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980956399004733,
+      "loss": 1.3186,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998075963036977,
+      "loss": 1.3164,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199805618513726,
+      "loss": 1.3296,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980363062033245,
+      "loss": 1.2896,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998016326237183,
+      "loss": 1.3336,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001997996245240858,
+      "loss": 1.2898,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979760632163825,
+      "loss": 1.2896,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019979557801657992,
+      "loss": 1.2952,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019979353960911613,
+      "loss": 1.3279,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997914910994533,
+      "loss": 1.2662,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978943248779874,
+      "loss": 1.3744,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978736377436086,
+      "loss": 1.2922,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978528495934907,
+      "loss": 1.318,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978319604297383,
+      "loss": 1.2542,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978109702544657,
+      "loss": 1.3483,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977898790697984,
+      "loss": 1.2653,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997768686877871,
+      "loss": 1.3122,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977473936808285,
+      "loss": 1.3274,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997725999480827,
+      "loss": 1.3068,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997704504280032,
+      "loss": 1.3673,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976829080806197,
+      "loss": 1.2779,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976612108847764,
+      "loss": 1.3051,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997639412694698,
+      "loss": 1.2454,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997617513512591,
+      "loss": 1.3314,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975955133406736,
+      "loss": 1.2455,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975734121811716,
+      "loss": 1.3522,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975512100363225,
+      "loss": 1.3038,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975289069083742,
+      "loss": 1.322,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975065027995844,
+      "loss": 1.3199,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974839977122212,
+      "loss": 1.3187,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974613916485626,
+      "loss": 1.3023,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019974386846108967,
+      "loss": 1.3018,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019974158766015228,
+      "loss": 1.3496,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973929676227496,
+      "loss": 1.4142,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997369957676896,
+      "loss": 1.2917,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973468467662914,
+      "loss": 1.3225,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973236348932755,
+      "loss": 1.2973,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973003220601977,
+      "loss": 1.2988,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972769082694184,
+      "loss": 1.3309,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972533935233078,
+      "loss": 1.3369,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997229777824246,
+      "loss": 1.3007,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972060611746234,
+      "loss": 1.3084,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971822435768417,
+      "loss": 1.3819,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971583250333113,
+      "loss": 1.2561,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997134305546454,
+      "loss": 1.289,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971101851187006,
+      "loss": 1.3273,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970859637524933,
+      "loss": 1.3416,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970616414502845,
+      "loss": 1.2857,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970372182145357,
+      "loss": 1.356,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970126940477192,
+      "loss": 1.357,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969880689523178,
+      "loss": 1.3616,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969633429308247,
+      "loss": 1.3066,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969385159857427,
+      "loss": 1.3862,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996913588119585,
+      "loss": 1.2751,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968885593348747,
+      "loss": 1.3257,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968634296341464,
+      "loss": 1.2905,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996838199019943,
+      "loss": 1.271,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019968128674948194,
+      "loss": 1.3336,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967874350613395,
+      "loss": 1.3723,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996761901722078,
+      "loss": 1.3537,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967362674796197,
+      "loss": 1.2853,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967105323365593,
+      "loss": 1.3079,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966846962955024,
+      "loss": 1.2629,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966587593590643,
+      "loss": 1.2555,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966327215298704,
+      "loss": 1.3623,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966065828105567,
+      "loss": 1.3465,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965803432037694,
+      "loss": 1.2978,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965540027121643,
+      "loss": 1.2844,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965275613384082,
+      "loss": 1.267,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965010190851779,
+      "loss": 1.2744,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000199647437595516,
+      "loss": 1.3007,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996447631951052,
+      "loss": 1.2534,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964207870755608,
+      "loss": 1.238,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996393841331404,
+      "loss": 1.2361,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963667947213098,
+      "loss": 1.3005,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963396472480157,
+      "loss": 1.3269,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963123989142698,
+      "loss": 1.2984,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962850497228307,
+      "loss": 1.2677,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996257599676467,
+      "loss": 1.2651,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962300487779574,
+      "loss": 1.2801,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962023970300909,
+      "loss": 1.3015,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996174644435667,
+      "loss": 1.3411,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961467909974947,
+      "loss": 1.3715,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019961188367183935,
+      "loss": 1.2963,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996090781601194,
+      "loss": 1.2831,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960626256487358,
+      "loss": 1.2962,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960343688638687,
+      "loss": 1.31,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960060112494542,
+      "loss": 1.2575,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995977552808362,
+      "loss": 1.2543,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959489935434733,
+      "loss": 1.3437,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959203334576794,
+      "loss": 1.2947,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958915725538817,
+      "loss": 1.2679,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995862710834991,
+      "loss": 1.349,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958337483039296,
+      "loss": 1.3337,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958046849636292,
+      "loss": 1.3522,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957755208170318,
+      "loss": 1.3161,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957462558670903,
+      "loss": 1.3157,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957168901167666,
+      "loss": 1.2809,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956874235690334,
+      "loss": 1.2995,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995657856226874,
+      "loss": 1.3021,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995628188093281,
+      "loss": 1.3236,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955984191712584,
+      "loss": 1.279,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955685494638193,
+      "loss": 1.2316,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955385789739878,
+      "loss": 1.3316,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955085077047975,
+      "loss": 1.2821,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019954783356592929,
+      "loss": 1.2859,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019954480628405277,
+      "loss": 1.2928,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995417689251567,
+      "loss": 1.2794,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953872148954855,
+      "loss": 1.3457,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995356639775368,
+      "loss": 1.3328,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953259638943096,
+      "loss": 1.2835,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952951872554157,
+      "loss": 1.3368,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952643098618023,
+      "loss": 1.301,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952333317165946,
+      "loss": 1.3062,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952022528229283,
+      "loss": 1.3078,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951710731839502,
+      "loss": 1.2849,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951397928028163,
+      "loss": 1.2238,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951084116826934,
+      "loss": 1.2676,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950769298267583,
+      "loss": 1.2755,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950453472381973,
+      "loss": 1.3405,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950136639202082,
+      "loss": 1.3736,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949818798759982,
+      "loss": 1.3394,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949499951087845,
+      "loss": 1.2847,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949180096217947,
+      "loss": 1.3388,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948859234182676,
+      "loss": 1.2751,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948537365014506,
+      "loss": 1.34,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948214488746022,
+      "loss": 1.3292,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994789060540991,
+      "loss": 1.3368,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947565715038956,
+      "loss": 1.2979,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947239817666047,
+      "loss": 1.2761,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019946912913324178,
+      "loss": 1.3148,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019946585002046438,
+      "loss": 1.2694,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019946256083866026,
+      "loss": 1.3103,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019945926158816235,
+      "loss": 1.2328,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019945595226930464,
+      "loss": 1.3425,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019945263288242217,
+      "loss": 1.2759,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944930342785094,
+      "loss": 1.3026,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944596390592797,
+      "loss": 1.2774,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944261431699135,
+      "loss": 1.2709,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943925466138019,
+      "loss": 1.2859,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943588493943453,
+      "loss": 1.2858,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943250515149554,
+      "loss": 1.2678,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942911529790533,
+      "loss": 1.288,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942571537900707,
+      "loss": 1.2569,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942230539514495,
+      "loss": 1.2367,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941888534666413,
+      "loss": 1.2557,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941545523391086,
+      "loss": 1.297,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941201505723234,
+      "loss": 1.372,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019940856481697687,
+      "loss": 1.2763,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001994051045134937,
+      "loss": 1.3077,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001994016341471331,
+      "loss": 1.3199,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939815371824637,
+      "loss": 1.3023,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939466322718588,
+      "loss": 1.3003,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939116267430498,
+      "loss": 1.2722,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019938765205995798,
+      "loss": 1.3298,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993841313845003,
+      "loss": 1.2837,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019938060064828833,
+      "loss": 1.2862,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993770598516795,
+      "loss": 1.2721,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019937350899503224,
+      "loss": 1.3444,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019936994807870603,
+      "loss": 1.3297,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993663771030613,
+      "loss": 1.2758,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993627960684596,
+      "loss": 1.3245,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019935920497526338,
+      "loss": 1.2612,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993556038238362,
+      "loss": 1.2508,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019935199261454263,
+      "loss": 1.3108,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993483713477482,
+      "loss": 1.2759,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934474002381952,
+      "loss": 1.2954,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934109864312415,
+      "loss": 1.2834,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933744720603078,
+      "loss": 1.3442,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933378571290898,
+      "loss": 1.2512,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933011416412946,
+      "loss": 1.3251,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019932643256006388,
+      "loss": 1.2842,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019932274090108488,
+      "loss": 1.2782,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931903918756624,
+      "loss": 1.3031,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931532741988267,
+      "loss": 1.3311,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931160559840994,
+      "loss": 1.3046,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019930787372352473,
+      "loss": 1.2609,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993041317956049,
+      "loss": 1.2191,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993003798150292,
+      "loss": 1.2206,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001992966177821775,
+      "loss": 1.2764,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992928456974306,
+      "loss": 1.2906,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992890635611703,
+      "loss": 1.2453,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992852713737796,
+      "loss": 1.3011,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019928146913564227,
+      "loss": 1.2416,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019927765684714328,
+      "loss": 1.2548,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992738345086685,
+      "loss": 1.2528,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019927000212060492,
+      "loss": 1.2716,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019926615968334047,
+      "loss": 1.274,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019926230719726412,
+      "loss": 1.255,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019925844466276587,
+      "loss": 1.2496,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019925457208023673,
+      "loss": 1.3053,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992506894500687,
+      "loss": 1.3164,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019924679677265486,
+      "loss": 1.2593,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992428940483893,
+      "loss": 1.3557,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923898127766697,
+      "loss": 1.2256,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992350584608841,
+      "loss": 1.2643,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923112559843773,
+      "loss": 1.2636,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922718269072603,
+      "loss": 1.3445,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922322973814807,
+      "loss": 1.2602,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019921926674110408,
+      "loss": 1.2941,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019921529369999521,
+      "loss": 1.2942,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019921131061522367,
+      "loss": 1.3006,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019920731748719264,
+      "loss": 1.295,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992033143163064,
+      "loss": 1.2905,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019919930110297017,
+      "loss": 1.3427,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019919527784759017,
+      "loss": 1.2728,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019919124455057373,
+      "loss": 1.2741,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019918720121232916,
+      "loss": 1.3396,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019918314783326572,
+      "loss": 1.3171,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917908441379378,
+      "loss": 1.2261,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917501095432464,
+      "loss": 1.3132,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917092745527068,
+      "loss": 1.3109,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991668339170453,
+      "loss": 1.2604,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019916273034006286,
+      "loss": 1.3483,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019915861672473878,
+      "loss": 1.3245,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019915449307148948,
+      "loss": 1.3262,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019915035938073244,
+      "loss": 1.3259,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019914621565288605,
+      "loss": 1.3456,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019914206188836986,
+      "loss": 1.2276,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019913789808760429,
+      "loss": 1.2755,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019913372425101088,
+      "loss": 1.2682,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019912954037901216,
+      "loss": 1.2511,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019912534647203165,
+      "loss": 1.2496,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991211425304939,
+      "loss": 1.3068,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991169285548245,
+      "loss": 1.2954,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019911270454545004,
+      "loss": 1.3333,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019910847050279808,
+      "loss": 1.2251,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019910422642729726,
+      "loss": 1.2724,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001990999723193772,
+      "loss": 1.2668,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990957081794686,
+      "loss": 1.3367,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019909143400800305,
+      "loss": 1.2696,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019908714980541325,
+      "loss": 1.2307,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019908285557213292,
+      "loss": 1.3324,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019907855130859675,
+      "loss": 1.3288,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019907423701524048,
+      "loss": 1.2765,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019906991269250086,
+      "loss": 1.1315,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990655783408156,
+      "loss": 1.2954,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990612339606235,
+      "loss": 1.298,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019905687955236432,
+      "loss": 1.3027,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019905251511647892,
+      "loss": 1.3432,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019904814065340904,
+      "loss": 1.2642,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019904375616359757,
+      "loss": 1.3014,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019903936164748833,
+      "loss": 1.2544,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990349571055262,
+      "loss": 1.2831,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019903054253815703,
+      "loss": 1.306,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019902611794582776,
+      "loss": 1.3082,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019902168332898624,
+      "loss": 1.2773,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990172386880814,
+      "loss": 1.2833,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019901278402356318,
+      "loss": 1.3505,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019900831933588253,
+      "loss": 1.3253,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019900384462549146,
+      "loss": 1.2544,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019899935989284287,
+      "loss": 1.3221,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019899486513839083,
+      "loss": 1.2336,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019899036036259032,
+      "loss": 1.2869,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019898584556589732,
+      "loss": 1.2978,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019898132074876892,
+      "loss": 1.2609,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019897678591166318,
+      "loss": 1.323,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989722410550391,
+      "loss": 1.1814,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019896768617935683,
+      "loss": 1.2768,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989631212850774,
+      "loss": 1.2864,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000198958546372663,
+      "loss": 1.2694,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989539614425767,
+      "loss": 1.2719,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019894936649528263,
+      "loss": 1.3139,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019894476153124597,
+      "loss": 1.345,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019894014655093284,
+      "loss": 1.3013,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989355215548105,
+      "loss": 1.2357,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019893088654334706,
+      "loss": 1.3159,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019892624151701175,
+      "loss": 1.2079,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019892158647627484,
+      "loss": 1.2971,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019891692142160752,
+      "loss": 1.2965,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000198912246353482,
+      "loss": 1.2646,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019890756127237164,
+      "loss": 1.2496,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989028661787506,
+      "loss": 1.3169,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019889816107309427,
+      "loss": 1.3451,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019889344595587892,
+      "loss": 1.3021,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019888872082758185,
+      "loss": 1.3061,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019888398568868138,
+      "loss": 1.2029,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988792405396569,
+      "loss": 1.2688,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988744853809887,
+      "loss": 1.3058,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988697202131582,
+      "loss": 1.1953,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019886494503664777,
+      "loss": 1.2438,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988601598519408,
+      "loss": 1.25,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019885536465952171,
+      "loss": 1.3355,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019885055945987592,
+      "loss": 1.3007,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019884574425348985,
+      "loss": 1.2665,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019884091904085098,
+      "loss": 1.2656,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988360838224477,
+      "loss": 1.2942,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988312385987696,
+      "loss": 1.2793,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019882638337030705,
+      "loss": 1.2902,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019882151813755162,
+      "loss": 1.2604,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019881664290099583,
+      "loss": 1.2123,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019881175766113312,
+      "loss": 1.3131,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.9085344076156616,
+      "eval_runtime": 500.9525,
+      "eval_samples_per_second": 51.702,
+      "eval_steps_per_second": 25.851,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019880686241845812,
+      "loss": 1.2923,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019880195717346635,
+      "loss": 1.2369,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019879704192665436,
+      "loss": 1.3472,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019879211667851973,
+      "loss": 1.2518,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019878718142956106,
+      "loss": 1.2682,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019878223618027794,
+      "loss": 1.2916,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019877728093117098,
+      "loss": 1.1845,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987723156827418,
+      "loss": 1.1866,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019876734043549306,
+      "loss": 1.1516,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987623551899284,
+      "loss": 1.2406,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019875735994655244,
+      "loss": 1.3324,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019875235470587091,
+      "loss": 1.2714,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019874733946839048,
+      "loss": 1.3033,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019874231423461884,
+      "loss": 1.2705,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019873727900506473,
+      "loss": 1.3042,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001987322337802378,
+      "loss": 1.3242,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019872717856064888,
+      "loss": 1.35,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001987221133468096,
+      "loss": 1.2572,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019871703813923282,
+      "loss": 1.2216,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019871195293843225,
+      "loss": 1.2907,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001987068577449227,
+      "loss": 1.3026,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001987017525592199,
+      "loss": 1.3666,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019869663738184077,
+      "loss": 1.2819,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019869151221330302,
+      "loss": 1.2837,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986863770541255,
+      "loss": 1.2532,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019868123190482808,
+      "loss": 1.3197,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019867607676593156,
+      "loss": 1.2813,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019867091163795785,
+      "loss": 1.2461,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986657365214298,
+      "loss": 1.2643,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986605514168713,
+      "loss": 1.2318,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986553563248072,
+      "loss": 1.2511,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019865015124576348,
+      "loss": 1.2484,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019864493618026695,
+      "loss": 1.256,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019863971112884566,
+      "loss": 1.3308,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019863447609202845,
+      "loss": 1.2864,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019862923107034534,
+      "loss": 1.3536,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019862397606432722,
+      "loss": 1.302,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019861871107450612,
+      "loss": 1.2262,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019861343610141495,
+      "loss": 1.278,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001986081511455878,
+      "loss": 1.2957,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019860285620755958,
+      "loss": 1.2685,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019859755128786633,
+      "loss": 1.2876,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019859223638704512,
+      "loss": 1.263,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001985869115056339,
+      "loss": 1.2406,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019858157664417179,
+      "loss": 1.2357,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019857623180319876,
+      "loss": 1.2603,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019857087698325597,
+      "loss": 1.2867,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019856551218488543,
+      "loss": 1.271,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019856013740863023,
+      "loss": 1.2259,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019855475265503447,
+      "loss": 1.2588,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019854935792464325,
+      "loss": 1.2132,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001985439532180027,
+      "loss": 1.2943,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019853853853565997,
+      "loss": 1.2647,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019853311387816312,
+      "loss": 1.2775,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019852767924606135,
+      "loss": 1.2626,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019852223463990478,
+      "loss": 1.3448,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001985167800602446,
+      "loss": 1.277,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019851131550763294,
+      "loss": 1.2931,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019850584098262306,
+      "loss": 1.3256,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019850035648576907,
+      "loss": 1.2222,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001984948620176262,
+      "loss": 1.2762,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019848935757875072,
+      "loss": 1.2136,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019848384316969975,
+      "loss": 1.2212,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019847831879103156,
+      "loss": 1.3068,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019847278444330542,
+      "loss": 1.2447,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019846724012708154,
+      "loss": 1.2495,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019846168584292117,
+      "loss": 1.3138,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019845612159138663,
+      "loss": 1.2421,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019845054737304115,
+      "loss": 1.2955,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000198444963188449,
+      "loss": 1.2113,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984393690381755,
+      "loss": 1.3096,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019843376492278692,
+      "loss": 1.2271,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984281508428506,
+      "loss": 1.3097,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984225267989349,
+      "loss": 1.2155,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984168927916091,
+      "loss": 1.2741,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984112488214435,
+      "loss": 1.2755,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984055948890095,
+      "loss": 1.2586,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019839993099487942,
+      "loss": 1.2435,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019839425713962666,
+      "loss": 1.2541,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019838857332382556,
+      "loss": 1.3138,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983828795480515,
+      "loss": 1.2315,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983771758128809,
+      "loss": 1.2108,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983714621188911,
+      "loss": 1.3389,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019836573846666053,
+      "loss": 1.2536,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983600048567686,
+      "loss": 1.3008,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019835426128979574,
+      "loss": 1.2423,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019834850776632336,
+      "loss": 1.2945,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983427442869339,
+      "loss": 1.2649,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019833697085221082,
+      "loss": 1.2797,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019833118746273852,
+      "loss": 1.2391,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019832539411910254,
+      "loss": 1.2546,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983195908218893,
+      "loss": 1.2617,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019831377757168622,
+      "loss": 1.2839,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019830795436908186,
+      "loss": 1.2804,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983021212146657,
+      "loss": 1.3039,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982962781090282,
+      "loss": 1.3072,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982904250527609,
+      "loss": 1.2879,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019828456204645627,
+      "loss": 1.2633,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982786890907079,
+      "loss": 1.2862,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019827280618611018,
+      "loss": 1.3338,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019826691333325877,
+      "loss": 1.3186,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982610105327502,
+      "loss": 1.2664,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019825509778518193,
+      "loss": 1.2779,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019824917509115263,
+      "loss": 1.2781,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019824324245126172,
+      "loss": 1.2319,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982372998661099,
+      "loss": 1.2804,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019823134733629869,
+      "loss": 1.2495,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019822538486243064,
+      "loss": 1.2724,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019821941244510936,
+      "loss": 1.2846,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982134300849395,
+      "loss": 1.2631,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019820743778252656,
+      "loss": 1.228,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019820143553847723,
+      "loss": 1.3081,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019819542335339907,
+      "loss": 1.24,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019818940122790074,
+      "loss": 1.2593,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019818336916259182,
+      "loss": 1.2975,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019817732715808297,
+      "loss": 1.2752,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019817127521498583,
+      "loss": 1.2587,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019816521333391305,
+      "loss": 1.2446,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019815914151547826,
+      "loss": 1.2193,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019815305976029612,
+      "loss": 1.2483,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981469680689823,
+      "loss": 1.2975,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019814086644215345,
+      "loss": 1.3177,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019813475488042728,
+      "loss": 1.2576,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019812863338442244,
+      "loss": 1.2032,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019812250195475864,
+      "loss": 1.2619,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019811636059205654,
+      "loss": 1.3143,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019811020929693784,
+      "loss": 1.2795,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019810404807002528,
+      "loss": 1.2414,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019809787691194253,
+      "loss": 1.2569,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019809169582331432,
+      "loss": 1.3244,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019808550480476632,
+      "loss": 1.2842,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019807930385692534,
+      "loss": 1.3542,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019807309298041904,
+      "loss": 1.2466,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980668721758762,
+      "loss": 1.2621,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980606414439265,
+      "loss": 1.1873,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019805440078520073,
+      "loss": 1.2565,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019804815020033062,
+      "loss": 1.2641,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019804188968994892,
+      "loss": 1.3378,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001980356192546894,
+      "loss": 1.1973,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019802933889518683,
+      "loss": 1.2489,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019802304861207693,
+      "loss": 1.2222,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019801674840599654,
+      "loss": 1.291,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019801043827758337,
+      "loss": 1.2231,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019800411822747625,
+      "loss": 1.303,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019799778825631495,
+      "loss": 1.2579,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019799144836474026,
+      "loss": 1.2573,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019798509855339396,
+      "loss": 1.2714,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019797873882291885,
+      "loss": 1.2407,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019797236917395875,
+      "loss": 1.2945,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019796598960715843,
+      "loss": 1.3181,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019795960012316378,
+      "loss": 1.3005,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979532007226215,
+      "loss": 1.2549,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979467914061795,
+      "loss": 1.2691,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979403721744866,
+      "loss": 1.2302,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019793394302819253,
+      "loss": 1.2165,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019792750396794822,
+      "loss": 1.2674,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019792105499440545,
+      "loss": 1.2131,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019791459610821708,
+      "loss": 1.2303,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019790812731003692,
+      "loss": 1.328,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019790164860051984,
+      "loss": 1.2963,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019789515998032168,
+      "loss": 1.2656,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978886614500993,
+      "loss": 1.2891,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978821530105105,
+      "loss": 1.3094,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019787563466221425,
+      "loss": 1.3339,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978691064058703,
+      "loss": 1.23,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019786256824213952,
+      "loss": 1.2492,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019785602017168382,
+      "loss": 1.29,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019784946219516606,
+      "loss": 1.345,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978428943132501,
+      "loss": 1.1837,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978363165266008,
+      "loss": 1.2603,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978297288358841,
+      "loss": 1.3206,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019782313124176677,
+      "loss": 1.2823,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019781652374491676,
+      "loss": 1.3745,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019780990634600296,
+      "loss": 1.2939,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019780327904569522,
+      "loss": 1.2507,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019779664184466447,
+      "loss": 1.2417,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019778999474358256,
+      "loss": 1.2865,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001977833377431224,
+      "loss": 1.2819,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019777667084395787,
+      "loss": 1.3007,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001977699940467639,
+      "loss": 1.2968,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019776330735221637,
+      "loss": 1.3161,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019775661076099218,
+      "loss": 1.2854,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019774990427376925,
+      "loss": 1.2675,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019774318789122646,
+      "loss": 1.2107,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001977364616140437,
+      "loss": 1.3864,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001977297254429019,
+      "loss": 1.2809,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000197722979378483,
+      "loss": 1.274,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019771622342146987,
+      "loss": 1.3034,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019770945757254644,
+      "loss": 1.2733,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001977026818323976,
+      "loss": 1.2531,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976958962017093,
+      "loss": 1.2801,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019768910068116845,
+      "loss": 1.2697,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976822952714629,
+      "loss": 1.2841,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019767547997328165,
+      "loss": 1.2681,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976686547873146,
+      "loss": 1.2316,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976618197142527,
+      "loss": 1.2799,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976549747547878,
+      "loss": 1.2308,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019764811990961287,
+      "loss": 1.2562,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976412551794218,
+      "loss": 1.3316,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019763438056490955,
+      "loss": 1.1787,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019762749606677202,
+      "loss": 1.2364,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019762060168570616,
+      "loss": 1.2454,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019761369742240986,
+      "loss": 1.2051,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019760678327758206,
+      "loss": 1.2805,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019759985925192267,
+      "loss": 1.2891,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019759292534613266,
+      "loss": 1.2088,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019758598156091391,
+      "loss": 1.2534,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975790278969694,
+      "loss": 1.2058,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019757206435500301,
+      "loss": 1.2363,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975650909357197,
+      "loss": 1.1937,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019755810763982535,
+      "loss": 1.2731,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019755111446802693,
+      "loss": 1.2373,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975441114210323,
+      "loss": 1.2135,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019753709849955046,
+      "loss": 1.2655,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019753007570429132,
+      "loss": 1.2298,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019752304303596578,
+      "loss": 1.3126,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975160004952858,
+      "loss": 1.2315,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019750894808296426,
+      "loss": 1.2472,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975018857997151,
+      "loss": 1.3233,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019749481364625323,
+      "loss": 1.2237,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974877316232946,
+      "loss": 1.2496,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974806397315561,
+      "loss": 1.204,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974735379717557,
+      "loss": 1.2628,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019746642634461227,
+      "loss": 1.2837,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019745930485084575,
+      "loss": 1.2484,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000197452173491177,
+      "loss": 1.2631,
+      "step": 721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019744503226632801,
+      "loss": 1.273,
+      "step": 722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019743788117702166,
+      "loss": 1.2769,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974307202239819,
+      "loss": 1.2539,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019742354940793355,
+      "loss": 1.2798,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974163687296026,
+      "loss": 1.2283,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019740917818971597,
+      "loss": 1.3153,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019740197778900148,
+      "loss": 1.2884,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019739476752818807,
+      "loss": 1.2646,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019738754740800566,
+      "loss": 1.2597,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001973803174291852,
+      "loss": 1.3122,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019737307759245845,
+      "loss": 1.276,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019736582789855843,
+      "loss": 1.2645,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000197358568348219,
+      "loss": 1.3548,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019735129894217504,
+      "loss": 1.3041,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019734401968116242,
+      "loss": 1.2623,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019733673056591806,
+      "loss": 1.2637,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019732943159717984,
+      "loss": 1.2582,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001973221227756866,
+      "loss": 1.2393,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019731480410217832,
+      "loss": 1.2333,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019730747557739572,
+      "loss": 1.2978,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019730013720208086,
+      "loss": 1.2034,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019729278897697645,
+      "loss": 1.2125,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019728543090282642,
+      "loss": 1.1789,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019727806298037567,
+      "loss": 1.1934,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019727068521037,
+      "loss": 1.2522,
+      "step": 746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019726329759355636,
+      "loss": 1.2068,
+      "step": 747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972559001306825,
+      "loss": 1.2197,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019724849282249732,
+      "loss": 1.3504,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019724107566975065,
+      "loss": 1.2604,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972336486731934,
+      "loss": 1.29,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972262118335773,
+      "loss": 1.2471,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019721876515165531,
+      "loss": 1.2494,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019721130862818118,
+      "loss": 1.2522,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019720384226390978,
+      "loss": 1.2316,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019719636605959693,
+      "loss": 1.2634,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971888800159994,
+      "loss": 1.3167,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019718138413387515,
+      "loss": 1.2669,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019717387841398286,
+      "loss": 1.3001,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019716636285708238,
+      "loss": 1.3199,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019715883746393456,
+      "loss": 1.214,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019715130223530111,
+      "loss": 1.3098,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019714375717194493,
+      "loss": 1.285,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019713620227462974,
+      "loss": 1.3189,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019712863754412042,
+      "loss": 1.2941,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019712106298118266,
+      "loss": 1.2651,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019711347858658324,
+      "loss": 1.2868,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019710588436109002,
+      "loss": 1.29,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970982803054717,
+      "loss": 1.1869,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019709066642049808,
+      "loss": 1.2313,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970830427069399,
+      "loss": 1.259,
+      "step": 771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019707540916556892,
+      "loss": 1.335,
+      "step": 772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970677657971579,
+      "loss": 1.2768,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970601126024806,
+      "loss": 1.2309,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019705244958231175,
+      "loss": 1.2897,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019704477673742704,
+      "loss": 1.2152,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019703709406860328,
+      "loss": 1.1895,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019702940157661814,
+      "loss": 1.3445,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019702169926225034,
+      "loss": 1.2387,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019701398712627962,
+      "loss": 1.1964,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019700626516948666,
+      "loss": 1.1775,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019699853339265317,
+      "loss": 1.2422,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019699079179656182,
+      "loss": 1.2553,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019698304038199638,
+      "loss": 1.2082,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019697527914974147,
+      "loss": 1.3171,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019696750810058275,
+      "loss": 1.2424,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019695972723530694,
+      "loss": 1.2508,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969519365547017,
+      "loss": 1.3512,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019694413605955563,
+      "loss": 1.1838,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019693632575065845,
+      "loss": 1.259,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969285056288008,
+      "loss": 1.198,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019692067569477425,
+      "loss": 1.2831,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019691283594937153,
+      "loss": 1.2532,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019690498639338618,
+      "loss": 1.289,
+      "step": 794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019689712702761287,
+      "loss": 1.2667,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019688925785284718,
+      "loss": 1.1851,
+      "step": 796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019688137886988573,
+      "loss": 1.3116,
+      "step": 797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019687349007952617,
+      "loss": 1.2206,
+      "step": 798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019686559148256695,
+      "loss": 1.2303,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019685768307980782,
+      "loss": 1.2603,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001968497648720492,
+      "loss": 1.24,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001968418368600928,
+      "loss": 1.1905,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019683389904474104,
+      "loss": 1.2439,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019682595142679762,
+      "loss": 1.3056,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000196817994007067,
+      "loss": 1.2582,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001968100267863547,
+      "loss": 1.2858,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001968020497654673,
+      "loss": 1.1891,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019679406294521227,
+      "loss": 1.223,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019678606632639819,
+      "loss": 1.2934,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019677805990983452,
+      "loss": 1.2971,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019677004369633174,
+      "loss": 1.2616,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001967620176867014,
+      "loss": 1.3108,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019675398188175595,
+      "loss": 1.2065,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019674593628230883,
+      "loss": 1.2476,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019673788088917456,
+      "loss": 1.3489,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019672981570316856,
+      "loss": 1.2412,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001967217407251073,
+      "loss": 1.2021,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019671365595580817,
+      "loss": 1.2096,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019670556139608967,
+      "loss": 1.2686,
+      "step": 819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966974570467712,
+      "loss": 1.2585,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019668934290867308,
+      "loss": 1.2777,
+      "step": 821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019668121898261685,
+      "loss": 1.2555,
+      "step": 822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019667308526942484,
+      "loss": 1.2675,
+      "step": 823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019666494176992042,
+      "loss": 1.2517,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019665678848492798,
+      "loss": 1.2405,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019664862541527288,
+      "loss": 1.2378,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019664045256178147,
+      "loss": 1.2973,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019663226992528113,
+      "loss": 1.2639,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019662407750660016,
+      "loss": 1.2226,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001966158753065679,
+      "loss": 1.3091,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019660766332601468,
+      "loss": 1.1779,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019659944156577176,
+      "loss": 1.3078,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019659121002667149,
+      "loss": 1.2605,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019658296870954714,
+      "loss": 1.202,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019657471761523298,
+      "loss": 1.2575,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019656645674456428,
+      "loss": 1.2576,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965581860983773,
+      "loss": 1.2257,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019654990567750927,
+      "loss": 1.3553,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019654161548279846,
+      "loss": 1.2705,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019653331551508404,
+      "loss": 1.3242,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965250057752063,
+      "loss": 1.2141,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019651668626400636,
+      "loss": 1.2668,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019650835698232645,
+      "loss": 1.3077,
+      "step": 843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965000179310098,
+      "loss": 1.3308,
+      "step": 844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019649166911090048,
+      "loss": 1.278,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019648331052284372,
+      "loss": 1.232,
+      "step": 846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019647494216768564,
+      "loss": 1.2486,
+      "step": 847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964665640462734,
+      "loss": 1.3109,
+      "step": 848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964581761594551,
+      "loss": 1.2522,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019644977850807988,
+      "loss": 1.2829,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019644137109299785,
+      "loss": 1.3071,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019643295391506005,
+      "loss": 1.2744,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001964245269751186,
+      "loss": 1.2742,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019641609027402655,
+      "loss": 1.279,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019640764381263797,
+      "loss": 1.2099,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963991875918079,
+      "loss": 1.2505,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019639072161239235,
+      "loss": 1.2486,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963822458752484,
+      "loss": 1.2919,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019637376038123398,
+      "loss": 1.3166,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019636526513120813,
+      "loss": 1.2561,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019635676012603082,
+      "loss": 1.2388,
+      "step": 861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019634824536656306,
+      "loss": 1.2534,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963397208536667,
+      "loss": 1.2602,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963311865882048,
+      "loss": 1.3376,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019632264257104124,
+      "loss": 1.2627,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019631408880304093,
+      "loss": 1.2848,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019630552528506983,
+      "loss": 1.2866,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019629695201799478,
+      "loss": 1.2385,
+      "step": 868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962883690026837,
+      "loss": 1.2435,
+      "step": 869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019627977624000537,
+      "loss": 1.3032,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019627117373082975,
+      "loss": 1.2801,
+      "step": 871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019626256147602762,
+      "loss": 1.2801,
+      "step": 872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019625393947647087,
+      "loss": 1.3409,
+      "step": 873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019624530773303223,
+      "loss": 1.274,
+      "step": 874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019623666624658554,
+      "loss": 1.2841,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962280150180056,
+      "loss": 1.2098,
+      "step": 876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019621935404816817,
+      "loss": 1.2621,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019621068333795,
+      "loss": 1.2658,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019620200288822883,
+      "loss": 1.2466,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019619331269988343,
+      "loss": 1.247,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019618461277379345,
+      "loss": 1.2575,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019617590311083968,
+      "loss": 1.2935,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001961671837119037,
+      "loss": 1.2487,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019615845457786826,
+      "loss": 1.2655,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019614971570961701,
+      "loss": 1.2749,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001961409671080346,
+      "loss": 1.2365,
+      "step": 886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019613220877400664,
+      "loss": 1.311,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019612344070841971,
+      "loss": 1.2095,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001961146629121615,
+      "loss": 1.233,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019610587538612052,
+      "loss": 1.2786,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019609707813118637,
+      "loss": 1.282,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001960882711482496,
+      "loss": 1.2826,
+      "step": 892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019607945443820171,
+      "loss": 1.2703,
+      "step": 893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019607062800193532,
+      "loss": 1.2766,
+      "step": 894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019606179184034387,
+      "loss": 1.2269,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019605294595432184,
+      "loss": 1.2812,
+      "step": 896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019604409034476477,
+      "loss": 1.2736,
+      "step": 897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019603522501256905,
+      "loss": 1.2466,
+      "step": 898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001960263499586322,
+      "loss": 1.254,
+      "step": 899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019601746518385258,
+      "loss": 1.2524,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019600857068912964,
+      "loss": 1.2129,
+      "step": 901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019599966647536378,
+      "loss": 1.2667,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019599075254345637,
+      "loss": 1.2897,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959818288943098,
+      "loss": 1.2826,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019597289552882738,
+      "loss": 1.2543,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019596395244791351,
+      "loss": 1.2443,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959549996524734,
+      "loss": 1.2656,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019594603714341346,
+      "loss": 1.2368,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019593706492164093,
+      "loss": 1.2607,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019592808298806404,
+      "loss": 1.2467,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019591909134359213,
+      "loss": 1.2207,
+      "step": 911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019591008998913532,
+      "loss": 1.3032,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959010789256049,
+      "loss": 1.2871,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019589205815391308,
+      "loss": 1.2004,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000195883027674973,
+      "loss": 1.2485,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019587398748969887,
+      "loss": 1.2599,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019586493759900578,
+      "loss": 1.2139,
+      "step": 917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019585587800380986,
+      "loss": 1.2254,
+      "step": 918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019584680870502827,
+      "loss": 1.2011,
+      "step": 919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019583772970357908,
+      "loss": 1.2146,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019582864100038138,
+      "loss": 1.2395,
+      "step": 921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958195425963552,
+      "loss": 1.2976,
+      "step": 922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958104344924216,
+      "loss": 1.2051,
+      "step": 923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019580131668950262,
+      "loss": 1.2763,
+      "step": 924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001957921891885212,
+      "loss": 1.2369,
+      "step": 925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019578305199040142,
+      "loss": 1.2824,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019577390509606816,
+      "loss": 1.2207,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001957647485064474,
+      "loss": 1.2716,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001957555822224661,
+      "loss": 1.239,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019574640624505217,
+      "loss": 1.2746,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019573722057513444,
+      "loss": 1.2351,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019572802521364285,
+      "loss": 1.2394,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001957188201615082,
+      "loss": 1.1881,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019570960541966242,
+      "loss": 1.2322,
+      "step": 934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001957003809890382,
+      "loss": 1.2975,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019569114687056945,
+      "loss": 1.2514,
+      "step": 936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956819030651909,
+      "loss": 1.2772,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019567264957383827,
+      "loss": 1.2402,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019566338639744838,
+      "loss": 1.2286,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956541135369589,
+      "loss": 1.2851,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019564483099330856,
+      "loss": 1.1644,
+      "step": 941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019563553876743704,
+      "loss": 1.2463,
+      "step": 942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019562623686028493,
+      "loss": 1.2692,
+      "step": 943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000195616925272794,
+      "loss": 1.2475,
+      "step": 944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019560760400590677,
+      "loss": 1.2991,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019559827306056687,
+      "loss": 1.3185,
+      "step": 946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019558893243771885,
+      "loss": 1.2726,
+      "step": 947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019557958213830835,
+      "loss": 1.2568,
+      "step": 948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019557022216328183,
+      "loss": 1.2041,
+      "step": 949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019556085251358686,
+      "loss": 1.2014,
+      "step": 950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955514731901719,
+      "loss": 1.2661,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019554208419398646,
+      "loss": 1.287,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000195532685525981,
+      "loss": 1.2075,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019552327718710692,
+      "loss": 1.1615,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019551385917831668,
+      "loss": 1.2321,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019550443150056362,
+      "loss": 1.201,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954949941548022,
+      "loss": 1.32,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019548554714198768,
+      "loss": 1.2544,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019547609046307644,
+      "loss": 1.2095,
+      "step": 959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019546662411902576,
+      "loss": 1.2862,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019545714811079396,
+      "loss": 1.2082,
+      "step": 961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954476624393403,
+      "loss": 1.1913,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000195438167105625,
+      "loss": 1.1941,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019542866211060927,
+      "loss": 1.2464,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019541914745525537,
+      "loss": 1.2322,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019540962314052644,
+      "loss": 1.246,
+      "step": 966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019540008916738667,
+      "loss": 1.2886,
+      "step": 967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019539054553680113,
+      "loss": 1.2868,
+      "step": 968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019538099224973595,
+      "loss": 1.2605,
+      "step": 969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019537142930715825,
+      "loss": 1.2409,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019536185671003607,
+      "loss": 1.2351,
+      "step": 971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019535227445933849,
+      "loss": 1.2545,
+      "step": 972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019534268255603547,
+      "loss": 1.2268,
+      "step": 973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001953330810010981,
+      "loss": 1.2486,
+      "step": 974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019532346979549825,
+      "loss": 1.2202,
+      "step": 975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019531384894020892,
+      "loss": 1.2729,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019530421843620403,
+      "loss": 1.1667,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019529457828445853,
+      "loss": 1.2926,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019528492848594825,
+      "loss": 1.2704,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019527526904165003,
+      "loss": 1.2108,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019526559995254177,
+      "loss": 1.2976,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019525592121960226,
+      "loss": 1.2465,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019524623284381126,
+      "loss": 1.2297,
+      "step": 983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019523653482614957,
+      "loss": 1.2168,
+      "step": 984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019522682716759888,
+      "loss": 1.2267,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019521710986914194,
+      "loss": 1.2766,
+      "step": 986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019520738293176247,
+      "loss": 1.2758,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951976463564451,
+      "loss": 1.2292,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019518790014417547,
+      "loss": 1.2653,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019517814429594022,
+      "loss": 1.2626,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.8773365616798401,
+      "eval_runtime": 501.214,
+      "eval_samples_per_second": 51.675,
+      "eval_steps_per_second": 25.837,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019516837881272694,
+      "loss": 1.2903,
+      "step": 991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951586036955242,
+      "loss": 1.2252,
+      "step": 992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951488189453215,
+      "loss": 1.2549,
+      "step": 993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019513902456310943,
+      "loss": 1.2065,
+      "step": 994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951292205498794,
+      "loss": 1.2446,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000195119406906624,
+      "loss": 1.276,
+      "step": 996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019510958363433658,
+      "loss": 1.2299,
+      "step": 997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001950997507340116,
+      "loss": 1.2432,
+      "step": 998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001950899082066444,
+      "loss": 1.2464,
+      "step": 999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019508005605323143,
+      "loss": 1.272,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019507019427477,
+      "loss": 1.2423,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019506032287225838,
+      "loss": 1.3064,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019505044184669595,
+      "loss": 1.2174,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019504055119908288,
+      "loss": 1.2381,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001950306509304205,
+      "loss": 1.2454,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019502074104171097,
+      "loss": 1.2739,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001950108215339575,
+      "loss": 1.2433,
+      "step": 1007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019500089240816424,
+      "loss": 1.2475,
+      "step": 1008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019499095366533632,
+      "loss": 1.2339,
+      "step": 1009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019498100530647986,
+      "loss": 1.2403,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019497104733260195,
+      "loss": 1.2849,
+      "step": 1011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001949610797447106,
+      "loss": 1.2307,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019495110254381492,
+      "loss": 1.298,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019494111573092483,
+      "loss": 1.2564,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019493111930705137,
+      "loss": 1.3115,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001949211132732064,
+      "loss": 1.1724,
+      "step": 1016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019491109763040298,
+      "loss": 1.2254,
+      "step": 1017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019490107237965488,
+      "loss": 1.2453,
+      "step": 1018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019489103752197702,
+      "loss": 1.223,
+      "step": 1019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948809930583852,
+      "loss": 1.2269,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948709389898963,
+      "loss": 1.2461,
+      "step": 1021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000194860875317528,
+      "loss": 1.1869,
+      "step": 1022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019485080204229919,
+      "loss": 1.2199,
+      "step": 1023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019484071916522948,
+      "loss": 1.1977,
+      "step": 1024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948306266873396,
+      "loss": 1.3396,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019482052460965127,
+      "loss": 1.289,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001948104129331871,
+      "loss": 1.2734,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001948002916589707,
+      "loss": 1.2267,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019479016078802666,
+      "loss": 1.2703,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019478002032138054,
+      "loss": 1.2294,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019476987026005886,
+      "loss": 1.2836,
+      "step": 1031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019475971060508912,
+      "loss": 1.2841,
+      "step": 1032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019474954135749982,
+      "loss": 1.2552,
+      "step": 1033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019473936251832038,
+      "loss": 1.229,
+      "step": 1034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019472917408858124,
+      "loss": 1.2154,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019471897606931373,
+      "loss": 1.2647,
+      "step": 1036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019470876846155023,
+      "loss": 1.2422,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946985512663241,
+      "loss": 1.2126,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019468832448466962,
+      "loss": 1.2506,
+      "step": 1039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019467808811762205,
+      "loss": 1.211,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946678421662176,
+      "loss": 1.2671,
+      "step": 1041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946575866314935,
+      "loss": 1.2488,
+      "step": 1042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000194647321514488,
+      "loss": 1.1989,
+      "step": 1043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019463704681624012,
+      "loss": 1.2653,
+      "step": 1044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019462676253779003,
+      "loss": 1.2629,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019461646868017884,
+      "loss": 1.2956,
+      "step": 1046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019460616524444863,
+      "loss": 1.2384,
+      "step": 1047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019459585223164237,
+      "loss": 1.294,
+      "step": 1048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019458552964280406,
+      "loss": 1.2262,
+      "step": 1049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019457519747897867,
+      "loss": 1.279,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019456485574121217,
+      "loss": 1.2562,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019455450443055144,
+      "loss": 1.3087,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019454414354804438,
+      "loss": 1.2508,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001945337730947398,
+      "loss": 1.2273,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001945233930716875,
+      "loss": 1.2809,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001945130034799383,
+      "loss": 1.2512,
+      "step": 1056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001945026043205439,
+      "loss": 1.2429,
+      "step": 1057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019449219559455706,
+      "loss": 1.2339,
+      "step": 1058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019448177730303147,
+      "loss": 1.2606,
+      "step": 1059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019447134944702174,
+      "loss": 1.2286,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019446091202758353,
+      "loss": 1.2333,
+      "step": 1061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001944504650457734,
+      "loss": 1.2281,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019444000850264893,
+      "loss": 1.2484,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019442954239926868,
+      "loss": 1.2481,
+      "step": 1064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019441906673669208,
+      "loss": 1.2278,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019440858151597963,
+      "loss": 1.2626,
+      "step": 1066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019439808673819272,
+      "loss": 1.2856,
+      "step": 1067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019438758240439378,
+      "loss": 1.2613,
+      "step": 1068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001943770685156462,
+      "loss": 1.1983,
+      "step": 1069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019436654507301427,
+      "loss": 1.2306,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019435601207756328,
+      "loss": 1.2418,
+      "step": 1071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019434546953035954,
+      "loss": 1.245,
+      "step": 1072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019433491743247024,
+      "loss": 1.2332,
+      "step": 1073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019432435578496356,
+      "loss": 1.2749,
+      "step": 1074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019431378458890876,
+      "loss": 1.2468,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001943032038453759,
+      "loss": 1.2804,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019429261355543607,
+      "loss": 1.2706,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001942820137201614,
+      "loss": 1.3057,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019427140434062485,
+      "loss": 1.307,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019426078541790048,
+      "loss": 1.2314,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019425015695306315,
+      "loss": 1.246,
+      "step": 1081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019423951894718895,
+      "loss": 1.252,
+      "step": 1082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019422887140135465,
+      "loss": 1.2475,
+      "step": 1083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019421821431663813,
+      "loss": 1.2566,
+      "step": 1084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019420754769411828,
+      "loss": 1.2505,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019419687153487484,
+      "loss": 1.2879,
+      "step": 1086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019418618583998858,
+      "loss": 1.2624,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019417549061054122,
+      "loss": 1.2727,
+      "step": 1088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019416478584761545,
+      "loss": 1.2337,
+      "step": 1089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019415407155229492,
+      "loss": 1.2841,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019414334772566424,
+      "loss": 1.2816,
+      "step": 1091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019413261436880905,
+      "loss": 1.2789,
+      "step": 1092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019412187148281582,
+      "loss": 1.2592,
+      "step": 1093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941111190687721,
+      "loss": 1.2755,
+      "step": 1094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019410035712776638,
+      "loss": 1.2031,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001940895856608881,
+      "loss": 1.2728,
+      "step": 1096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019407880466922763,
+      "loss": 1.286,
+      "step": 1097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019406801415387638,
+      "loss": 1.2774,
+      "step": 1098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019405721411592665,
+      "loss": 1.2242,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019404640455647178,
+      "loss": 1.2327,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000194035585476606,
+      "loss": 1.239,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019402475687742456,
+      "loss": 1.268,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019401391876002361,
+      "loss": 1.2722,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019400307112550037,
+      "loss": 1.2956,
+      "step": 1104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019399221397495292,
+      "loss": 1.2322,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939813473094803,
+      "loss": 1.2704,
+      "step": 1106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939704711301826,
+      "loss": 1.271,
+      "step": 1107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019395958543816082,
+      "loss": 1.2497,
+      "step": 1108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019394869023451694,
+      "loss": 1.2779,
+      "step": 1109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019393778552035388,
+      "loss": 1.2713,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939268712967755,
+      "loss": 1.2954,
+      "step": 1111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939159475648867,
+      "loss": 1.1946,
+      "step": 1112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939050143257933,
+      "loss": 1.3073,
+      "step": 1113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019389407158060208,
+      "loss": 1.1861,
+      "step": 1114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938831193304208,
+      "loss": 1.2397,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938721575763581,
+      "loss": 1.2862,
+      "step": 1116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019386118631952373,
+      "loss": 1.2505,
+      "step": 1117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019385020556102826,
+      "loss": 1.232,
+      "step": 1118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938392153019833,
+      "loss": 1.2306,
+      "step": 1119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019382821554350148,
+      "loss": 1.2621,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938172062866962,
+      "loss": 1.2491,
+      "step": 1121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019380618753268202,
+      "loss": 1.2802,
+      "step": 1122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019379515928257432,
+      "loss": 1.2295,
+      "step": 1123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019378412153748956,
+      "loss": 1.2178,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019377307429854502,
+      "loss": 1.2567,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019376201756685912,
+      "loss": 1.2988,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019375095134355108,
+      "loss": 1.2241,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019373987562974117,
+      "loss": 1.195,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019372879042655057,
+      "loss": 1.2715,
+      "step": 1129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001937176957351015,
+      "loss": 1.3213,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019370659155651702,
+      "loss": 1.2405,
+      "step": 1131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019369547789192126,
+      "loss": 1.2779,
+      "step": 1132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019368435474243926,
+      "loss": 1.2355,
+      "step": 1133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019367322210919703,
+      "loss": 1.2562,
+      "step": 1134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001936620799933215,
+      "loss": 1.2286,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019365092839594063,
+      "loss": 1.2015,
+      "step": 1136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019363976731818334,
+      "loss": 1.2196,
+      "step": 1137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019362859676117942,
+      "loss": 1.2777,
+      "step": 1138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019361741672605968,
+      "loss": 1.2644,
+      "step": 1139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019360622721395596,
+      "loss": 1.2464,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019359502822600087,
+      "loss": 1.2254,
+      "step": 1141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935838197633282,
+      "loss": 1.2701,
+      "step": 1142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019357260182707253,
+      "loss": 1.2725,
+      "step": 1143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935613744183695,
+      "loss": 1.2252,
+      "step": 1144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935501375383556,
+      "loss": 1.3444,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019353889118816847,
+      "loss": 1.2814,
+      "step": 1146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019352763536894648,
+      "loss": 1.2454,
+      "step": 1147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019351637008182914,
+      "loss": 1.2565,
+      "step": 1148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019350509532795678,
+      "loss": 1.2696,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001934938111084708,
+      "loss": 1.3077,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019348251742451353,
+      "loss": 1.1634,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001934712142772282,
+      "loss": 1.3003,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019345990166775906,
+      "loss": 1.2555,
+      "step": 1153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019344857959725128,
+      "loss": 1.2618,
+      "step": 1154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019343724806685102,
+      "loss": 1.2085,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019342590707770537,
+      "loss": 1.2923,
+      "step": 1156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019341455663096242,
+      "loss": 1.2357,
+      "step": 1157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019340319672777115,
+      "loss": 1.2567,
+      "step": 1158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019339182736928154,
+      "loss": 1.2831,
+      "step": 1159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019338044855664454,
+      "loss": 1.2485,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019336906029101202,
+      "loss": 1.2513,
+      "step": 1161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019335766257353685,
+      "loss": 1.1641,
+      "step": 1162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001933462554053728,
+      "loss": 1.2255,
+      "step": 1163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019333483878767466,
+      "loss": 1.2731,
+      "step": 1164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001933234127215981,
+      "loss": 1.1675,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019331197720829984,
+      "loss": 1.2912,
+      "step": 1166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001933005322489375,
+      "loss": 1.2713,
+      "step": 1167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019328907784466962,
+      "loss": 1.2367,
+      "step": 1168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932776139966558,
+      "loss": 1.2864,
+      "step": 1169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019326614070605652,
+      "loss": 1.2529,
+      "step": 1170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932546579740332,
+      "loss": 1.2959,
+      "step": 1171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019324316580174831,
+      "loss": 1.3618,
+      "step": 1172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019323166419036518,
+      "loss": 1.2637,
+      "step": 1173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932201531410481,
+      "loss": 1.2153,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019320863265496238,
+      "loss": 1.1965,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019319710273327424,
+      "loss": 1.263,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019318556337715085,
+      "loss": 1.296,
+      "step": 1177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001931740145877604,
+      "loss": 1.2161,
+      "step": 1178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019316245636627195,
+      "loss": 1.2728,
+      "step": 1179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019315088871385552,
+      "loss": 1.2468,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001931393116316822,
+      "loss": 1.1732,
+      "step": 1181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019312772512092386,
+      "loss": 1.2837,
+      "step": 1182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019311612918275346,
+      "loss": 1.2397,
+      "step": 1183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001931045238183449,
+      "loss": 1.2437,
+      "step": 1184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001930929090288729,
+      "loss": 1.2312,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001930812848155133,
+      "loss": 1.2854,
+      "step": 1186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019306965117944286,
+      "loss": 1.2767,
+      "step": 1187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019305800812183924,
+      "loss": 1.1937,
+      "step": 1188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019304635564388107,
+      "loss": 1.2051,
+      "step": 1189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019303469374674795,
+      "loss": 1.2942,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019302302243162038,
+      "loss": 1.2188,
+      "step": 1191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019301134169967993,
+      "loss": 1.253,
+      "step": 1192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019299965155210905,
+      "loss": 1.1721,
+      "step": 1193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019298795199009105,
+      "loss": 1.2772,
+      "step": 1194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001929762430148104,
+      "loss": 1.2482,
+      "step": 1195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019296452462745238,
+      "loss": 1.2536,
+      "step": 1196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001929527968292032,
+      "loss": 1.2427,
+      "step": 1197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019294105962125015,
+      "loss": 1.2393,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019292931300478135,
+      "loss": 1.2783,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019291755698098593,
+      "loss": 1.2457,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000192905791551054,
+      "loss": 1.2367,
+      "step": 1201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019289401671617653,
+      "loss": 1.2257,
+      "step": 1202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019288223247754553,
+      "loss": 1.222,
+      "step": 1203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019287043883635394,
+      "loss": 1.2191,
+      "step": 1204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001928586357937956,
+      "loss": 1.1836,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001928468233510654,
+      "loss": 1.2058,
+      "step": 1206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001928350015093591,
+      "loss": 1.2944,
+      "step": 1207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019282317026987343,
+      "loss": 1.1986,
+      "step": 1208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019281132963380607,
+      "loss": 1.235,
+      "step": 1209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019279947960235567,
+      "loss": 1.2229,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019278762017672184,
+      "loss": 1.2395,
+      "step": 1211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019277575135810508,
+      "loss": 1.2769,
+      "step": 1212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019276387314770693,
+      "loss": 1.2212,
+      "step": 1213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019275198554672977,
+      "loss": 1.3124,
+      "step": 1214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019274008855637708,
+      "loss": 1.2188,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019272818217785313,
+      "loss": 1.2737,
+      "step": 1216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019271626641236326,
+      "loss": 1.1458,
+      "step": 1217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019270434126111366,
+      "loss": 1.2637,
+      "step": 1218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019269240672531156,
+      "loss": 1.2297,
+      "step": 1219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019268046280616514,
+      "loss": 1.2869,
+      "step": 1220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019266850950488347,
+      "loss": 1.2469,
+      "step": 1221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019265654682267656,
+      "loss": 1.258,
+      "step": 1222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019264457476075544,
+      "loss": 1.2348,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019263259332033203,
+      "loss": 1.2339,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019262060250261923,
+      "loss": 1.3235,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019260860230883089,
+      "loss": 1.3068,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019259659274018182,
+      "loss": 1.2458,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019258457379788771,
+      "loss": 1.2014,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019257254548316532,
+      "loss": 1.3056,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019256050779723223,
+      "loss": 1.2782,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019254846074130705,
+      "loss": 1.2829,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925364043166093,
+      "loss": 1.3102,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019252433852435947,
+      "loss": 1.2447,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000192512263365779,
+      "loss": 1.2171,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925001788420903,
+      "loss": 1.2048,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924880849545166,
+      "loss": 1.2795,
+      "step": 1236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019247598170428227,
+      "loss": 1.2925,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019246386909261252,
+      "loss": 1.2455,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924517471207335,
+      "loss": 1.3049,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019243961578987233,
+      "loss": 1.246,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924274751012571,
+      "loss": 1.2615,
+      "step": 1241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924153250561168,
+      "loss": 1.3048,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019240316565568137,
+      "loss": 1.2497,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019239099690118177,
+      "loss": 1.2223,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019237881879384983,
+      "loss": 1.2639,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019236663133491834,
+      "loss": 1.2938,
+      "step": 1246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001923544345256211,
+      "loss": 1.2695,
+      "step": 1247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019234222836719275,
+      "loss": 1.2371,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019233001286086894,
+      "loss": 1.2365,
+      "step": 1249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019231778800788627,
+      "loss": 1.2383,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019230555380948225,
+      "loss": 1.1979,
+      "step": 1251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019229331026689538,
+      "loss": 1.2407,
+      "step": 1252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019228105738136513,
+      "loss": 1.1868,
+      "step": 1253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922687951541318,
+      "loss": 1.1796,
+      "step": 1254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019225652358643674,
+      "loss": 1.2161,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922442426795222,
+      "loss": 1.2459,
+      "step": 1256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922319524346314,
+      "loss": 1.2506,
+      "step": 1257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019221965285300848,
+      "loss": 1.2421,
+      "step": 1258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922073439358986,
+      "loss": 1.2026,
+      "step": 1259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019219502568454768,
+      "loss": 1.2468,
+      "step": 1260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001921826981002028,
+      "loss": 1.2294,
+      "step": 1261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019217036118411191,
+      "loss": 1.1733,
+      "step": 1262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019215801493752385,
+      "loss": 1.2323,
+      "step": 1263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019214565936168842,
+      "loss": 1.2595,
+      "step": 1264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019213329445785644,
+      "loss": 1.2033,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019212092022727957,
+      "loss": 1.2295,
+      "step": 1266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001921085366712105,
+      "loss": 1.2363,
+      "step": 1267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019209614379090285,
+      "loss": 1.2115,
+      "step": 1268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920837415876111,
+      "loss": 1.2607,
+      "step": 1269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920713300625908,
+      "loss": 1.2335,
+      "step": 1270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920589092170983,
+      "loss": 1.3169,
+      "step": 1271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019204647905239107,
+      "loss": 1.2677,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019203403956972738,
+      "loss": 1.2682,
+      "step": 1273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001920215907703665,
+      "loss": 1.2938,
+      "step": 1274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019200913265556862,
+      "loss": 1.2025,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001919966652265949,
+      "loss": 1.3,
+      "step": 1276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019198418848470744,
+      "loss": 1.2462,
+      "step": 1277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019197170243116923,
+      "loss": 1.2247,
+      "step": 1278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019195920706724427,
+      "loss": 1.1734,
+      "step": 1279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019194670239419754,
+      "loss": 1.2325,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001919341884132948,
+      "loss": 1.2369,
+      "step": 1281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019192166512580294,
+      "loss": 1.3141,
+      "step": 1282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019190913253298964,
+      "loss": 1.2508,
+      "step": 1283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001918965906361236,
+      "loss": 1.2458,
+      "step": 1284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019188403943647445,
+      "loss": 1.2303,
+      "step": 1285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019187147893531277,
+      "loss": 1.2636,
+      "step": 1286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019185890913391013,
+      "loss": 1.2552,
+      "step": 1287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019184633003353887,
+      "loss": 1.2443,
+      "step": 1288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019183374163547244,
+      "loss": 1.2843,
+      "step": 1289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001918211439409852,
+      "loss": 1.2453,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019180853695135238,
+      "loss": 1.2584,
+      "step": 1291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019179592066785023,
+      "loss": 1.2308,
+      "step": 1292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019178329509175588,
+      "loss": 1.2285,
+      "step": 1293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019177066022434746,
+      "loss": 1.2685,
+      "step": 1294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000191758016066904,
+      "loss": 1.289,
+      "step": 1295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001917453626207055,
+      "loss": 1.3093,
+      "step": 1296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019173269988703284,
+      "loss": 1.1709,
+      "step": 1297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001917200278671679,
+      "loss": 1.2228,
+      "step": 1298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019170734656239347,
+      "loss": 1.278,
+      "step": 1299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019169465597399334,
+      "loss": 1.2332,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019168195610325212,
+      "loss": 1.1356,
+      "step": 1301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019166924695145548,
+      "loss": 1.2747,
+      "step": 1302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019165652851988997,
+      "loss": 1.2379,
+      "step": 1303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001916438008098431,
+      "loss": 1.2327,
+      "step": 1304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019163106382260328,
+      "loss": 1.2323,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019161831755945988,
+      "loss": 1.1926,
+      "step": 1306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019160556202170326,
+      "loss": 1.2607,
+      "step": 1307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019159279721062464,
+      "loss": 1.2246,
+      "step": 1308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019158002312751624,
+      "loss": 1.2142,
+      "step": 1309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001915672397736712,
+      "loss": 1.2078,
+      "step": 1310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019155444715038354,
+      "loss": 1.2431,
+      "step": 1311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019154164525894828,
+      "loss": 1.2871,
+      "step": 1312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019152883410066145,
+      "loss": 1.2114,
+      "step": 1313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019151601367681985,
+      "loss": 1.2744,
+      "step": 1314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019150318398872133,
+      "loss": 1.279,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019149034503766464,
+      "loss": 1.2599,
+      "step": 1316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019147749682494954,
+      "loss": 1.2431,
+      "step": 1317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019146463935187658,
+      "loss": 1.1927,
+      "step": 1318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019145177261974738,
+      "loss": 1.1897,
+      "step": 1319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019143889662986445,
+      "loss": 1.2564,
+      "step": 1320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019142601138353124,
+      "loss": 1.1986,
+      "step": 1321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019141311688205214,
+      "loss": 1.203,
+      "step": 1322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019140021312673245,
+      "loss": 1.2394,
+      "step": 1323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019138730011887848,
+      "loss": 1.2919,
+      "step": 1324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019137437785979737,
+      "loss": 1.2282,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019136144635079728,
+      "loss": 1.1822,
+      "step": 1326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019134850559318726,
+      "loss": 1.2634,
+      "step": 1327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019133555558827737,
+      "loss": 1.2312,
+      "step": 1328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019132259633737847,
+      "loss": 1.3632,
+      "step": 1329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001913096278418025,
+      "loss": 1.2651,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019129665010286226,
+      "loss": 1.2664,
+      "step": 1331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019128366312187148,
+      "loss": 1.291,
+      "step": 1332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019127066690014487,
+      "loss": 1.2786,
+      "step": 1333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019125766143899804,
+      "loss": 1.278,
+      "step": 1334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019124464673974756,
+      "loss": 1.2576,
+      "step": 1335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019123162280371085,
+      "loss": 1.2714,
+      "step": 1336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019121858963220645,
+      "loss": 1.2404,
+      "step": 1337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019120554722655363,
+      "loss": 1.2803,
+      "step": 1338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019119249558807274,
+      "loss": 1.2759,
+      "step": 1339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000191179434718085,
+      "loss": 1.2549,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019116636461791254,
+      "loss": 1.2621,
+      "step": 1341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001911532852888785,
+      "loss": 1.2557,
+      "step": 1342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019114019673230684,
+      "loss": 1.1802,
+      "step": 1343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019112709894952262,
+      "loss": 1.253,
+      "step": 1344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019111399194185174,
+      "loss": 1.2679,
+      "step": 1345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019110087571062097,
+      "loss": 1.3045,
+      "step": 1346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001910877502571581,
+      "loss": 1.1894,
+      "step": 1347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019107461558279186,
+      "loss": 1.2303,
+      "step": 1348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001910614716888519,
+      "loss": 1.2443,
+      "step": 1349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001910483185766687,
+      "loss": 1.2468,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019103515624757384,
+      "loss": 1.1883,
+      "step": 1351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019102198470289977,
+      "loss": 1.2405,
+      "step": 1352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019100880394397977,
+      "loss": 1.2552,
+      "step": 1353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019099561397214825,
+      "loss": 1.2432,
+      "step": 1354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019098241478874038,
+      "loss": 1.2405,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001909692063950923,
+      "loss": 1.2497,
+      "step": 1356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019095598879254118,
+      "loss": 1.2725,
+      "step": 1357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019094276198242498,
+      "loss": 1.19,
+      "step": 1358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019092952596608273,
+      "loss": 1.1766,
+      "step": 1359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001909162807448543,
+      "loss": 1.2454,
+      "step": 1360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019090302632008047,
+      "loss": 1.2506,
+      "step": 1361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001908897626931031,
+      "loss": 1.2342,
+      "step": 1362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001908764898652648,
+      "loss": 1.2584,
+      "step": 1363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001908632078379092,
+      "loss": 1.2392,
+      "step": 1364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019084991661238085,
+      "loss": 1.2516,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019083661619002527,
+      "loss": 1.2816,
+      "step": 1366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019082330657218884,
+      "loss": 1.2044,
+      "step": 1367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019080998776021898,
+      "loss": 1.2436,
+      "step": 1368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019079665975546382,
+      "loss": 1.2493,
+      "step": 1369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001907833225592727,
+      "loss": 1.206,
+      "step": 1370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001907699761729957,
+      "loss": 1.3174,
+      "step": 1371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001907566205979839,
+      "loss": 1.2199,
+      "step": 1372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001907432558355893,
+      "loss": 1.2175,
+      "step": 1373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019072988188716483,
+      "loss": 1.1898,
+      "step": 1374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019071649875406434,
+      "loss": 1.2125,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001907031064376426,
+      "loss": 1.2539,
+      "step": 1376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019068970493925537,
+      "loss": 1.2747,
+      "step": 1377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019067629426025928,
+      "loss": 1.2327,
+      "step": 1378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906628744020119,
+      "loss": 1.2755,
+      "step": 1379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019064944536587172,
+      "loss": 1.2544,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906360071531982,
+      "loss": 1.2706,
+      "step": 1381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906225597653517,
+      "loss": 1.1581,
+      "step": 1382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906091032036935,
+      "loss": 1.199,
+      "step": 1383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019059563746958585,
+      "loss": 1.204,
+      "step": 1384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019058216256439187,
+      "loss": 1.2813,
+      "step": 1385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019056867848947563,
+      "loss": 1.2508,
+      "step": 1386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019055518524620216,
+      "loss": 1.2035,
+      "step": 1387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001905416828359374,
+      "loss": 1.2073,
+      "step": 1388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001905281712600482,
+      "loss": 1.2501,
+      "step": 1389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019051465051990235,
+      "loss": 1.2669,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019050112061686853,
+      "loss": 1.2793,
+      "step": 1391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019048758155231648,
+      "loss": 1.2071,
+      "step": 1392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019047403332761667,
+      "loss": 1.2074,
+      "step": 1393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019046047594414067,
+      "loss": 1.2402,
+      "step": 1394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001904469094032609,
+      "loss": 1.2256,
+      "step": 1395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001904333337063507,
+      "loss": 1.2349,
+      "step": 1396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001904197488547843,
+      "loss": 1.248,
+      "step": 1397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000190406154849937,
+      "loss": 1.2082,
+      "step": 1398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019039255169318487,
+      "loss": 1.2397,
+      "step": 1399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000190378939385905,
+      "loss": 1.2122,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001903653179294754,
+      "loss": 1.305,
+      "step": 1401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019035168732527494,
+      "loss": 1.2938,
+      "step": 1402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019033804757468344,
+      "loss": 1.2914,
+      "step": 1403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019032439867908173,
+      "loss": 1.3479,
+      "step": 1404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001903107406398515,
+      "loss": 1.2241,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019029707345837529,
+      "loss": 1.1608,
+      "step": 1406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019028339713603672,
+      "loss": 1.3098,
+      "step": 1407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001902697116742202,
+      "loss": 1.2696,
+      "step": 1408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019025601707431115,
+      "loss": 1.2027,
+      "step": 1409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001902423133376959,
+      "loss": 1.2309,
+      "step": 1410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019022860046576166,
+      "loss": 1.2685,
+      "step": 1411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019021487845989667,
+      "loss": 1.2523,
+      "step": 1412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019020114732148995,
+      "loss": 1.1835,
+      "step": 1413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019018740705193152,
+      "loss": 1.23,
+      "step": 1414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019017365765261235,
+      "loss": 1.247,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001901598991249243,
+      "loss": 1.3226,
+      "step": 1416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019014613147026017,
+      "loss": 1.2003,
+      "step": 1417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019013235469001364,
+      "loss": 1.2266,
+      "step": 1418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019011856878557934,
+      "loss": 1.2884,
+      "step": 1419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019010477375835288,
+      "loss": 1.2505,
+      "step": 1420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019009096960973074,
+      "loss": 1.2806,
+      "step": 1421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019007715634111025,
+      "loss": 1.2302,
+      "step": 1422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019006333395388985,
+      "loss": 1.2297,
+      "step": 1423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019004950244946872,
+      "loss": 1.2725,
+      "step": 1424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019003566182924705,
+      "loss": 1.183,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000190021812094626,
+      "loss": 1.1858,
+      "step": 1426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001900079532470075,
+      "loss": 1.2497,
+      "step": 1427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001899940852877945,
+      "loss": 1.1903,
+      "step": 1428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018998020821839098,
+      "loss": 1.3021,
+      "step": 1429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018996632204020162,
+      "loss": 1.2193,
+      "step": 1430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018995242675463216,
+      "loss": 1.3373,
+      "step": 1431
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001899385223630892,
+      "loss": 1.225,
+      "step": 1432
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001899246088669804,
+      "loss": 1.2587,
+      "step": 1433
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018991068626771415,
+      "loss": 1.2665,
+      "step": 1434
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018989675456669987,
+      "loss": 1.2216,
+      "step": 1435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018988281376534785,
+      "loss": 1.2152,
+      "step": 1436
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018986886386506942,
+      "loss": 1.2692,
+      "step": 1437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018985490486727664,
+      "loss": 1.3146,
+      "step": 1438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018984093677338262,
+      "loss": 1.2622,
+      "step": 1439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898269595848014,
+      "loss": 1.2262,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898129733029479,
+      "loss": 1.2743,
+      "step": 1441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018979897792923792,
+      "loss": 1.2239,
+      "step": 1442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018978497346508827,
+      "loss": 1.2782,
+      "step": 1443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018977095991191663,
+      "loss": 1.2634,
+      "step": 1444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018975693727114158,
+      "loss": 1.2534,
+      "step": 1445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018974290554418265,
+      "loss": 1.2608,
+      "step": 1446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018972886473246034,
+      "loss": 1.2465,
+      "step": 1447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018971481483739594,
+      "loss": 1.2751,
+      "step": 1448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001897007558604118,
+      "loss": 1.2377,
+      "step": 1449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018968668780293106,
+      "loss": 1.307,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001896726106663779,
+      "loss": 1.2106,
+      "step": 1451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018965852445217732,
+      "loss": 1.2473,
+      "step": 1452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001896444291617553,
+      "loss": 1.2172,
+      "step": 1453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018963032479653874,
+      "loss": 1.2241,
+      "step": 1454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001896162113579554,
+      "loss": 1.2265,
+      "step": 1455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018960208884743403,
+      "loss": 1.2384,
+      "step": 1456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018958795726640423,
+      "loss": 1.2476,
+      "step": 1457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001895738166162966,
+      "loss": 1.2384,
+      "step": 1458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018955966689854255,
+      "loss": 1.2168,
+      "step": 1459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018954550811457456,
+      "loss": 1.2193,
+      "step": 1460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018953134026582588,
+      "loss": 1.2786,
+      "step": 1461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001895171633537307,
+      "loss": 1.1463,
+      "step": 1462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018950297737972425,
+      "loss": 1.2082,
+      "step": 1463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018948878234524252,
+      "loss": 1.1818,
+      "step": 1464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018947457825172252,
+      "loss": 1.1553,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018946036510060211,
+      "loss": 1.2737,
+      "step": 1466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018944614289332016,
+      "loss": 1.2735,
+      "step": 1467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018943191163131638,
+      "loss": 1.2218,
+      "step": 1468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018941767131603138,
+      "loss": 1.2088,
+      "step": 1469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001894034219489067,
+      "loss": 1.2249,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018938916353138492,
+      "loss": 1.2106,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018937489606490936,
+      "loss": 1.2229,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018936061955092433,
+      "loss": 1.2435,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018934633399087506,
+      "loss": 1.2083,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018933203938620774,
+      "loss": 1.2667,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018931773573836936,
+      "loss": 1.2506,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001893034230488079,
+      "loss": 1.2426,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018928910131897232,
+      "loss": 1.2109,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018927477055031232,
+      "loss": 1.2189,
+      "step": 1479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001892604307442787,
+      "loss": 1.2,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018924608190232304,
+      "loss": 1.2392,
+      "step": 1481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018923172402589793,
+      "loss": 1.2007,
+      "step": 1482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018921735711645678,
+      "loss": 1.2691,
+      "step": 1483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018920298117545403,
+      "loss": 1.251,
+      "step": 1484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018918859620434492,
+      "loss": 1.1712,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.8611011505126953,
+      "eval_runtime": 500.7324,
+      "eval_samples_per_second": 51.724,
+      "eval_steps_per_second": 25.862,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891742022045857,
+      "loss": 1.2516,
+      "step": 1486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018915979917763344,
+      "loss": 1.2357,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891453871249462,
+      "loss": 1.1796,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018913096604798287,
+      "loss": 1.2458,
+      "step": 1489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018911653594820346,
+      "loss": 1.2337,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018910209682706857,
+      "loss": 1.2456,
+      "step": 1491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018908764868603997,
+      "loss": 1.2529,
+      "step": 1492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018907319152658026,
+      "loss": 1.1828,
+      "step": 1493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018905872535015297,
+      "loss": 1.2024,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018904425015822245,
+      "loss": 1.2213,
+      "step": 1495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018902976595225413,
+      "loss": 1.2361,
+      "step": 1496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018901527273371417,
+      "loss": 1.2127,
+      "step": 1497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018900077050406982,
+      "loss": 1.2185,
+      "step": 1498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018898625926478912,
+      "loss": 1.2546,
+      "step": 1499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018897173901734102,
+      "loss": 1.2651,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001889572097631955,
+      "loss": 1.1954,
+      "step": 1501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018894267150382324,
+      "loss": 1.2945,
+      "step": 1502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018892812424069612,
+      "loss": 1.2893,
+      "step": 1503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018891356797528665,
+      "loss": 1.2992,
+      "step": 1504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018889900270906845,
+      "loss": 1.257,
+      "step": 1505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018888442844351597,
+      "loss": 1.3043,
+      "step": 1506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018886984518010454,
+      "loss": 1.2835,
+      "step": 1507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018885525292031043,
+      "loss": 1.1789,
+      "step": 1508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001888406516656109,
+      "loss": 1.2448,
+      "step": 1509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018882604141748396,
+      "loss": 1.2597,
+      "step": 1510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018881142217740867,
+      "loss": 1.193,
+      "step": 1511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018879679394686496,
+      "loss": 1.2008,
+      "step": 1512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018878215672733366,
+      "loss": 1.2282,
+      "step": 1513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018876751052029645,
+      "loss": 1.2334,
+      "step": 1514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018875285532723605,
+      "loss": 1.2564,
+      "step": 1515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000188738191149636,
+      "loss": 1.2527,
+      "step": 1516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018872351798898077,
+      "loss": 1.2288,
+      "step": 1517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001887088358467557,
+      "loss": 1.2904,
+      "step": 1518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018869414472444715,
+      "loss": 1.2131,
+      "step": 1519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018867944462354225,
+      "loss": 1.2416,
+      "step": 1520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018866473554552914,
+      "loss": 1.2173,
+      "step": 1521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018865001749189682,
+      "loss": 1.2844,
+      "step": 1522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018863529046413526,
+      "loss": 1.2385,
+      "step": 1523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001886205544637352,
+      "loss": 1.1896,
+      "step": 1524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018860580949218848,
+      "loss": 1.2408,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018859105555098766,
+      "loss": 1.219,
+      "step": 1526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018857629264162635,
+      "loss": 1.1958,
+      "step": 1527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000188561520765599,
+      "loss": 1.1924,
+      "step": 1528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018854673992440099,
+      "loss": 1.2238,
+      "step": 1529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018853195011952857,
+      "loss": 1.1993,
+      "step": 1530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018851715135247897,
+      "loss": 1.2742,
+      "step": 1531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018850234362475024,
+      "loss": 1.2608,
+      "step": 1532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018848752693784144,
+      "loss": 1.2565,
+      "step": 1533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018847270129325244,
+      "loss": 1.3133,
+      "step": 1534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018845786669248403,
+      "loss": 1.273,
+      "step": 1535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018844302313703794,
+      "loss": 1.3008,
+      "step": 1536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018842817062841686,
+      "loss": 1.3006,
+      "step": 1537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018841330916812425,
+      "loss": 1.2351,
+      "step": 1538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018839843875766458,
+      "loss": 1.2351,
+      "step": 1539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018838355939854322,
+      "loss": 1.24,
+      "step": 1540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001883686710922664,
+      "loss": 1.2048,
+      "step": 1541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018835377384034124,
+      "loss": 1.2348,
+      "step": 1542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018833886764427584,
+      "loss": 1.2342,
+      "step": 1543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001883239525055792,
+      "loss": 1.2679,
+      "step": 1544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018830902842576115,
+      "loss": 1.1946,
+      "step": 1545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018829409540633248,
+      "loss": 1.2816,
+      "step": 1546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018827915344880487,
+      "loss": 1.1683,
+      "step": 1547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018826420255469093,
+      "loss": 1.1844,
+      "step": 1548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018824924272550412,
+      "loss": 1.2292,
+      "step": 1549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018823427396275888,
+      "loss": 1.1631,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001882192962679705,
+      "loss": 1.3208,
+      "step": 1551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018820430964265513,
+      "loss": 1.2349,
+      "step": 1552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018818931408832995,
+      "loss": 1.2021,
+      "step": 1553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000188174309606513,
+      "loss": 1.3181,
+      "step": 1554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001881592961987231,
+      "loss": 1.1824,
+      "step": 1555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018814427386648012,
+      "loss": 1.3131,
+      "step": 1556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001881292426113048,
+      "loss": 1.1927,
+      "step": 1557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001881142024347188,
+      "loss": 1.1794,
+      "step": 1558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018809915333824458,
+      "loss": 1.3065,
+      "step": 1559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018808409532340562,
+      "loss": 1.3085,
+      "step": 1560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018806902839172622,
+      "loss": 1.1936,
+      "step": 1561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018805395254473166,
+      "loss": 1.198,
+      "step": 1562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018803886778394807,
+      "loss": 1.2477,
+      "step": 1563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018802377411090247,
+      "loss": 1.1677,
+      "step": 1564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018800867152712284,
+      "loss": 1.2603,
+      "step": 1565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018799356003413803,
+      "loss": 1.2714,
+      "step": 1566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018797843963347776,
+      "loss": 1.268,
+      "step": 1567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018796331032667273,
+      "loss": 1.2945,
+      "step": 1568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018794817211525446,
+      "loss": 1.2032,
+      "step": 1569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001879330250007554,
+      "loss": 1.2854,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018791786898470893,
+      "loss": 1.139,
+      "step": 1571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018790270406864932,
+      "loss": 1.2356,
+      "step": 1572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018788753025411167,
+      "loss": 1.3009,
+      "step": 1573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018787234754263213,
+      "loss": 1.2751,
+      "step": 1574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018785715593574756,
+      "loss": 1.2078,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878419554349959,
+      "loss": 1.2622,
+      "step": 1576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878267460419159,
+      "loss": 1.2109,
+      "step": 1577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878115277580472,
+      "loss": 1.3006,
+      "step": 1578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018779630058493038,
+      "loss": 1.2074,
+      "step": 1579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018778106452410685,
+      "loss": 1.2542,
+      "step": 1580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018776581957711904,
+      "loss": 1.1864,
+      "step": 1581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018775056574551018,
+      "loss": 1.2773,
+      "step": 1582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018773530303082444,
+      "loss": 1.2246,
+      "step": 1583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018772003143460687,
+      "loss": 1.2362,
+      "step": 1584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018770475095840344,
+      "loss": 1.2215,
+      "step": 1585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018768946160376096,
+      "loss": 1.2767,
+      "step": 1586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018767416337222727,
+      "loss": 1.1698,
+      "step": 1587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018765885626535097,
+      "loss": 1.242,
+      "step": 1588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018764354028468164,
+      "loss": 1.2888,
+      "step": 1589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018762821543176966,
+      "loss": 1.2729,
+      "step": 1590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001876128817081665,
+      "loss": 1.24,
+      "step": 1591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018759753911542433,
+      "loss": 1.2818,
+      "step": 1592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018758218765509632,
+      "loss": 1.232,
+      "step": 1593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018756682732873648,
+      "loss": 1.261,
+      "step": 1594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001875514581378998,
+      "loss": 1.3113,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001875360800841421,
+      "loss": 1.2141,
+      "step": 1596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018752069316902006,
+      "loss": 1.263,
+      "step": 1597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001875052973940914,
+      "loss": 1.283,
+      "step": 1598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001874898927609146,
+      "loss": 1.1788,
+      "step": 1599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018747447927104912,
+      "loss": 1.208,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018745905692605524,
+      "loss": 1.3063,
+      "step": 1601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018744362572749423,
+      "loss": 1.266,
+      "step": 1602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018742818567692816,
+      "loss": 1.222,
+      "step": 1603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018741273677592006,
+      "loss": 1.1688,
+      "step": 1604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018739727902603384,
+      "loss": 1.2496,
+      "step": 1605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018738181242883428,
+      "loss": 1.2662,
+      "step": 1606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018736633698588712,
+      "loss": 1.2063,
+      "step": 1607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018735085269875893,
+      "loss": 1.2507,
+      "step": 1608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001873353595690172,
+      "loss": 1.2471,
+      "step": 1609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018731985759823033,
+      "loss": 1.2144,
+      "step": 1610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018730434678796754,
+      "loss": 1.3024,
+      "step": 1611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872888271397991,
+      "loss": 1.2693,
+      "step": 1612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000187273298655296,
+      "loss": 1.2226,
+      "step": 1613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018725776133603027,
+      "loss": 1.25,
+      "step": 1614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872422151835747,
+      "loss": 1.2008,
+      "step": 1615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872266601995031,
+      "loss": 1.2315,
+      "step": 1616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872110963853901,
+      "loss": 1.1504,
+      "step": 1617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001871955237428112,
+      "loss": 1.181,
+      "step": 1618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018717994227334287,
+      "loss": 1.2828,
+      "step": 1619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018716435197856245,
+      "loss": 1.2054,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018714875286004814,
+      "loss": 1.1594,
+      "step": 1621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018713314491937907,
+      "loss": 1.2461,
+      "step": 1622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018711752815813525,
+      "loss": 1.2413,
+      "step": 1623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018710190257789755,
+      "loss": 1.2167,
+      "step": 1624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018708626818024775,
+      "loss": 1.209,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018707062496676859,
+      "loss": 1.1833,
+      "step": 1626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018705497293904365,
+      "loss": 1.2465,
+      "step": 1627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018703931209865734,
+      "loss": 1.2086,
+      "step": 1628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018702364244719508,
+      "loss": 1.2185,
+      "step": 1629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018700796398624306,
+      "loss": 1.25,
+      "step": 1630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018699227671738853,
+      "loss": 1.3126,
+      "step": 1631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001869765806422194,
+      "loss": 1.3034,
+      "step": 1632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018696087576232473,
+      "loss": 1.3339,
+      "step": 1633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018694516207929423,
+      "loss": 1.3173,
+      "step": 1634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018692943959471866,
+      "loss": 1.2107,
+      "step": 1635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018691370831018964,
+      "loss": 1.1942,
+      "step": 1636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018689796822729963,
+      "loss": 1.1882,
+      "step": 1637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018688221934764202,
+      "loss": 1.288,
+      "step": 1638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018686646167281111,
+      "loss": 1.1881,
+      "step": 1639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018685069520440206,
+      "loss": 1.2012,
+      "step": 1640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018683491994401087,
+      "loss": 1.2473,
+      "step": 1641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018681913589323456,
+      "loss": 1.2152,
+      "step": 1642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001868033430536709,
+      "loss": 1.2152,
+      "step": 1643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018678754142691868,
+      "loss": 1.2516,
+      "step": 1644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018677173101457748,
+      "loss": 1.2057,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001867559118182478,
+      "loss": 1.1898,
+      "step": 1646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000186740083839531,
+      "loss": 1.2289,
+      "step": 1647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018672424708002943,
+      "loss": 1.2664,
+      "step": 1648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018670840154134626,
+      "loss": 1.2632,
+      "step": 1649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018669254722508552,
+      "loss": 1.2103,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001866766841328521,
+      "loss": 1.2665,
+      "step": 1651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018666081226625196,
+      "loss": 1.208,
+      "step": 1652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001866449316268917,
+      "loss": 1.2701,
+      "step": 1653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018662904221637905,
+      "loss": 1.1832,
+      "step": 1654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018661314403632244,
+      "loss": 1.2951,
+      "step": 1655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018659723708833124,
+      "loss": 1.1713,
+      "step": 1656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018658132137401579,
+      "loss": 1.1882,
+      "step": 1657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001865653968949872,
+      "loss": 1.2419,
+      "step": 1658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018654946365285758,
+      "loss": 1.2448,
+      "step": 1659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001865335216492398,
+      "loss": 1.2325,
+      "step": 1660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018651757088574772,
+      "loss": 1.2013,
+      "step": 1661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018650161136399604,
+      "loss": 1.2763,
+      "step": 1662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018648564308560038,
+      "loss": 1.2244,
+      "step": 1663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018646966605217723,
+      "loss": 1.2156,
+      "step": 1664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018645368026534392,
+      "loss": 1.2298,
+      "step": 1665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018643768572671873,
+      "loss": 1.22,
+      "step": 1666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001864216824379208,
+      "loss": 1.2367,
+      "step": 1667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018640567040057016,
+      "loss": 1.2141,
+      "step": 1668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018638964961628773,
+      "loss": 1.2727,
+      "step": 1669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001863736200866953,
+      "loss": 1.2248,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018635758181341557,
+      "loss": 1.2288,
+      "step": 1671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001863415347980721,
+      "loss": 1.2306,
+      "step": 1672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018632547904228933,
+      "loss": 1.2385,
+      "step": 1673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018630941454769262,
+      "loss": 1.2153,
+      "step": 1674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018629334131590818,
+      "loss": 1.3313,
+      "step": 1675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018627725934856315,
+      "loss": 1.2592,
+      "step": 1676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001862611686472855,
+      "loss": 1.2337,
+      "step": 1677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001862450692137041,
+      "loss": 1.2798,
+      "step": 1678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018622896104944874,
+      "loss": 1.2888,
+      "step": 1679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018621284415615004,
+      "loss": 1.1777,
+      "step": 1680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018619671853543955,
+      "loss": 1.2156,
+      "step": 1681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018618058418894964,
+      "loss": 1.2774,
+      "step": 1682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018616444111831367,
+      "loss": 1.2208,
+      "step": 1683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018614828932516573,
+      "loss": 1.2365,
+      "step": 1684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000186132128811141,
+      "loss": 1.2342,
+      "step": 1685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018611595957787532,
+      "loss": 1.2198,
+      "step": 1686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001860997816270056,
+      "loss": 1.2639,
+      "step": 1687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018608359496016946,
+      "loss": 1.2117,
+      "step": 1688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018606739957900556,
+      "loss": 1.2297,
+      "step": 1689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018605119548515339,
+      "loss": 1.2703,
+      "step": 1690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001860349826802532,
+      "loss": 1.2481,
+      "step": 1691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018601876116594636,
+      "loss": 1.3281,
+      "step": 1692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001860025309438749,
+      "loss": 1.2361,
+      "step": 1693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001859862920156819,
+      "loss": 1.2772,
+      "step": 1694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018597004438301117,
+      "loss": 1.1468,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018595378804750747,
+      "loss": 1.1707,
+      "step": 1696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001859375230108165,
+      "loss": 1.2688,
+      "step": 1697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018592124927458475,
+      "loss": 1.2975,
+      "step": 1698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018590496684045964,
+      "loss": 1.2488,
+      "step": 1699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018588867571008946,
+      "loss": 1.2244,
+      "step": 1700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018587237588512333,
+      "loss": 1.1869,
+      "step": 1701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001858560673672114,
+      "loss": 1.2624,
+      "step": 1702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018583975015800452,
+      "loss": 1.3048,
+      "step": 1703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001858234242591545,
+      "loss": 1.2807,
+      "step": 1704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018580708967231403,
+      "loss": 1.2536,
+      "step": 1705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018579074639913672,
+      "loss": 1.2104,
+      "step": 1706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018577439444127696,
+      "loss": 1.2458,
+      "step": 1707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018575803380039006,
+      "loss": 1.2183,
+      "step": 1708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018574166447813233,
+      "loss": 1.2385,
+      "step": 1709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018572528647616075,
+      "loss": 1.187,
+      "step": 1710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001857088997961333,
+      "loss": 1.2588,
+      "step": 1711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018569250443970886,
+      "loss": 1.2471,
+      "step": 1712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018567610040854713,
+      "loss": 1.2058,
+      "step": 1713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001856596877043087,
+      "loss": 1.2402,
+      "step": 1714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000185643266328655,
+      "loss": 1.2473,
+      "step": 1715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018562683628324845,
+      "loss": 1.3298,
+      "step": 1716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018561039756975227,
+      "loss": 1.253,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018559395018983055,
+      "loss": 1.2101,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018557749414514828,
+      "loss": 1.2222,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001855610294373713,
+      "loss": 1.1573,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001855445560681664,
+      "loss": 1.2647,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001855280740392011,
+      "loss": 1.2712,
+      "step": 1722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000185511583352144,
+      "loss": 1.249,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018549508400866443,
+      "loss": 1.2767,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018547857601043264,
+      "loss": 1.2481,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018546205935911973,
+      "loss": 1.1727,
+      "step": 1726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001854455340563977,
+      "loss": 1.2452,
+      "step": 1727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018542900010393943,
+      "loss": 1.2574,
+      "step": 1728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018541245750341867,
+      "loss": 1.2345,
+      "step": 1729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018539590625651003,
+      "loss": 1.1614,
+      "step": 1730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018537934636488903,
+      "loss": 1.199,
+      "step": 1731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018536277783023203,
+      "loss": 1.2489,
+      "step": 1732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001853462006542163,
+      "loss": 1.2524,
+      "step": 1733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018532961483851997,
+      "loss": 1.1938,
+      "step": 1734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000185313020384822,
+      "loss": 1.2113,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018529641729480225,
+      "loss": 1.2252,
+      "step": 1736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018527980557014152,
+      "loss": 1.2356,
+      "step": 1737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018526318521252142,
+      "loss": 1.2865,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018524655622362442,
+      "loss": 1.2558,
+      "step": 1739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001852299186051339,
+      "loss": 1.2595,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018521327235873413,
+      "loss": 1.2579,
+      "step": 1741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018519661748611017,
+      "loss": 1.2398,
+      "step": 1742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018517995398894808,
+      "loss": 1.2487,
+      "step": 1743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018516328186893465,
+      "loss": 1.2152,
+      "step": 1744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018514660112775767,
+      "loss": 1.2681,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001851299117671057,
+      "loss": 1.2611,
+      "step": 1746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001851132137886683,
+      "loss": 1.2233,
+      "step": 1747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018509650719413572,
+      "loss": 1.1737,
+      "step": 1748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018507979198519926,
+      "loss": 1.291,
+      "step": 1749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000185063068163551,
+      "loss": 1.2729,
+      "step": 1750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001850463357308839,
+      "loss": 1.2004,
+      "step": 1751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018502959468889177,
+      "loss": 1.2901,
+      "step": 1752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018501284503926938,
+      "loss": 1.2699,
+      "step": 1753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018499608678371229,
+      "loss": 1.2393,
+      "step": 1754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018497931992391697,
+      "loss": 1.1824,
+      "step": 1755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849625444615807,
+      "loss": 1.242,
+      "step": 1756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018494576039840171,
+      "loss": 1.2257,
+      "step": 1757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018492896773607907,
+      "loss": 1.2477,
+      "step": 1758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849121664763127,
+      "loss": 1.1805,
+      "step": 1759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018489535662080347,
+      "loss": 1.2588,
+      "step": 1760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018487853817125295,
+      "loss": 1.2754,
+      "step": 1761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848617111293638,
+      "loss": 1.2364,
+      "step": 1762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018484487549683934,
+      "loss": 1.2875,
+      "step": 1763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848280312753839,
+      "loss": 1.1623,
+      "step": 1764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018481117846670266,
+      "loss": 1.2081,
+      "step": 1765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018479431707250164,
+      "loss": 1.1854,
+      "step": 1766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001847774470944877,
+      "loss": 1.1821,
+      "step": 1767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018476056853436863,
+      "loss": 1.1538,
+      "step": 1768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018474368139385306,
+      "loss": 1.2111,
+      "step": 1769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001847267856746505,
+      "loss": 1.2607,
+      "step": 1770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001847098813784713,
+      "loss": 1.2231,
+      "step": 1771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018469296850702672,
+      "loss": 1.1754,
+      "step": 1772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018467604706202883,
+      "loss": 1.2809,
+      "step": 1773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018465911704519067,
+      "loss": 1.2487,
+      "step": 1774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018464217845822603,
+      "loss": 1.1793,
+      "step": 1775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018462523130284962,
+      "loss": 1.2789,
+      "step": 1776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018460827558077706,
+      "loss": 1.2268,
+      "step": 1777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018459131129372475,
+      "loss": 1.2864,
+      "step": 1778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018457433844341003,
+      "loss": 1.2048,
+      "step": 1779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001845573570315511,
+      "loss": 1.2613,
+      "step": 1780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018454036705986692,
+      "loss": 1.3062,
+      "step": 1781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018452336853007746,
+      "loss": 1.1829,
+      "step": 1782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001845063614439035,
+      "loss": 1.2248,
+      "step": 1783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018448934580306668,
+      "loss": 1.3004,
+      "step": 1784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001844723216092895,
+      "loss": 1.2165,
+      "step": 1785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018445528886429534,
+      "loss": 1.221,
+      "step": 1786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018443824756980844,
+      "loss": 1.2094,
+      "step": 1787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018442119772755394,
+      "loss": 1.1897,
+      "step": 1788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018440413933925776,
+      "loss": 1.2514,
+      "step": 1789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018438707240664675,
+      "loss": 1.2586,
+      "step": 1790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018436999693144864,
+      "loss": 1.2781,
+      "step": 1791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018435291291539198,
+      "loss": 1.2159,
+      "step": 1792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018433582036020621,
+      "loss": 1.1751,
+      "step": 1793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018431871926762164,
+      "loss": 1.2594,
+      "step": 1794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018430160963936936,
+      "loss": 1.2016,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001842844914771815,
+      "loss": 1.32,
+      "step": 1796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018426736478279084,
+      "loss": 1.2395,
+      "step": 1797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001842502295579312,
+      "loss": 1.2665,
+      "step": 1798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001842330858043372,
+      "loss": 1.2477,
+      "step": 1799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018421593352374428,
+      "loss": 1.2033,
+      "step": 1800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001841987727178888,
+      "loss": 1.1984,
+      "step": 1801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018418160338850797,
+      "loss": 1.2576,
+      "step": 1802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001841644255373399,
+      "loss": 1.2212,
+      "step": 1803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001841472391661234,
+      "loss": 1.1995,
+      "step": 1804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018413004427659843,
+      "loss": 1.2023,
+      "step": 1805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001841128408705055,
+      "loss": 1.2403,
+      "step": 1806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018409562894958618,
+      "loss": 1.269,
+      "step": 1807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840784085155829,
+      "loss": 1.1914,
+      "step": 1808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018406117957023882,
+      "loss": 1.3035,
+      "step": 1809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018404394211529807,
+      "loss": 1.2318,
+      "step": 1810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018402669615250565,
+      "loss": 1.2485,
+      "step": 1811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018400944168360734,
+      "loss": 1.2446,
+      "step": 1812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839921787103499,
+      "loss": 1.2779,
+      "step": 1813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018397490723448073,
+      "loss": 1.2034,
+      "step": 1814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839576272577484,
+      "loss": 1.2324,
+      "step": 1815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018394033878190209,
+      "loss": 1.3093,
+      "step": 1816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018392304180869196,
+      "loss": 1.2576,
+      "step": 1817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018390573633986898,
+      "loss": 1.2511,
+      "step": 1818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018388842237718504,
+      "loss": 1.251,
+      "step": 1819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001838710999223928,
+      "loss": 1.2321,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018385376897724586,
+      "loss": 1.1996,
+      "step": 1821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018383642954349864,
+      "loss": 1.2946,
+      "step": 1822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018381908162290643,
+      "loss": 1.2469,
+      "step": 1823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018380172521722537,
+      "loss": 1.2348,
+      "step": 1824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018378436032821246,
+      "loss": 1.2054,
+      "step": 1825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018376698695762557,
+      "loss": 1.2548,
+      "step": 1826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018374960510722342,
+      "loss": 1.2077,
+      "step": 1827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018373221477876566,
+      "loss": 1.2652,
+      "step": 1828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018371481597401264,
+      "loss": 1.2125,
+      "step": 1829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018369740869472566,
+      "loss": 1.2503,
+      "step": 1830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018367999294266696,
+      "loss": 1.2674,
+      "step": 1831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018366256871959945,
+      "loss": 1.2367,
+      "step": 1832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836451360272871,
+      "loss": 1.2118,
+      "step": 1833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018362769486749455,
+      "loss": 1.2089,
+      "step": 1834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018361024524198746,
+      "loss": 1.2924,
+      "step": 1835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001835927871525322,
+      "loss": 1.2539,
+      "step": 1836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018357532060089612,
+      "loss": 1.1396,
+      "step": 1837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018355784558884736,
+      "loss": 1.2372,
+      "step": 1838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018354036211815498,
+      "loss": 1.2642,
+      "step": 1839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018352287019058876,
+      "loss": 1.2802,
+      "step": 1840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001835053698079195,
+      "loss": 1.3,
+      "step": 1841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018348786097191876,
+      "loss": 1.2537,
+      "step": 1842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018347034368435892,
+      "loss": 1.2297,
+      "step": 1843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018345281794701336,
+      "loss": 1.1964,
+      "step": 1844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001834352837616562,
+      "loss": 1.1448,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001834177411300624,
+      "loss": 1.2239,
+      "step": 1846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001834001900540079,
+      "loss": 1.203,
+      "step": 1847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018338263053526932,
+      "loss": 1.2271,
+      "step": 1848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001833650625756243,
+      "loss": 1.2378,
+      "step": 1849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001833474861768512,
+      "loss": 1.2052,
+      "step": 1850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018332990134072933,
+      "loss": 1.2675,
+      "step": 1851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018331230806903888,
+      "loss": 1.21,
+      "step": 1852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001832947063635607,
+      "loss": 1.2464,
+      "step": 1853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018327709622607675,
+      "loss": 1.226,
+      "step": 1854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018325947765836966,
+      "loss": 1.2391,
+      "step": 1855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000183241850662223,
+      "loss": 1.2277,
+      "step": 1856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018322421523942115,
+      "loss": 1.2176,
+      "step": 1857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018320657139174937,
+      "loss": 1.144,
+      "step": 1858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018318891912099376,
+      "loss": 1.2155,
+      "step": 1859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001831712584289413,
+      "loss": 1.3157,
+      "step": 1860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018315358931737973,
+      "loss": 1.2148,
+      "step": 1861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001831359117880978,
+      "loss": 1.3078,
+      "step": 1862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000183118225842885,
+      "loss": 1.1944,
+      "step": 1863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018310053148353167,
+      "loss": 1.1815,
+      "step": 1864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000183082828711829,
+      "loss": 1.2061,
+      "step": 1865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018306511752956914,
+      "loss": 1.192,
+      "step": 1866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018304739793854497,
+      "loss": 1.115,
+      "step": 1867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018302966994055025,
+      "loss": 1.2968,
+      "step": 1868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001830119335373796,
+      "loss": 1.1799,
+      "step": 1869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001829941887308285,
+      "loss": 1.2018,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018297643552269333,
+      "loss": 1.2695,
+      "step": 1871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018295867391477115,
+      "loss": 1.3009,
+      "step": 1872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001829409039088601,
+      "loss": 1.246,
+      "step": 1873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000182923125506759,
+      "loss": 1.2428,
+      "step": 1874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001829053387102676,
+      "loss": 1.2084,
+      "step": 1875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001828875435211864,
+      "loss": 1.2299,
+      "step": 1876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001828697399413169,
+      "loss": 1.295,
+      "step": 1877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018285192797246137,
+      "loss": 1.2795,
+      "step": 1878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018283410761642293,
+      "loss": 1.1701,
+      "step": 1879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018281627887500553,
+      "loss": 1.1944,
+      "step": 1880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000182798441750014,
+      "loss": 1.2868,
+      "step": 1881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000182780596243254,
+      "loss": 1.2758,
+      "step": 1882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001827627423565321,
+      "loss": 1.2535,
+      "step": 1883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018274488009165557,
+      "loss": 1.252,
+      "step": 1884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018272700945043272,
+      "loss": 1.259,
+      "step": 1885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018270913043467256,
+      "loss": 1.2537,
+      "step": 1886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018269124304618503,
+      "loss": 1.2226,
+      "step": 1887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018267334728678088,
+      "loss": 1.2828,
+      "step": 1888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001826554431582717,
+      "loss": 1.2485,
+      "step": 1889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018263753066246995,
+      "loss": 1.2303,
+      "step": 1890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018261960980118892,
+      "loss": 1.2448,
+      "step": 1891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018260168057624275,
+      "loss": 1.1922,
+      "step": 1892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001825837429894465,
+      "loss": 1.2023,
+      "step": 1893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018256579704261597,
+      "loss": 1.2032,
+      "step": 1894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001825478427375678,
+      "loss": 1.1654,
+      "step": 1895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018252988007611955,
+      "loss": 1.2123,
+      "step": 1896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001825119090600896,
+      "loss": 1.1708,
+      "step": 1897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018249392969129722,
+      "loss": 1.2234,
+      "step": 1898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018247594197156244,
+      "loss": 1.2735,
+      "step": 1899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018245794590270614,
+      "loss": 1.2276,
+      "step": 1900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018243994148655016,
+      "loss": 1.2131,
+      "step": 1901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000182421928724917,
+      "loss": 1.2433,
+      "step": 1902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018240390761963023,
+      "loss": 1.1505,
+      "step": 1903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018238587817251406,
+      "loss": 1.208,
+      "step": 1904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018236784038539366,
+      "loss": 1.2049,
+      "step": 1905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000182349794260095,
+      "loss": 1.1766,
+      "step": 1906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001823317397984449,
+      "loss": 1.3081,
+      "step": 1907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018231367700227103,
+      "loss": 1.2085,
+      "step": 1908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018229560587340197,
+      "loss": 1.2709,
+      "step": 1909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000182277526413667,
+      "loss": 1.2031,
+      "step": 1910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018225943862489635,
+      "loss": 1.252,
+      "step": 1911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018224134250892105,
+      "loss": 1.2818,
+      "step": 1912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018222323806757303,
+      "loss": 1.2053,
+      "step": 1913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018220512530268498,
+      "loss": 1.2092,
+      "step": 1914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018218700421609047,
+      "loss": 1.159,
+      "step": 1915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018216887480962396,
+      "loss": 1.1613,
+      "step": 1916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018215073708512068,
+      "loss": 1.2568,
+      "step": 1917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001821325910444167,
+      "loss": 1.2018,
+      "step": 1918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000182114436689349,
+      "loss": 1.2084,
+      "step": 1919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018209627402175535,
+      "loss": 1.2707,
+      "step": 1920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001820781030434744,
+      "loss": 1.2661,
+      "step": 1921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018205992375634557,
+      "loss": 1.2785,
+      "step": 1922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018204173616220918,
+      "loss": 1.2539,
+      "step": 1923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018202354026290642,
+      "loss": 1.1825,
+      "step": 1924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018200533606027924,
+      "loss": 1.3025,
+      "step": 1925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001819871235561705,
+      "loss": 1.2745,
+      "step": 1926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018196890275242383,
+      "loss": 1.1989,
+      "step": 1927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018195067365088375,
+      "loss": 1.2047,
+      "step": 1928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018193243625339562,
+      "loss": 1.2559,
+      "step": 1929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018191419056180563,
+      "loss": 1.1766,
+      "step": 1930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001818959365779608,
+      "loss": 1.1904,
+      "step": 1931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018187767430370903,
+      "loss": 1.1434,
+      "step": 1932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000181859403740899,
+      "loss": 1.1883,
+      "step": 1933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018184112489138025,
+      "loss": 1.2634,
+      "step": 1934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001818228377570032,
+      "loss": 1.2432,
+      "step": 1935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018180454233961902,
+      "loss": 1.2279,
+      "step": 1936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018178623864107985,
+      "loss": 1.1665,
+      "step": 1937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018176792666323852,
+      "loss": 1.2132,
+      "step": 1938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018174960640794877,
+      "loss": 1.2241,
+      "step": 1939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018173127787706524,
+      "loss": 1.1817,
+      "step": 1940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018171294107244333,
+      "loss": 1.2645,
+      "step": 1941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018169459599593925,
+      "loss": 1.1882,
+      "step": 1942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001816762426494101,
+      "loss": 1.2224,
+      "step": 1943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018165788103471385,
+      "loss": 1.2058,
+      "step": 1944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018163951115370922,
+      "loss": 1.187,
+      "step": 1945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018162113300825583,
+      "loss": 1.1899,
+      "step": 1946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001816027466002141,
+      "loss": 1.2214,
+      "step": 1947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018158435193144535,
+      "loss": 1.2278,
+      "step": 1948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001815659490038116,
+      "loss": 1.2015,
+      "step": 1949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018154753781917591,
+      "loss": 1.243,
+      "step": 1950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000181529118379402,
+      "loss": 1.2336,
+      "step": 1951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018151069068635448,
+      "loss": 1.227,
+      "step": 1952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018149225474189886,
+      "loss": 1.2318,
+      "step": 1953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018147381054790137,
+      "loss": 1.211,
+      "step": 1954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018145535810622912,
+      "loss": 1.2039,
+      "step": 1955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018143689741875017,
+      "loss": 1.196,
+      "step": 1956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001814184284873332,
+      "loss": 1.2161,
+      "step": 1957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018139995131384788,
+      "loss": 1.2234,
+      "step": 1958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001813814659001647,
+      "loss": 1.2932,
+      "step": 1959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018136297224815494,
+      "loss": 1.2586,
+      "step": 1960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001813444703596907,
+      "loss": 1.2498,
+      "step": 1961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018132596023664498,
+      "loss": 1.2366,
+      "step": 1962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001813074418808916,
+      "loss": 1.1951,
+      "step": 1963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018128891529430513,
+      "loss": 1.1514,
+      "step": 1964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018127038047876105,
+      "loss": 1.2893,
+      "step": 1965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018125183743613569,
+      "loss": 1.1829,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018123328616830614,
+      "loss": 1.276,
+      "step": 1967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001812147266771504,
+      "loss": 1.1692,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018119615896454726,
+      "loss": 1.2471,
+      "step": 1969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018117758303237634,
+      "loss": 1.2956,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001811589988825181,
+      "loss": 1.2168,
+      "step": 1971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018114040651685383,
+      "loss": 1.2094,
+      "step": 1972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018112180593726564,
+      "loss": 1.2681,
+      "step": 1973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018110319714563652,
+      "loss": 1.2503,
+      "step": 1974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018108458014385023,
+      "loss": 1.2138,
+      "step": 1975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018106595493379136,
+      "loss": 1.1992,
+      "step": 1976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018104732151734542,
+      "loss": 1.281,
+      "step": 1977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001810286798963987,
+      "loss": 1.2281,
+      "step": 1978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001810100300728382,
+      "loss": 1.1316,
+      "step": 1979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018099137204855202,
+      "loss": 1.2129,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.8508283495903015,
+      "eval_runtime": 500.5157,
+      "eval_samples_per_second": 51.747,
+      "eval_steps_per_second": 25.873,
+      "step": 1980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018097270582542879,
+      "loss": 1.2396,
+      "step": 1981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018095403140535815,
+      "loss": 1.2075,
+      "step": 1982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018093534879023055,
+      "loss": 1.2388,
+      "step": 1983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018091665798193723,
+      "loss": 1.2327,
+      "step": 1984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001808979589823703,
+      "loss": 1.2325,
+      "step": 1985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001808792517934227,
+      "loss": 1.1849,
+      "step": 1986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018086053641698814,
+      "loss": 1.2704,
+      "step": 1987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018084181285496118,
+      "loss": 1.2738,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018082308110923726,
+      "loss": 1.2473,
+      "step": 1989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001808043411817126,
+      "loss": 1.2618,
+      "step": 1990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018078559307428427,
+      "loss": 1.2637,
+      "step": 1991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018076683678885014,
+      "loss": 1.2178,
+      "step": 1992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018074807232730894,
+      "loss": 1.2299,
+      "step": 1993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001807292996915602,
+      "loss": 1.2038,
+      "step": 1994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018071051888350436,
+      "loss": 1.2153,
+      "step": 1995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018069172990504252,
+      "loss": 1.2836,
+      "step": 1996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018067293275807677,
+      "loss": 1.223,
+      "step": 1997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018065412744450992,
+      "loss": 1.233,
+      "step": 1998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018063531396624572,
+      "loss": 1.2489,
+      "step": 1999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001806164923251886,
+      "loss": 1.1977,
+      "step": 2000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018059766252324398,
+      "loss": 1.2772,
+      "step": 2001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018057882456231793,
+      "loss": 1.2066,
+      "step": 2002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018055997844431746,
+      "loss": 1.2268,
+      "step": 2003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001805411241711504,
+      "loss": 1.2848,
+      "step": 2004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001805222617447254,
+      "loss": 1.2095,
+      "step": 2005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018050339116695193,
+      "loss": 1.2113,
+      "step": 2006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001804845124397402,
+      "loss": 1.2578,
+      "step": 2007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018046562556500143,
+      "loss": 1.2054,
+      "step": 2008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001804467305446474,
+      "loss": 1.2079,
+      "step": 2009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018042782738059108,
+      "loss": 1.2312,
+      "step": 2010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001804089160747459,
+      "loss": 1.2512,
+      "step": 2011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018038999662902633,
+      "loss": 1.2237,
+      "step": 2012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001803710690453476,
+      "loss": 1.2626,
+      "step": 2013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018035213332562574,
+      "loss": 1.2083,
+      "step": 2014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018033318947177765,
+      "loss": 1.1677,
+      "step": 2015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018031423748572105,
+      "loss": 1.2782,
+      "step": 2016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018029527736937445,
+      "loss": 1.2297,
+      "step": 2017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018027630912465723,
+      "loss": 1.2208,
+      "step": 2018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001802573327534895,
+      "loss": 1.2459,
+      "step": 2019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018023834825779233,
+      "loss": 1.2363,
+      "step": 2020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018021935563948752,
+      "loss": 1.2137,
+      "step": 2021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018020035490049768,
+      "loss": 1.2376,
+      "step": 2022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001801813460427463,
+      "loss": 1.2558,
+      "step": 2023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018016232906815764,
+      "loss": 1.2186,
+      "step": 2024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018014330397865688,
+      "loss": 1.1849,
+      "step": 2025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018012427077616984,
+      "loss": 1.2733,
+      "step": 2026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001801052294626233,
+      "loss": 1.2732,
+      "step": 2027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018008618003994493,
+      "loss": 1.2226,
+      "step": 2028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000180067122510063,
+      "loss": 1.2436,
+      "step": 2029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018004805687490678,
+      "loss": 1.2792,
+      "step": 2030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018002898313640632,
+      "loss": 1.2288,
+      "step": 2031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018000990129649243,
+      "loss": 1.2285,
+      "step": 2032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017999081135709679,
+      "loss": 1.23,
+      "step": 2033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001799717133201519,
+      "loss": 1.2027,
+      "step": 2034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017995260718759107,
+      "loss": 1.256,
+      "step": 2035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017993349296134849,
+      "loss": 1.2491,
+      "step": 2036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017991437064335903,
+      "loss": 1.1714,
+      "step": 2037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017989524023555848,
+      "loss": 1.2093,
+      "step": 2038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017987610173988348,
+      "loss": 1.2654,
+      "step": 2039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017985695515827137,
+      "loss": 1.1699,
+      "step": 2040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001798378004926604,
+      "loss": 1.2246,
+      "step": 2041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017981863774498965,
+      "loss": 1.1893,
+      "step": 2042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000179799466917199,
+      "loss": 1.2379,
+      "step": 2043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000179780288011229,
+      "loss": 1.2484,
+      "step": 2044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017976110102902132,
+      "loss": 1.2864,
+      "step": 2045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001797419059725182,
+      "loss": 1.2607,
+      "step": 2046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017972270284366272,
+      "loss": 1.1964,
+      "step": 2047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001797034916443989,
+      "loss": 1.1906,
+      "step": 2048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001796842723766715,
+      "loss": 1.2337,
+      "step": 2049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001796650450424261,
+      "loss": 1.2778,
+      "step": 2050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001796458096436091,
+      "loss": 1.223,
+      "step": 2051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017962656618216775,
+      "loss": 1.2217,
+      "step": 2052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017960731466005003,
+      "loss": 1.2882,
+      "step": 2053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001795880550792048,
+      "loss": 1.2127,
+      "step": 2054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017956878744158176,
+      "loss": 1.2523,
+      "step": 2055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001795495117491314,
+      "loss": 1.2216,
+      "step": 2056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017953022800380495,
+      "loss": 1.2484,
+      "step": 2057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017951093620755456,
+      "loss": 1.2513,
+      "step": 2058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017949163636233318,
+      "loss": 1.2363,
+      "step": 2059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017947232847009457,
+      "loss": 1.257,
+      "step": 2060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001794530125327932,
+      "loss": 1.2287,
+      "step": 2061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017943368855238452,
+      "loss": 1.2917,
+      "step": 2062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017941435653082467,
+      "loss": 1.196,
+      "step": 2063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001793950164700707,
+      "loss": 1.1973,
+      "step": 2064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017937566837208035,
+      "loss": 1.2118,
+      "step": 2065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017935631223881232,
+      "loss": 1.2336,
+      "step": 2066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000179336948072226,
+      "loss": 1.2424,
+      "step": 2067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017931757587428168,
+      "loss": 1.239,
+      "step": 2068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001792981956469404,
+      "loss": 1.1858,
+      "step": 2069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017927880739216409,
+      "loss": 1.2975,
+      "step": 2070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017925941111191534,
+      "loss": 1.1769,
+      "step": 2071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017924000680815777,
+      "loss": 1.2918,
+      "step": 2072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017922059448285558,
+      "loss": 1.2643,
+      "step": 2073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000179201174137974,
+      "loss": 1.2509,
+      "step": 2074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017918174577547896,
+      "loss": 1.1692,
+      "step": 2075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017916230939733714,
+      "loss": 1.3075,
+      "step": 2076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001791428650055162,
+      "loss": 1.2389,
+      "step": 2077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017912341260198442,
+      "loss": 1.2605,
+      "step": 2078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017910395218871105,
+      "loss": 1.1557,
+      "step": 2079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017908448376766607,
+      "loss": 1.2018,
+      "step": 2080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001790650073408203,
+      "loss": 1.1497,
+      "step": 2081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017904552291014532,
+      "loss": 1.2438,
+      "step": 2082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001790260304776136,
+      "loss": 1.2386,
+      "step": 2083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017900653004519835,
+      "loss": 1.2944,
+      "step": 2084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017898702161487363,
+      "loss": 1.2142,
+      "step": 2085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017896750518861429,
+      "loss": 1.3227,
+      "step": 2086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017894798076839601,
+      "loss": 1.1666,
+      "step": 2087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017892844835619526,
+      "loss": 1.2212,
+      "step": 2088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001789089079539893,
+      "loss": 1.2989,
+      "step": 2089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001788893595637563,
+      "loss": 1.2253,
+      "step": 2090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001788698031874751,
+      "loss": 1.2171,
+      "step": 2091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001788502388271254,
+      "loss": 1.2658,
+      "step": 2092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017883066648468775,
+      "loss": 1.1943,
+      "step": 2093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001788110861621435,
+      "loss": 1.2056,
+      "step": 2094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017879149786147473,
+      "loss": 1.2592,
+      "step": 2095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017877190158466442,
+      "loss": 1.172,
+      "step": 2096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017875229733369628,
+      "loss": 1.1497,
+      "step": 2097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017873268511055495,
+      "loss": 1.2614,
+      "step": 2098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001787130649172257,
+      "loss": 1.2651,
+      "step": 2099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017869343675569477,
+      "loss": 1.1639,
+      "step": 2100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001786738006279491,
+      "loss": 1.2322,
+      "step": 2101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001786541565359765,
+      "loss": 1.2059,
+      "step": 2102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017863450448176552,
+      "loss": 1.2365,
+      "step": 2103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017861484446730563,
+      "loss": 1.2339,
+      "step": 2104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017859517649458696,
+      "loss": 1.2578,
+      "step": 2105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017857550056560054,
+      "loss": 1.2089,
+      "step": 2106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017855581668233824,
+      "loss": 1.2067,
+      "step": 2107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017853612484679257,
+      "loss": 1.2523,
+      "step": 2108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017851642506095704,
+      "loss": 1.2356,
+      "step": 2109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017849671732682586,
+      "loss": 1.1865,
+      "step": 2110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017847700164639406,
+      "loss": 1.1805,
+      "step": 2111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017845727802165746,
+      "loss": 1.1461,
+      "step": 2112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017843754645461272,
+      "loss": 1.1539,
+      "step": 2113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001784178069472573,
+      "loss": 1.1716,
+      "step": 2114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001783980595015894,
+      "loss": 1.2757,
+      "step": 2115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001783783041196081,
+      "loss": 1.2062,
+      "step": 2116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017835854080331328,
+      "loss": 1.1758,
+      "step": 2117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017833876955470558,
+      "loss": 1.2303,
+      "step": 2118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001783189903757865,
+      "loss": 1.2245,
+      "step": 2119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017829920326855824,
+      "loss": 1.221,
+      "step": 2120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001782794082350239,
+      "loss": 1.1992,
+      "step": 2121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017825960527718734,
+      "loss": 1.2755,
+      "step": 2122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017823979439705328,
+      "loss": 1.241,
+      "step": 2123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017821997559662713,
+      "loss": 1.1971,
+      "step": 2124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001782001488779152,
+      "loss": 1.1926,
+      "step": 2125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017818031424292457,
+      "loss": 1.1919,
+      "step": 2126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017816047169366312,
+      "loss": 1.2546,
+      "step": 2127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017814062123213951,
+      "loss": 1.2226,
+      "step": 2128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017812076286036327,
+      "loss": 1.3043,
+      "step": 2129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001781008965803446,
+      "loss": 1.2969,
+      "step": 2130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001780810223940947,
+      "loss": 1.1971,
+      "step": 2131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017806114030362533,
+      "loss": 1.2147,
+      "step": 2132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017804125031094925,
+      "loss": 1.2398,
+      "step": 2133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017802135241807993,
+      "loss": 1.182,
+      "step": 2134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017800144662703164,
+      "loss": 1.2649,
+      "step": 2135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017798153293981946,
+      "loss": 1.2151,
+      "step": 2136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017796161135845933,
+      "loss": 1.2346,
+      "step": 2137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017794168188496782,
+      "loss": 1.1644,
+      "step": 2138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017792174452136252,
+      "loss": 1.2119,
+      "step": 2139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017790179926966166,
+      "loss": 1.1659,
+      "step": 2140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001778818461318843,
+      "loss": 1.2273,
+      "step": 2141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017786188511005038,
+      "loss": 1.2362,
+      "step": 2142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017784191620618048,
+      "loss": 1.2585,
+      "step": 2143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017782193942229616,
+      "loss": 1.3014,
+      "step": 2144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017780195476041962,
+      "loss": 1.178,
+      "step": 2145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017778196222257398,
+      "loss": 1.298,
+      "step": 2146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017776196181078308,
+      "loss": 1.2462,
+      "step": 2147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017774195352707162,
+      "loss": 1.2048,
+      "step": 2148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000177721937373465,
+      "loss": 1.2907,
+      "step": 2149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001777019133519895,
+      "loss": 1.2375,
+      "step": 2150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017768188146467215,
+      "loss": 1.1804,
+      "step": 2151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017766184171354085,
+      "loss": 1.1892,
+      "step": 2152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001776417941006242,
+      "loss": 1.2203,
+      "step": 2153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017762173862795165,
+      "loss": 1.2191,
+      "step": 2154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017760167529755344,
+      "loss": 1.2244,
+      "step": 2155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001775816041114606,
+      "loss": 1.1804,
+      "step": 2156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017756152507170496,
+      "loss": 1.2344,
+      "step": 2157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017754143818031915,
+      "loss": 1.2061,
+      "step": 2158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017752134343933658,
+      "loss": 1.2395,
+      "step": 2159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001775012408507914,
+      "loss": 1.2535,
+      "step": 2160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017748113041671872,
+      "loss": 1.2068,
+      "step": 2161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017746101213915428,
+      "loss": 1.2188,
+      "step": 2162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017744088602013466,
+      "loss": 1.1918,
+      "step": 2163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017742075206169725,
+      "loss": 1.1741,
+      "step": 2164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017740061026588029,
+      "loss": 1.3031,
+      "step": 2165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017738046063472268,
+      "loss": 1.254,
+      "step": 2166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017736030317026425,
+      "loss": 1.2764,
+      "step": 2167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001773401378745455,
+      "loss": 1.2284,
+      "step": 2168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017731996474960787,
+      "loss": 1.2594,
+      "step": 2169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017729978379749336,
+      "loss": 1.2335,
+      "step": 2170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017727959502024504,
+      "loss": 1.1933,
+      "step": 2171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001772593984199066,
+      "loss": 1.2152,
+      "step": 2172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017723919399852258,
+      "loss": 1.2141,
+      "step": 2173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017721898175813826,
+      "loss": 1.2166,
+      "step": 2174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017719876170079973,
+      "loss": 1.2319,
+      "step": 2175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017717853382855393,
+      "loss": 1.1994,
+      "step": 2176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017715829814344853,
+      "loss": 1.2895,
+      "step": 2177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000177138054647532,
+      "loss": 1.2168,
+      "step": 2178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017711780334285363,
+      "loss": 1.3022,
+      "step": 2179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017709754423146345,
+      "loss": 1.2829,
+      "step": 2180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017707727731541235,
+      "loss": 1.2443,
+      "step": 2181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017705700259675194,
+      "loss": 1.1784,
+      "step": 2182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017703672007753464,
+      "loss": 1.1908,
+      "step": 2183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017701642975981373,
+      "loss": 1.2156,
+      "step": 2184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017699613164564313,
+      "loss": 1.2551,
+      "step": 2185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001769758257370777,
+      "loss": 1.2433,
+      "step": 2186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017695551203617303,
+      "loss": 1.1937,
+      "step": 2187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017693519054498547,
+      "loss": 1.2548,
+      "step": 2188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017691486126557217,
+      "loss": 1.1629,
+      "step": 2189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017689452419999112,
+      "loss": 1.2149,
+      "step": 2190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017687417935030104,
+      "loss": 1.1828,
+      "step": 2191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017685382671856148,
+      "loss": 1.2538,
+      "step": 2192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017683346630683274,
+      "loss": 1.259,
+      "step": 2193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017681309811717593,
+      "loss": 1.1739,
+      "step": 2194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017679272215165295,
+      "loss": 1.2594,
+      "step": 2195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017677233841232646,
+      "loss": 1.1889,
+      "step": 2196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001767519469012599,
+      "loss": 1.2399,
+      "step": 2197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001767315476205176,
+      "loss": 1.2317,
+      "step": 2198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017671114057216455,
+      "loss": 1.2355,
+      "step": 2199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017669072575826657,
+      "loss": 1.2841,
+      "step": 2200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017667030318089033,
+      "loss": 1.1962,
+      "step": 2201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017664987284210314,
+      "loss": 1.2426,
+      "step": 2202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017662943474397325,
+      "loss": 1.2437,
+      "step": 2203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017660898888856958,
+      "loss": 1.2302,
+      "step": 2204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017658853527796194,
+      "loss": 1.1671,
+      "step": 2205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017656807391422082,
+      "loss": 1.25,
+      "step": 2206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017654760479941757,
+      "loss": 1.2469,
+      "step": 2207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001765271279356243,
+      "loss": 1.1753,
+      "step": 2208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001765066433249139,
+      "loss": 1.2349,
+      "step": 2209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017648615096936,
+      "loss": 1.2595,
+      "step": 2210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017646565087103716,
+      "loss": 1.2862,
+      "step": 2211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017644514303202054,
+      "loss": 1.1908,
+      "step": 2212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001764246274543862,
+      "loss": 1.2639,
+      "step": 2213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017640410414021095,
+      "loss": 1.2369,
+      "step": 2214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001763835730915724,
+      "loss": 1.1842,
+      "step": 2215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017636303431054891,
+      "loss": 1.2247,
+      "step": 2216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017634248779921965,
+      "loss": 1.2177,
+      "step": 2217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017632193355966453,
+      "loss": 1.1729,
+      "step": 2218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017630137159396433,
+      "loss": 1.2416,
+      "step": 2219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017628080190420053,
+      "loss": 1.1864,
+      "step": 2220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017626022449245544,
+      "loss": 1.2471,
+      "step": 2221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017623963936081211,
+      "loss": 1.2246,
+      "step": 2222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017621904651135437,
+      "loss": 1.2002,
+      "step": 2223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017619844594616693,
+      "loss": 1.1855,
+      "step": 2224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017617783766733514,
+      "loss": 1.2394,
+      "step": 2225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001761572216769452,
+      "loss": 1.192,
+      "step": 2226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001761365979770841,
+      "loss": 1.2235,
+      "step": 2227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017611596656983965,
+      "loss": 1.1924,
+      "step": 2228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017609532745730032,
+      "loss": 1.2254,
+      "step": 2229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001760746806415554,
+      "loss": 1.2024,
+      "step": 2230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017605402612469508,
+      "loss": 1.1955,
+      "step": 2231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017603336390881018,
+      "loss": 1.1866,
+      "step": 2232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001760126939959924,
+      "loss": 1.2357,
+      "step": 2233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017599201638833408,
+      "loss": 1.1964,
+      "step": 2234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017597133108792856,
+      "loss": 1.1688,
+      "step": 2235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017595063809686972,
+      "loss": 1.2207,
+      "step": 2236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001759299374172524,
+      "loss": 1.2142,
+      "step": 2237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017590922905117218,
+      "loss": 1.1931,
+      "step": 2238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017588851300072528,
+      "loss": 1.1684,
+      "step": 2239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001758677892680089,
+      "loss": 1.2337,
+      "step": 2240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001758470578551209,
+      "loss": 1.263,
+      "step": 2241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017582631876415992,
+      "loss": 1.2938,
+      "step": 2242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017580557199722542,
+      "loss": 1.278,
+      "step": 2243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017578481755641762,
+      "loss": 1.2296,
+      "step": 2244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017576405544383748,
+      "loss": 1.2863,
+      "step": 2245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017574328566158682,
+      "loss": 1.173,
+      "step": 2246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017572250821176812,
+      "loss": 1.2114,
+      "step": 2247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017570172309648477,
+      "loss": 1.251,
+      "step": 2248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017568093031784083,
+      "loss": 1.2104,
+      "step": 2249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017566012987794117,
+      "loss": 1.2579,
+      "step": 2250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017563932177889146,
+      "loss": 1.2576,
+      "step": 2251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017561850602279812,
+      "loss": 1.2075,
+      "step": 2252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017559768261176832,
+      "loss": 1.1809,
+      "step": 2253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017557685154791008,
+      "loss": 1.2426,
+      "step": 2254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017555601283333213,
+      "loss": 1.2599,
+      "step": 2255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017553516647014398,
+      "loss": 1.2024,
+      "step": 2256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017551431246045596,
+      "loss": 1.2461,
+      "step": 2257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017549345080637908,
+      "loss": 1.194,
+      "step": 2258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017547258151002525,
+      "loss": 1.1911,
+      "step": 2259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001754517045735071,
+      "loss": 1.2372,
+      "step": 2260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017543081999893794,
+      "loss": 1.2897,
+      "step": 2261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000175409927788432,
+      "loss": 1.1853,
+      "step": 2262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017538902794410422,
+      "loss": 1.2686,
+      "step": 2263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001753681204680703,
+      "loss": 1.216,
+      "step": 2264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001753472053624467,
+      "loss": 1.215,
+      "step": 2265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017532628262935073,
+      "loss": 1.22,
+      "step": 2266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017530535227090036,
+      "loss": 1.2153,
+      "step": 2267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017528441428921442,
+      "loss": 1.1926,
+      "step": 2268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001752634686864125,
+      "loss": 1.1434,
+      "step": 2269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001752425154646149,
+      "loss": 1.2092,
+      "step": 2270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001752215546259428,
+      "loss": 1.1871,
+      "step": 2271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017520058617251803,
+      "loss": 1.3012,
+      "step": 2272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017517961010646326,
+      "loss": 1.2113,
+      "step": 2273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017515862642990192,
+      "loss": 1.1778,
+      "step": 2274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017513763514495823,
+      "loss": 1.1934,
+      "step": 2275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017511663625375713,
+      "loss": 1.281,
+      "step": 2276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017509562975842434,
+      "loss": 1.1976,
+      "step": 2277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017507461566108645,
+      "loss": 1.181,
+      "step": 2278
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017505359396387067,
+      "loss": 1.245,
+      "step": 2279
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017503256466890502,
+      "loss": 1.2144,
+      "step": 2280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017501152777831843,
+      "loss": 1.2645,
+      "step": 2281
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017499048329424035,
+      "loss": 1.2261,
+      "step": 2282
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017496943121880122,
+      "loss": 1.2246,
+      "step": 2283
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017494837155413215,
+      "loss": 1.1509,
+      "step": 2284
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000174927304302365,
+      "loss": 1.1825,
+      "step": 2285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017490622946563246,
+      "loss": 1.2544,
+      "step": 2286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017488514704606798,
+      "loss": 1.2246,
+      "step": 2287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017486405704580569,
+      "loss": 1.2788,
+      "step": 2288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001748429594669806,
+      "loss": 1.2532,
+      "step": 2289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001748218543117284,
+      "loss": 1.2017,
+      "step": 2290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017480074158218563,
+      "loss": 1.1525,
+      "step": 2291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001747796212804895,
+      "loss": 1.2184,
+      "step": 2292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001747584934087781,
+      "loss": 1.276,
+      "step": 2293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017473735796919018,
+      "loss": 1.2203,
+      "step": 2294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001747162149638653,
+      "loss": 1.2206,
+      "step": 2295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017469506439494382,
+      "loss": 1.3057,
+      "step": 2296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017467390626456684,
+      "loss": 1.2675,
+      "step": 2297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017465274057487618,
+      "loss": 1.2206,
+      "step": 2298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017463156732801447,
+      "loss": 1.2089,
+      "step": 2299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017461038652612508,
+      "loss": 1.2177,
+      "step": 2300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017458919817135224,
+      "loss": 1.1678,
+      "step": 2301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017456800226584078,
+      "loss": 1.1705,
+      "step": 2302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017454679881173646,
+      "loss": 1.2053,
+      "step": 2303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017452558781118569,
+      "loss": 1.233,
+      "step": 2304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017450436926633563,
+      "loss": 1.2172,
+      "step": 2305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017448314317933434,
+      "loss": 1.2079,
+      "step": 2306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017446190955233053,
+      "loss": 1.2803,
+      "step": 2307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001744406683874737,
+      "loss": 1.2171,
+      "step": 2308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017441941968691406,
+      "loss": 1.2081,
+      "step": 2309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017439816345280273,
+      "loss": 1.2334,
+      "step": 2310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017437689968729142,
+      "loss": 1.2387,
+      "step": 2311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017435562839253273,
+      "loss": 1.2474,
+      "step": 2312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017433434957067997,
+      "loss": 1.2312,
+      "step": 2313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001743130632238872,
+      "loss": 1.2133,
+      "step": 2314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017429176935430927,
+      "loss": 1.1752,
+      "step": 2315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017427046796410176,
+      "loss": 1.2313,
+      "step": 2316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017424915905542106,
+      "loss": 1.1622,
+      "step": 2317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017422784263042428,
+      "loss": 1.1597,
+      "step": 2318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017420651869126926,
+      "loss": 1.2356,
+      "step": 2319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017418518724011476,
+      "loss": 1.2482,
+      "step": 2320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017416384827912004,
+      "loss": 1.2592,
+      "step": 2321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017414250181044537,
+      "loss": 1.2292,
+      "step": 2322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017412114783625158,
+      "loss": 1.2177,
+      "step": 2323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017409978635870045,
+      "loss": 1.2396,
+      "step": 2324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001740784173799544,
+      "loss": 1.169,
+      "step": 2325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001740570409021766,
+      "loss": 1.1993,
+      "step": 2326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017403565692753102,
+      "loss": 1.2114,
+      "step": 2327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017401426545818238,
+      "loss": 1.2537,
+      "step": 2328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001739928664962962,
+      "loss": 1.2809,
+      "step": 2329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017397146004403865,
+      "loss": 1.2176,
+      "step": 2330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001739500461035768,
+      "loss": 1.2334,
+      "step": 2331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017392862467707833,
+      "loss": 1.2659,
+      "step": 2332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017390719576671182,
+      "loss": 1.2667,
+      "step": 2333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001738857593746465,
+      "loss": 1.294,
+      "step": 2334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017386431550305242,
+      "loss": 1.2235,
+      "step": 2335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017384286415410034,
+      "loss": 1.2113,
+      "step": 2336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017382140532996182,
+      "loss": 1.1306,
+      "step": 2337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017379993903280915,
+      "loss": 1.1374,
+      "step": 2338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001737784652648154,
+      "loss": 1.1657,
+      "step": 2339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017375698402815432,
+      "loss": 1.1884,
+      "step": 2340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017373549532500056,
+      "loss": 1.2273,
+      "step": 2341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001737139991575294,
+      "loss": 1.3299,
+      "step": 2342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017369249552791693,
+      "loss": 1.2983,
+      "step": 2343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017367098443833996,
+      "loss": 1.2655,
+      "step": 2344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001736494658909761,
+      "loss": 1.246,
+      "step": 2345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017362793988800373,
+      "loss": 1.169,
+      "step": 2346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017360640643160187,
+      "loss": 1.166,
+      "step": 2347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001735848655239504,
+      "loss": 1.2501,
+      "step": 2348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017356331716722998,
+      "loss": 1.1425,
+      "step": 2349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017354176136362193,
+      "loss": 1.2883,
+      "step": 2350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017352019811530832,
+      "loss": 1.1834,
+      "step": 2351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017349862742447214,
+      "loss": 1.1964,
+      "step": 2352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017347704929329688,
+      "loss": 1.2123,
+      "step": 2353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017345546372396698,
+      "loss": 1.2481,
+      "step": 2354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001734338707186676,
+      "loss": 1.2384,
+      "step": 2355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001734122702795846,
+      "loss": 1.2003,
+      "step": 2356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017339066240890452,
+      "loss": 1.2417,
+      "step": 2357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017336904710881488,
+      "loss": 1.178,
+      "step": 2358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017334742438150373,
+      "loss": 1.1613,
+      "step": 2359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017332579422916003,
+      "loss": 1.1971,
+      "step": 2360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017330415665397334,
+      "loss": 1.1961,
+      "step": 2361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017328251165813415,
+      "loss": 1.2282,
+      "step": 2362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017326085924383353,
+      "loss": 1.2725,
+      "step": 2363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017323919941326338,
+      "loss": 1.1966,
+      "step": 2364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001732175321686164,
+      "loss": 1.1653,
+      "step": 2365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001731958575120859,
+      "loss": 1.1921,
+      "step": 2366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017317417544586615,
+      "loss": 1.1604,
+      "step": 2367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017315248597215194,
+      "loss": 1.2143,
+      "step": 2368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017313078909313894,
+      "loss": 1.2235,
+      "step": 2369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017310908481102356,
+      "loss": 1.2533,
+      "step": 2370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017308737312800293,
+      "loss": 1.2439,
+      "step": 2371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017306565404627498,
+      "loss": 1.1791,
+      "step": 2372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017304392756803833,
+      "loss": 1.2499,
+      "step": 2373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017302219369549237,
+      "loss": 1.1717,
+      "step": 2374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017300045243083726,
+      "loss": 1.2488,
+      "step": 2375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017297870377627387,
+      "loss": 1.223,
+      "step": 2376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001729569477340039,
+      "loss": 1.1901,
+      "step": 2377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001729351843062296,
+      "loss": 1.2381,
+      "step": 2378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017291341349515423,
+      "loss": 1.2319,
+      "step": 2379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001728916353029816,
+      "loss": 1.2678,
+      "step": 2380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017286984973191637,
+      "loss": 1.2424,
+      "step": 2381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017284805678416393,
+      "loss": 1.1156,
+      "step": 2382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001728262564619304,
+      "loss": 1.1919,
+      "step": 2383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017280444876742256,
+      "loss": 1.2388,
+      "step": 2384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017278263370284816,
+      "loss": 1.2138,
+      "step": 2385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017276081127041545,
+      "loss": 1.1873,
+      "step": 2386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001727389814723336,
+      "loss": 1.1947,
+      "step": 2387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017271714431081245,
+      "loss": 1.2207,
+      "step": 2388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001726952997880626,
+      "loss": 1.2265,
+      "step": 2389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017267344790629534,
+      "loss": 1.2293,
+      "step": 2390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017265158866772281,
+      "loss": 1.2255,
+      "step": 2391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017262972207455785,
+      "loss": 1.2476,
+      "step": 2392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017260784812901404,
+      "loss": 1.1659,
+      "step": 2393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001725859668333057,
+      "loss": 1.2379,
+      "step": 2394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001725640781896478,
+      "loss": 1.2007,
+      "step": 2395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017254218220025627,
+      "loss": 1.196,
+      "step": 2396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001725202788673476,
+      "loss": 1.1917,
+      "step": 2397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001724983681931391,
+      "loss": 1.2828,
+      "step": 2398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017247645017984885,
+      "loss": 1.1814,
+      "step": 2399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001724545248296956,
+      "loss": 1.185,
+      "step": 2400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017243259214489882,
+      "loss": 1.1803,
+      "step": 2401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017241065212767885,
+      "loss": 1.2349,
+      "step": 2402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001723887047802567,
+      "loss": 1.2775,
+      "step": 2403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017236675010485407,
+      "loss": 1.24,
+      "step": 2404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017234478810369353,
+      "loss": 1.2814,
+      "step": 2405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001723228187789982,
+      "loss": 1.2061,
+      "step": 2406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017230084213299218,
+      "loss": 1.179,
+      "step": 2407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017227885816790012,
+      "loss": 1.2119,
+      "step": 2408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001722568668859475,
+      "loss": 1.1871,
+      "step": 2409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017223486828936047,
+      "loss": 1.2155,
+      "step": 2410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017221286238036608,
+      "loss": 1.2602,
+      "step": 2411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017219084916119187,
+      "loss": 1.2818,
+      "step": 2412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001721688286340664,
+      "loss": 1.1943,
+      "step": 2413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001721468008012187,
+      "loss": 1.2424,
+      "step": 2414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017212476566487873,
+      "loss": 1.2289,
+      "step": 2415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017210272322727715,
+      "loss": 1.1703,
+      "step": 2416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017208067349064535,
+      "loss": 1.2454,
+      "step": 2417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017205861645721538,
+      "loss": 1.2306,
+      "step": 2418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017203655212922011,
+      "loss": 1.2496,
+      "step": 2419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017201448050889318,
+      "loss": 1.2164,
+      "step": 2420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001719924015984689,
+      "loss": 1.2034,
+      "step": 2421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017197031540018234,
+      "loss": 1.2338,
+      "step": 2422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017194822191626927,
+      "loss": 1.2055,
+      "step": 2423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001719261211489663,
+      "loss": 1.2775,
+      "step": 2424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017190401310051065,
+      "loss": 1.1656,
+      "step": 2425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001718818977731404,
+      "loss": 1.2033,
+      "step": 2426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017185977516909423,
+      "loss": 1.2669,
+      "step": 2427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017183764529061173,
+      "loss": 1.2705,
+      "step": 2428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001718155081399331,
+      "loss": 1.2364,
+      "step": 2429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017179336371929923,
+      "loss": 1.205,
+      "step": 2430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017177121203095188,
+      "loss": 1.2004,
+      "step": 2431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017174905307713352,
+      "loss": 1.2488,
+      "step": 2432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017172688686008724,
+      "loss": 1.2492,
+      "step": 2433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017170471338205702,
+      "loss": 1.2423,
+      "step": 2434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017168253264528748,
+      "loss": 1.1996,
+      "step": 2435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000171660344652024,
+      "loss": 1.1801,
+      "step": 2436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017163814940451269,
+      "loss": 1.2296,
+      "step": 2437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001716159469050004,
+      "loss": 1.1769,
+      "step": 2438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017159373715573467,
+      "loss": 1.212,
+      "step": 2439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001715715201589639,
+      "loss": 1.1924,
+      "step": 2440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017154929591693704,
+      "loss": 1.1534,
+      "step": 2441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017152706443190397,
+      "loss": 1.202,
+      "step": 2442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017150482570611514,
+      "loss": 1.2347,
+      "step": 2443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017148257974182183,
+      "loss": 1.1489,
+      "step": 2444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017146032654127597,
+      "loss": 1.2438,
+      "step": 2445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017143806610673035,
+      "loss": 1.2117,
+      "step": 2446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017141579844043837,
+      "loss": 1.1832,
+      "step": 2447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001713935235446542,
+      "loss": 1.2864,
+      "step": 2448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017137124142163276,
+      "loss": 1.122,
+      "step": 2449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017134895207362972,
+      "loss": 1.2046,
+      "step": 2450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017132665550290144,
+      "loss": 1.2398,
+      "step": 2451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017130435171170495,
+      "loss": 1.2292,
+      "step": 2452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001712820407022982,
+      "loss": 1.2437,
+      "step": 2453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001712597224769397,
+      "loss": 1.2179,
+      "step": 2454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017123739703788876,
+      "loss": 1.2156,
+      "step": 2455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017121506438740535,
+      "loss": 1.2261,
+      "step": 2456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001711927245277503,
+      "loss": 1.2238,
+      "step": 2457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017117037746118506,
+      "loss": 1.1439,
+      "step": 2458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001711480231899719,
+      "loss": 1.2298,
+      "step": 2459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017112566171637364,
+      "loss": 1.1751,
+      "step": 2460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001711032930426541,
+      "loss": 1.2702,
+      "step": 2461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001710809171710776,
+      "loss": 1.257,
+      "step": 2462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017105853410390927,
+      "loss": 1.1943,
+      "step": 2463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000171036143843415,
+      "loss": 1.2161,
+      "step": 2464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017101374639186137,
+      "loss": 1.2115,
+      "step": 2465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017099134175151568,
+      "loss": 1.2316,
+      "step": 2466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000170968929924646,
+      "loss": 1.1586,
+      "step": 2467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017094651091352107,
+      "loss": 1.2007,
+      "step": 2468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001709240847204104,
+      "loss": 1.1299,
+      "step": 2469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017090165134758426,
+      "loss": 1.2438,
+      "step": 2470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017087921079731355,
+      "loss": 1.2278,
+      "step": 2471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017085676307186998,
+      "loss": 1.2249,
+      "step": 2472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017083430817352592,
+      "loss": 1.2405,
+      "step": 2473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017081184610455452,
+      "loss": 1.2207,
+      "step": 2474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017078937686722964,
+      "loss": 1.2631,
+      "step": 2475
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.842420756816864,
+      "eval_runtime": 500.9384,
+      "eval_samples_per_second": 51.703,
+      "eval_steps_per_second": 25.851,
+      "step": 2475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017076690046382588,
+      "loss": 1.2195,
+      "step": 2476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001707444168966185,
+      "loss": 1.1923,
+      "step": 2477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017072192616788358,
+      "loss": 1.2348,
+      "step": 2478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001706994282798979,
+      "loss": 1.2324,
+      "step": 2479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017067692323493886,
+      "loss": 1.2439,
+      "step": 2480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017065441103528476,
+      "loss": 1.1957,
+      "step": 2481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001706318916832144,
+      "loss": 1.2102,
+      "step": 2482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017060936518100762,
+      "loss": 1.2433,
+      "step": 2483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017058683153094466,
+      "loss": 1.1872,
+      "step": 2484
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017056429073530665,
+      "loss": 1.1538,
+      "step": 2485
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017054174279637548,
+      "loss": 1.2063,
+      "step": 2486
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017051918771643365,
+      "loss": 1.1508,
+      "step": 2487
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001704966254977644,
+      "loss": 1.1863,
+      "step": 2488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001704740561426518,
+      "loss": 1.1999,
+      "step": 2489
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017045147965338053,
+      "loss": 1.235,
+      "step": 2490
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000170428896032236,
+      "loss": 1.2861,
+      "step": 2491
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017040630528150446,
+      "loss": 1.2223,
+      "step": 2492
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017038370740347268,
+      "loss": 1.2855,
+      "step": 2493
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017036110240042836,
+      "loss": 1.2298,
+      "step": 2494
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001703384902746598,
+      "loss": 1.2043,
+      "step": 2495
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017031587102845603,
+      "loss": 1.2872,
+      "step": 2496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702932446641068,
+      "loss": 1.2557,
+      "step": 2497
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017027061118390265,
+      "loss": 1.2129,
+      "step": 2498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001702479705901348,
+      "loss": 1.2664,
+      "step": 2499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001702253228850951,
+      "loss": 1.1449,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017020266807107626,
+      "loss": 1.1557,
+      "step": 2501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017018000615037166,
+      "loss": 1.1687,
+      "step": 2502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017015733712527533,
+      "loss": 1.1421,
+      "step": 2503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017013466099808215,
+      "loss": 1.1677,
+      "step": 2504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001701119777710876,
+      "loss": 1.2201,
+      "step": 2505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017008928744658788,
+      "loss": 1.1837,
+      "step": 2506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017006659002688008,
+      "loss": 1.1599,
+      "step": 2507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001700438855142618,
+      "loss": 1.2342,
+      "step": 2508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017002117391103143,
+      "loss": 1.1457,
+      "step": 2509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016999845521948812,
+      "loss": 1.2131,
+      "step": 2510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001699757294419317,
+      "loss": 1.1454,
+      "step": 2511
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016995299658066274,
+      "loss": 1.1056,
+      "step": 2512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016993025663798243,
+      "loss": 1.2301,
+      "step": 2513
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016990750961619287,
+      "loss": 1.1536,
+      "step": 2514
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016988475551759667,
+      "loss": 1.1705,
+      "step": 2515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016986199434449736,
+      "loss": 1.1575,
+      "step": 2516
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001698392260991989,
+      "loss": 1.1458,
+      "step": 2517
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016981645078400633,
+      "loss": 1.1552,
+      "step": 2518
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016979366840122513,
+      "loss": 1.1508,
+      "step": 2519
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016977087895316158,
+      "loss": 1.1428,
+      "step": 2520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016974808244212266,
+      "loss": 1.2239,
+      "step": 2521
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016972527887041614,
+      "loss": 1.1399,
+      "step": 2522
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001697024682403504,
+      "loss": 1.1422,
+      "step": 2523
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016967965055423461,
+      "loss": 1.1632,
+      "step": 2524
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001696568258143786,
+      "loss": 1.1752,
+      "step": 2525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000169633994023093,
+      "loss": 1.165,
+      "step": 2526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016961115518268903,
+      "loss": 1.1908,
+      "step": 2527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016958830929547872,
+      "loss": 1.1366,
+      "step": 2528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016956545636377478,
+      "loss": 1.1251,
+      "step": 2529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016954259638989064,
+      "loss": 1.1275,
+      "step": 2530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016951972937614038,
+      "loss": 1.1413,
+      "step": 2531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016949685532483894,
+      "loss": 1.1419,
+      "step": 2532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016947397423830183,
+      "loss": 1.1966,
+      "step": 2533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016945108611884533,
+      "loss": 1.2063,
+      "step": 2534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016942819096878647,
+      "loss": 1.1929,
+      "step": 2535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001694052887904429,
+      "loss": 1.1327,
+      "step": 2536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000169382379586133,
+      "loss": 1.1341,
+      "step": 2537
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016935946335817597,
+      "loss": 1.1937,
+      "step": 2538
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001693365401088916,
+      "loss": 1.2007,
+      "step": 2539
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016931360984060042,
+      "loss": 1.1786,
+      "step": 2540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016929067255562372,
+      "loss": 1.2284,
+      "step": 2541
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016926772825628345,
+      "loss": 1.2203,
+      "step": 2542
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016924477694490228,
+      "loss": 1.1771,
+      "step": 2543
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001692218186238036,
+      "loss": 1.2364,
+      "step": 2544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016919885329531149,
+      "loss": 1.1546,
+      "step": 2545
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001691758809617508,
+      "loss": 1.2519,
+      "step": 2546
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016915290162544696,
+      "loss": 1.2424,
+      "step": 2547
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016912991528872625,
+      "loss": 1.2069,
+      "step": 2548
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016910692195391559,
+      "loss": 1.1883,
+      "step": 2549
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016908392162334263,
+      "loss": 1.1415,
+      "step": 2550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016906091429933566,
+      "loss": 1.1167,
+      "step": 2551
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016903789998422378,
+      "loss": 1.2099,
+      "step": 2552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001690148786803368,
+      "loss": 1.1558,
+      "step": 2553
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016899185039000506,
+      "loss": 1.2219,
+      "step": 2554
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016896881511555987,
+      "loss": 1.203,
+      "step": 2555
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016894577285933305,
+      "loss": 1.1765,
+      "step": 2556
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016892272362365718,
+      "loss": 1.1609,
+      "step": 2557
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016889966741086556,
+      "loss": 1.1508,
+      "step": 2558
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016887660422329224,
+      "loss": 1.1086,
+      "step": 2559
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016885353406327186,
+      "loss": 1.2032,
+      "step": 2560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001688304569331399,
+      "loss": 1.1981,
+      "step": 2561
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016880737283523245,
+      "loss": 1.189,
+      "step": 2562
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016878428177188632,
+      "loss": 1.2007,
+      "step": 2563
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016876118374543908,
+      "loss": 1.1679,
+      "step": 2564
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016873807875822893,
+      "loss": 1.1856,
+      "step": 2565
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016871496681259484,
+      "loss": 1.2227,
+      "step": 2566
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016869184791087641,
+      "loss": 1.1569,
+      "step": 2567
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001686687220554141,
+      "loss": 1.1574,
+      "step": 2568
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001686455892485488,
+      "loss": 1.2013,
+      "step": 2569
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016862244949262238,
+      "loss": 1.173,
+      "step": 2570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001685993027899773,
+      "loss": 1.1875,
+      "step": 2571
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016857614914295665,
+      "loss": 1.1,
+      "step": 2572
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016855298855390438,
+      "loss": 1.1438,
+      "step": 2573
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016852982102516498,
+      "loss": 1.2113,
+      "step": 2574
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001685066465590838,
+      "loss": 1.1841,
+      "step": 2575
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016848346515800676,
+      "loss": 1.1533,
+      "step": 2576
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001684602768242806,
+      "loss": 1.152,
+      "step": 2577
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001684370815602526,
+      "loss": 1.1656,
+      "step": 2578
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001684138793682709,
+      "loss": 1.1997,
+      "step": 2579
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016839067025068428,
+      "loss": 1.15,
+      "step": 2580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001683674542098422,
+      "loss": 1.1892,
+      "step": 2581
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016834423124809485,
+      "loss": 1.2654,
+      "step": 2582
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016832100136779317,
+      "loss": 1.1562,
+      "step": 2583
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016829776457128866,
+      "loss": 1.1769,
+      "step": 2584
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016827452086093364,
+      "loss": 1.2221,
+      "step": 2585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001682512702390811,
+      "loss": 1.1294,
+      "step": 2586
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016822801270808472,
+      "loss": 1.2191,
+      "step": 2587
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016820474827029886,
+      "loss": 1.1498,
+      "step": 2588
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016818147692807865,
+      "loss": 1.1794,
+      "step": 2589
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016815819868377984,
+      "loss": 1.1794,
+      "step": 2590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016813491353975888,
+      "loss": 1.1521,
+      "step": 2591
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016811162149837298,
+      "loss": 1.0951,
+      "step": 2592
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016808832256198005,
+      "loss": 1.204,
+      "step": 2593
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016806501673293864,
+      "loss": 1.1348,
+      "step": 2594
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016804170401360798,
+      "loss": 1.2047,
+      "step": 2595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001680183844063481,
+      "loss": 1.2251,
+      "step": 2596
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016799505791351962,
+      "loss": 1.1302,
+      "step": 2597
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016797172453748393,
+      "loss": 1.1267,
+      "step": 2598
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016794838428060306,
+      "loss": 1.1799,
+      "step": 2599
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016792503714523985,
+      "loss": 1.2168,
+      "step": 2600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016790168313375766,
+      "loss": 1.1624,
+      "step": 2601
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016787832224852067,
+      "loss": 1.1444,
+      "step": 2602
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016785495449189375,
+      "loss": 1.1753,
+      "step": 2603
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016783157986624244,
+      "loss": 1.1947,
+      "step": 2604
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016780819837393292,
+      "loss": 1.1634,
+      "step": 2605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016778481001733219,
+      "loss": 1.1864,
+      "step": 2606
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001677614147988078,
+      "loss": 1.1168,
+      "step": 2607
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016773801272072816,
+      "loss": 1.1908,
+      "step": 2608
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016771460378546223,
+      "loss": 1.2009,
+      "step": 2609
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016769118799537974,
+      "loss": 1.1377,
+      "step": 2610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016766776535285112,
+      "loss": 1.1558,
+      "step": 2611
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001676443358602474,
+      "loss": 1.1384,
+      "step": 2612
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016762089951994038,
+      "loss": 1.1303,
+      "step": 2613
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001675974563343026,
+      "loss": 1.2012,
+      "step": 2614
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001675740063057072,
+      "loss": 1.1599,
+      "step": 2615
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016755054943652805,
+      "loss": 1.1514,
+      "step": 2616
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016752708572913977,
+      "loss": 1.1797,
+      "step": 2617
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001675036151859175,
+      "loss": 1.2112,
+      "step": 2618
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001674801378092373,
+      "loss": 1.1941,
+      "step": 2619
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016745665360147572,
+      "loss": 1.1424,
+      "step": 2620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016743316256501013,
+      "loss": 1.199,
+      "step": 2621
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016740966470221857,
+      "loss": 1.1622,
+      "step": 2622
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016738616001547973,
+      "loss": 1.2288,
+      "step": 2623
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000167362648507173,
+      "loss": 1.1613,
+      "step": 2624
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001673391301796785,
+      "loss": 1.222,
+      "step": 2625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000167315605035377,
+      "loss": 1.2288,
+      "step": 2626
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016729207307664994,
+      "loss": 1.1984,
+      "step": 2627
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016726853430587953,
+      "loss": 1.1634,
+      "step": 2628
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016724498872544861,
+      "loss": 1.2228,
+      "step": 2629
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016722143633774076,
+      "loss": 1.2088,
+      "step": 2630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016719787714514014,
+      "loss": 1.1706,
+      "step": 2631
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016717431115003173,
+      "loss": 1.2101,
+      "step": 2632
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016715073835480107,
+      "loss": 1.1443,
+      "step": 2633
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016712715876183453,
+      "loss": 1.1906,
+      "step": 2634
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016710357237351907,
+      "loss": 1.1627,
+      "step": 2635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016707997919224234,
+      "loss": 1.1425,
+      "step": 2636
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016705637922039275,
+      "loss": 1.1488,
+      "step": 2637
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001670327724603593,
+      "loss": 1.1719,
+      "step": 2638
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016700915891453176,
+      "loss": 1.2172,
+      "step": 2639
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016698553858530055,
+      "loss": 1.1912,
+      "step": 2640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016696191147505673,
+      "loss": 1.1892,
+      "step": 2641
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016693827758619217,
+      "loss": 1.157,
+      "step": 2642
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001669146369210993,
+      "loss": 1.1831,
+      "step": 2643
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001668909894821713,
+      "loss": 1.225,
+      "step": 2644
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016686733527180206,
+      "loss": 1.1513,
+      "step": 2645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016684367429238605,
+      "loss": 1.2179,
+      "step": 2646
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016682000654631853,
+      "loss": 1.169,
+      "step": 2647
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016679633203599544,
+      "loss": 1.2207,
+      "step": 2648
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001667726507638133,
+      "loss": 1.1067,
+      "step": 2649
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001667489627321695,
+      "loss": 1.1347,
+      "step": 2650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016672526794346188,
+      "loss": 1.2052,
+      "step": 2651
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016670156640008916,
+      "loss": 1.2,
+      "step": 2652
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016667785810445067,
+      "loss": 1.2088,
+      "step": 2653
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016665414305894638,
+      "loss": 1.1637,
+      "step": 2654
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016663042126597704,
+      "loss": 1.2288,
+      "step": 2655
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000166606692727944,
+      "loss": 1.1974,
+      "step": 2656
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016658295744724933,
+      "loss": 1.2216,
+      "step": 2657
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016655921542629577,
+      "loss": 1.1633,
+      "step": 2658
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016653546666748676,
+      "loss": 1.2068,
+      "step": 2659
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001665117111732264,
+      "loss": 1.1878,
+      "step": 2660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001664879489459195,
+      "loss": 1.1896,
+      "step": 2661
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001664641799879715,
+      "loss": 1.206,
+      "step": 2662
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001664404043017886,
+      "loss": 1.171,
+      "step": 2663
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001664166218897776,
+      "loss": 1.1656,
+      "step": 2664
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016639283275434602,
+      "loss": 1.2267,
+      "step": 2665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001663690368979021,
+      "loss": 1.1341,
+      "step": 2666
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001663452343228547,
+      "loss": 1.1999,
+      "step": 2667
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001663214250316133,
+      "loss": 1.1532,
+      "step": 2668
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016629760902658823,
+      "loss": 1.2024,
+      "step": 2669
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001662737863101904,
+      "loss": 1.269,
+      "step": 2670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001662499568848314,
+      "loss": 1.1762,
+      "step": 2671
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016622612075292342,
+      "loss": 1.2509,
+      "step": 2672
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016620227791687954,
+      "loss": 1.2193,
+      "step": 2673
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016617842837911336,
+      "loss": 1.1803,
+      "step": 2674
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016615457214203913,
+      "loss": 1.2149,
+      "step": 2675
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016613070920807192,
+      "loss": 1.1973,
+      "step": 2676
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016610683957962735,
+      "loss": 1.1922,
+      "step": 2677
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001660829632591218,
+      "loss": 1.1178,
+      "step": 2678
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016605908024897227,
+      "loss": 1.1857,
+      "step": 2679
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016603519055159645,
+      "loss": 1.1747,
+      "step": 2680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016601129416941274,
+      "loss": 1.1484,
+      "step": 2681
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016598739110484017,
+      "loss": 1.2182,
+      "step": 2682
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001659634813602985,
+      "loss": 1.1729,
+      "step": 2683
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016593956493820815,
+      "loss": 1.197,
+      "step": 2684
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016591564184099017,
+      "loss": 1.219,
+      "step": 2685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001658917120710663,
+      "loss": 1.106,
+      "step": 2686
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016586777563085904,
+      "loss": 1.1802,
+      "step": 2687
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016584383252279142,
+      "loss": 1.1932,
+      "step": 2688
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016581988274928732,
+      "loss": 1.2022,
+      "step": 2689
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001657959263127711,
+      "loss": 1.2147,
+      "step": 2690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016577196321566797,
+      "loss": 1.1293,
+      "step": 2691
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016574799346040367,
+      "loss": 1.2328,
+      "step": 2692
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016572401704940476,
+      "loss": 1.1628,
+      "step": 2693
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016570003398509833,
+      "loss": 1.1743,
+      "step": 2694
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016567604426991224,
+      "loss": 1.1749,
+      "step": 2695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000165652047906275,
+      "loss": 1.2063,
+      "step": 2696
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016562804489661574,
+      "loss": 1.1891,
+      "step": 2697
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016560403524336437,
+      "loss": 1.2253,
+      "step": 2698
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001655800189489514,
+      "loss": 1.1177,
+      "step": 2699
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016555599601580795,
+      "loss": 1.2135,
+      "step": 2700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000165531966446366,
+      "loss": 1.195,
+      "step": 2701
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000165507930243058,
+      "loss": 1.1465,
+      "step": 2702
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001654838874083172,
+      "loss": 1.1868,
+      "step": 2703
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016545983794457746,
+      "loss": 1.1826,
+      "step": 2704
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016543578185427332,
+      "loss": 1.1697,
+      "step": 2705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016541171913984006,
+      "loss": 1.2247,
+      "step": 2706
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016538764980371352,
+      "loss": 1.1907,
+      "step": 2707
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016536357384833026,
+      "loss": 1.1678,
+      "step": 2708
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016533949127612752,
+      "loss": 1.25,
+      "step": 2709
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016531540208954325,
+      "loss": 1.1608,
+      "step": 2710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016529130629101596,
+      "loss": 1.1945,
+      "step": 2711
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001652672038829849,
+      "loss": 1.202,
+      "step": 2712
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016524309486789005,
+      "loss": 1.1559,
+      "step": 2713
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016521897924817188,
+      "loss": 1.2268,
+      "step": 2714
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016519485702627172,
+      "loss": 1.168,
+      "step": 2715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016517072820463143,
+      "loss": 1.2092,
+      "step": 2716
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016514659278569365,
+      "loss": 1.1671,
+      "step": 2717
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016512245077190163,
+      "loss": 1.2426,
+      "step": 2718
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016509830216569922,
+      "loss": 1.147,
+      "step": 2719
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001650741469695311,
+      "loss": 1.1922,
+      "step": 2720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016504998518584248,
+      "loss": 1.1988,
+      "step": 2721
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016502581681707924,
+      "loss": 1.1496,
+      "step": 2722
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016500164186568803,
+      "loss": 1.1845,
+      "step": 2723
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016497746033411606,
+      "loss": 1.1558,
+      "step": 2724
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001649532722248113,
+      "loss": 1.1668,
+      "step": 2725
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001649290775402223,
+      "loss": 1.1965,
+      "step": 2726
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001649048762827983,
+      "loss": 1.212,
+      "step": 2727
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016488066845498928,
+      "loss": 1.1818,
+      "step": 2728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016485645405924578,
+      "loss": 1.2008,
+      "step": 2729
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016483223309801903,
+      "loss": 1.2022,
+      "step": 2730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016480800557376095,
+      "loss": 1.2337,
+      "step": 2731
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016478377148892412,
+      "loss": 1.2083,
+      "step": 2732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016475953084596183,
+      "loss": 1.2422,
+      "step": 2733
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001647352836473279,
+      "loss": 1.2452,
+      "step": 2734
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016471102989547697,
+      "loss": 1.1764,
+      "step": 2735
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016468676959286422,
+      "loss": 1.2088,
+      "step": 2736
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016466250274194556,
+      "loss": 1.1314,
+      "step": 2737
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016463822934517758,
+      "loss": 1.1658,
+      "step": 2738
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016461394940501744,
+      "loss": 1.179,
+      "step": 2739
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001645896629239231,
+      "loss": 1.1753,
+      "step": 2740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016456536990435303,
+      "loss": 1.249,
+      "step": 2741
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016454107034876648,
+      "loss": 1.1965,
+      "step": 2742
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016451676425962327,
+      "loss": 1.1872,
+      "step": 2743
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016449245163938398,
+      "loss": 1.1569,
+      "step": 2744
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001644681324905098,
+      "loss": 1.2149,
+      "step": 2745
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016444380681546257,
+      "loss": 1.1593,
+      "step": 2746
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016441947461670478,
+      "loss": 1.1559,
+      "step": 2747
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016439513589669964,
+      "loss": 1.2667,
+      "step": 2748
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016437079065791097,
+      "loss": 1.2041,
+      "step": 2749
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016434643890280324,
+      "loss": 1.2038,
+      "step": 2750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001643220806338416,
+      "loss": 1.1581,
+      "step": 2751
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001642977158534919,
+      "loss": 1.2036,
+      "step": 2752
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001642733445642206,
+      "loss": 1.1687,
+      "step": 2753
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016424896676849483,
+      "loss": 1.1345,
+      "step": 2754
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016422458246878233,
+      "loss": 1.1882,
+      "step": 2755
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001642001916675516,
+      "loss": 1.1816,
+      "step": 2756
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016417579436727174,
+      "loss": 1.1845,
+      "step": 2757
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001641513905704125,
+      "loss": 1.2033,
+      "step": 2758
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016412698027944432,
+      "loss": 1.2055,
+      "step": 2759
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016410256349683826,
+      "loss": 1.2155,
+      "step": 2760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016407814022506605,
+      "loss": 1.147,
+      "step": 2761
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001640537104666001,
+      "loss": 1.1859,
+      "step": 2762
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016402927422391346,
+      "loss": 1.1805,
+      "step": 2763
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001640048314994798,
+      "loss": 1.1945,
+      "step": 2764
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001639803822957735,
+      "loss": 1.17,
+      "step": 2765
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016395592661526962,
+      "loss": 1.2228,
+      "step": 2766
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016393146446044376,
+      "loss": 1.1875,
+      "step": 2767
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001639069958337723,
+      "loss": 1.2165,
+      "step": 2768
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001638825207377322,
+      "loss": 1.2094,
+      "step": 2769
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016385803917480111,
+      "loss": 1.2485,
+      "step": 2770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016383355114745732,
+      "loss": 1.1219,
+      "step": 2771
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016380905665817975,
+      "loss": 1.1343,
+      "step": 2772
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016378455570944807,
+      "loss": 1.2388,
+      "step": 2773
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016376004830374248,
+      "loss": 1.1173,
+      "step": 2774
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001637355344435439,
+      "loss": 1.1513,
+      "step": 2775
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016371101413133388,
+      "loss": 1.1417,
+      "step": 2776
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016368648736959465,
+      "loss": 1.1615,
+      "step": 2777
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001636619541608091,
+      "loss": 1.1262,
+      "step": 2778
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001636374145074607,
+      "loss": 1.0915,
+      "step": 2779
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016361286841203368,
+      "loss": 1.1866,
+      "step": 2780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016358831587701283,
+      "loss": 1.1861,
+      "step": 2781
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001635637569048836,
+      "loss": 1.1642,
+      "step": 2782
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016353919149813222,
+      "loss": 1.1946,
+      "step": 2783
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016351461965924538,
+      "loss": 1.2476,
+      "step": 2784
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016349004139071052,
+      "loss": 1.2286,
+      "step": 2785
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016346545669501574,
+      "loss": 1.1534,
+      "step": 2786
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016344086557464976,
+      "loss": 1.1318,
+      "step": 2787
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000163416268032102,
+      "loss": 1.1652,
+      "step": 2788
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016339166406986248,
+      "loss": 1.1665,
+      "step": 2789
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016336705369042186,
+      "loss": 1.2052,
+      "step": 2790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001633424368962715,
+      "loss": 1.1909,
+      "step": 2791
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001633178136899033,
+      "loss": 1.2181,
+      "step": 2792
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016329318407381005,
+      "loss": 1.2175,
+      "step": 2793
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001632685480504849,
+      "loss": 1.1808,
+      "step": 2794
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016324390562242184,
+      "loss": 1.1529,
+      "step": 2795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016321925679211541,
+      "loss": 1.2036,
+      "step": 2796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016319460156206085,
+      "loss": 1.1568,
+      "step": 2797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016316993993475406,
+      "loss": 1.2056,
+      "step": 2798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016314527191269148,
+      "loss": 1.1489,
+      "step": 2799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001631205974983704,
+      "loss": 1.187,
+      "step": 2800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016309591669428858,
+      "loss": 1.108,
+      "step": 2801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001630712295029444,
+      "loss": 1.1964,
+      "step": 2802
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016304653592683709,
+      "loss": 1.1514,
+      "step": 2803
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016302183596846636,
+      "loss": 1.1612,
+      "step": 2804
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016299712963033263,
+      "loss": 1.1407,
+      "step": 2805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016297241691493687,
+      "loss": 1.1364,
+      "step": 2806
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016294769782478085,
+      "loss": 1.1878,
+      "step": 2807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016292297236236692,
+      "loss": 1.197,
+      "step": 2808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016289824053019797,
+      "loss": 1.1419,
+      "step": 2809
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016287350233077773,
+      "loss": 1.2023,
+      "step": 2810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016284875776661044,
+      "loss": 1.1659,
+      "step": 2811
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016282400684020098,
+      "loss": 1.1557,
+      "step": 2812
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016279924955405495,
+      "loss": 1.186,
+      "step": 2813
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016277448591067855,
+      "loss": 1.1906,
+      "step": 2814
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001627497159125786,
+      "loss": 1.0967,
+      "step": 2815
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016272493956226262,
+      "loss": 1.1841,
+      "step": 2816
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016270015686223874,
+      "loss": 1.1686,
+      "step": 2817
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016267536781501573,
+      "loss": 1.1615,
+      "step": 2818
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016265057242310303,
+      "loss": 1.1503,
+      "step": 2819
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001626257706890107,
+      "loss": 1.1679,
+      "step": 2820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001626009626152494,
+      "loss": 1.2229,
+      "step": 2821
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001625761482043305,
+      "loss": 1.2218,
+      "step": 2822
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000162551327458766,
+      "loss": 1.1895,
+      "step": 2823
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016252650038106853,
+      "loss": 1.1851,
+      "step": 2824
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016250166697375138,
+      "loss": 1.1984,
+      "step": 2825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001624768272393284,
+      "loss": 1.1599,
+      "step": 2826
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016245198118031417,
+      "loss": 1.144,
+      "step": 2827
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016242712879922387,
+      "loss": 1.1821,
+      "step": 2828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001624022700985734,
+      "loss": 1.2132,
+      "step": 2829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016237740508087914,
+      "loss": 1.1915,
+      "step": 2830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016235253374865827,
+      "loss": 1.215,
+      "step": 2831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016232765610442847,
+      "loss": 1.1675,
+      "step": 2832
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016230277215070817,
+      "loss": 1.1297,
+      "step": 2833
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001622778818900164,
+      "loss": 1.1794,
+      "step": 2834
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016225298532487282,
+      "loss": 1.1638,
+      "step": 2835
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016222808245779774,
+      "loss": 1.1708,
+      "step": 2836
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001622031732913121,
+      "loss": 1.183,
+      "step": 2837
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016217825782793748,
+      "loss": 1.1687,
+      "step": 2838
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016215333607019606,
+      "loss": 1.2159,
+      "step": 2839
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016212840802061072,
+      "loss": 1.1597,
+      "step": 2840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000162103473681705,
+      "loss": 1.1957,
+      "step": 2841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016207853305600295,
+      "loss": 1.2105,
+      "step": 2842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001620535861460294,
+      "loss": 1.1636,
+      "step": 2843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001620286329543097,
+      "loss": 1.1508,
+      "step": 2844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001620036734833699,
+      "loss": 1.1917,
+      "step": 2845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001619787077357367,
+      "loss": 1.2154,
+      "step": 2846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016195373571393736,
+      "loss": 1.2077,
+      "step": 2847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016192875742049987,
+      "loss": 1.2134,
+      "step": 2848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016190377285795276,
+      "loss": 1.1666,
+      "step": 2849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001618787820288253,
+      "loss": 1.1883,
+      "step": 2850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016185378493564732,
+      "loss": 1.1972,
+      "step": 2851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016182878158094925,
+      "loss": 1.2729,
+      "step": 2852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016180377196726222,
+      "loss": 1.1862,
+      "step": 2853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000161778756097118,
+      "loss": 1.2132,
+      "step": 2854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000161753733973049,
+      "loss": 1.2296,
+      "step": 2855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001617287055975882,
+      "loss": 1.2067,
+      "step": 2856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016170367097326923,
+      "loss": 1.1201,
+      "step": 2857
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016167863010262644,
+      "loss": 1.1789,
+      "step": 2858
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016165358298819466,
+      "loss": 1.1819,
+      "step": 2859
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016162852963250946,
+      "loss": 1.2032,
+      "step": 2860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000161603470038107,
+      "loss": 1.1853,
+      "step": 2861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016157840420752414,
+      "loss": 1.1888,
+      "step": 2862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016155333214329831,
+      "loss": 1.1845,
+      "step": 2863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016152825384796752,
+      "loss": 1.2449,
+      "step": 2864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016150316932407051,
+      "loss": 1.1916,
+      "step": 2865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016147807857414662,
+      "loss": 1.1592,
+      "step": 2866
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016145298160073583,
+      "loss": 1.1844,
+      "step": 2867
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016142787840637868,
+      "loss": 1.1808,
+      "step": 2868
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016140276899361642,
+      "loss": 1.2322,
+      "step": 2869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001613776533649909,
+      "loss": 1.1146,
+      "step": 2870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001613525315230446,
+      "loss": 1.1584,
+      "step": 2871
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001613274034703206,
+      "loss": 1.1,
+      "step": 2872
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001613022692093627,
+      "loss": 1.1651,
+      "step": 2873
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001612771287427152,
+      "loss": 1.1569,
+      "step": 2874
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016125198207292312,
+      "loss": 1.1223,
+      "step": 2875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001612268292025321,
+      "loss": 1.1335,
+      "step": 2876
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001612016701340884,
+      "loss": 1.1539,
+      "step": 2877
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016117650487013882,
+      "loss": 1.1943,
+      "step": 2878
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016115133341323095,
+      "loss": 1.2107,
+      "step": 2879
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001611261557659129,
+      "loss": 1.1335,
+      "step": 2880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001611009719307334,
+      "loss": 1.1804,
+      "step": 2881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016107578191024185,
+      "loss": 1.213,
+      "step": 2882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016105058570698827,
+      "loss": 1.162,
+      "step": 2883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016102538332352328,
+      "loss": 1.0941,
+      "step": 2884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016100017476239814,
+      "loss": 1.2245,
+      "step": 2885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001609749600261648,
+      "loss": 1.1728,
+      "step": 2886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016094973911737567,
+      "loss": 1.2135,
+      "step": 2887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016092451203858396,
+      "loss": 1.2,
+      "step": 2888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016089927879234342,
+      "loss": 1.1711,
+      "step": 2889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016087403938120845,
+      "loss": 1.1702,
+      "step": 2890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000160848793807734,
+      "loss": 1.1936,
+      "step": 2891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001608235420744758,
+      "loss": 1.1019,
+      "step": 2892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016079828418398998,
+      "loss": 1.1543,
+      "step": 2893
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016077302013883354,
+      "loss": 1.1146,
+      "step": 2894
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001607477499415639,
+      "loss": 1.1954,
+      "step": 2895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001607224735947393,
+      "loss": 1.225,
+      "step": 2896
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016069719110091833,
+      "loss": 1.1374,
+      "step": 2897
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016067190246266052,
+      "loss": 1.2381,
+      "step": 2898
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016064660768252574,
+      "loss": 1.1903,
+      "step": 2899
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016062130676307467,
+      "loss": 1.2113,
+      "step": 2900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016059599970686853,
+      "loss": 1.1718,
+      "step": 2901
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016057068651646918,
+      "loss": 1.1737,
+      "step": 2902
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001605453671944391,
+      "loss": 1.1586,
+      "step": 2903
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001605200417433414,
+      "loss": 1.1359,
+      "step": 2904
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001604947101657398,
+      "loss": 1.1831,
+      "step": 2905
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016046937246419861,
+      "loss": 1.1608,
+      "step": 2906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016044402864128285,
+      "loss": 1.1625,
+      "step": 2907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016041867869955804,
+      "loss": 1.1724,
+      "step": 2908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016039332264159039,
+      "loss": 1.1702,
+      "step": 2909
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016036796046994676,
+      "loss": 1.1641,
+      "step": 2910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016034259218719455,
+      "loss": 1.2057,
+      "step": 2911
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016031721779590183,
+      "loss": 1.2048,
+      "step": 2912
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016029183729863728,
+      "loss": 1.2215,
+      "step": 2913
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016026645069797018,
+      "loss": 1.202,
+      "step": 2914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016024105799647043,
+      "loss": 1.1694,
+      "step": 2915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016021565919670857,
+      "loss": 1.129,
+      "step": 2916
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016019025430125575,
+      "loss": 1.1898,
+      "step": 2917
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016016484331268376,
+      "loss": 1.1486,
+      "step": 2918
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016013942623356492,
+      "loss": 1.223,
+      "step": 2919
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016011400306647228,
+      "loss": 1.2136,
+      "step": 2920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016008857381397945,
+      "loss": 1.2407,
+      "step": 2921
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016006313847866062,
+      "loss": 1.1367,
+      "step": 2922
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016003769706309062,
+      "loss": 1.15,
+      "step": 2923
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016001224956984497,
+      "loss": 1.1591,
+      "step": 2924
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015998679600149975,
+      "loss": 1.1645,
+      "step": 2925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015996133636063156,
+      "loss": 1.1841,
+      "step": 2926
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015993587064981778,
+      "loss": 1.1995,
+      "step": 2927
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015991039887163635,
+      "loss": 1.2202,
+      "step": 2928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015988492102866572,
+      "loss": 1.1452,
+      "step": 2929
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001598594371234851,
+      "loss": 1.1756,
+      "step": 2930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015983394715867423,
+      "loss": 1.2215,
+      "step": 2931
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001598084511368135,
+      "loss": 1.1708,
+      "step": 2932
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015978294906048387,
+      "loss": 1.1255,
+      "step": 2933
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000159757440932267,
+      "loss": 1.1501,
+      "step": 2934
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000159731926754745,
+      "loss": 1.161,
+      "step": 2935
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015970640653050077,
+      "loss": 1.1798,
+      "step": 2936
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015968088026211774,
+      "loss": 1.1637,
+      "step": 2937
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015965534795217996,
+      "loss": 1.1322,
+      "step": 2938
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015962980960327206,
+      "loss": 1.1913,
+      "step": 2939
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015960426521797935,
+      "loss": 1.1306,
+      "step": 2940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015957871479888773,
+      "loss": 1.111,
+      "step": 2941
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015955315834858362,
+      "loss": 1.1764,
+      "step": 2942
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015952759586965418,
+      "loss": 1.1118,
+      "step": 2943
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001595020273646871,
+      "loss": 1.1528,
+      "step": 2944
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015947645283627076,
+      "loss": 1.1493,
+      "step": 2945
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015945087228699402,
+      "loss": 1.2227,
+      "step": 2946
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015942528571944646,
+      "loss": 1.1994,
+      "step": 2947
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015939969313621822,
+      "loss": 1.2091,
+      "step": 2948
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001593740945399001,
+      "loss": 1.2497,
+      "step": 2949
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015934848993308344,
+      "loss": 1.2296,
+      "step": 2950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015932287931836022,
+      "loss": 1.1595,
+      "step": 2951
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015929726269832303,
+      "loss": 1.2081,
+      "step": 2952
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001592716400755651,
+      "loss": 1.1803,
+      "step": 2953
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015924601145268015,
+      "loss": 1.1818,
+      "step": 2954
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015922037683226267,
+      "loss": 1.2159,
+      "step": 2955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015919473621690765,
+      "loss": 1.1684,
+      "step": 2956
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001591690896092107,
+      "loss": 1.1687,
+      "step": 2957
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001591434370117681,
+      "loss": 1.2372,
+      "step": 2958
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015911777842717663,
+      "loss": 1.2731,
+      "step": 2959
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001590921138580338,
+      "loss": 1.2296,
+      "step": 2960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001590664433069376,
+      "loss": 1.1775,
+      "step": 2961
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015904076677648665,
+      "loss": 1.1698,
+      "step": 2962
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001590150842692803,
+      "loss": 1.1069,
+      "step": 2963
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001589893957879184,
+      "loss": 1.1226,
+      "step": 2964
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015896370133500142,
+      "loss": 1.2212,
+      "step": 2965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015893800091313035,
+      "loss": 1.1128,
+      "step": 2966
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000158912294524907,
+      "loss": 1.2022,
+      "step": 2967
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015888658217293357,
+      "loss": 1.1465,
+      "step": 2968
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015886086385981295,
+      "loss": 1.1232,
+      "step": 2969
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015883513958814865,
+      "loss": 1.1303,
+      "step": 2970
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 0.8377401232719421,
+      "eval_runtime": 503.6304,
+      "eval_samples_per_second": 51.427,
+      "eval_steps_per_second": 25.713,
+      "step": 2970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015880940936054477,
+      "loss": 1.1802,
+      "step": 2971
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000158783673179606,
+      "loss": 1.1688,
+      "step": 2972
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015875793104793762,
+      "loss": 1.1531,
+      "step": 2973
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015873218296814557,
+      "loss": 1.2327,
+      "step": 2974
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015870642894283635,
+      "loss": 1.107,
+      "step": 2975
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015868066897461699,
+      "loss": 1.1862,
+      "step": 2976
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015865490306609528,
+      "loss": 1.1659,
+      "step": 2977
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001586291312198795,
+      "loss": 1.1569,
+      "step": 2978
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015860335343857856,
+      "loss": 1.1107,
+      "step": 2979
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015857756972480197,
+      "loss": 1.2467,
+      "step": 2980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015855178008115987,
+      "loss": 1.0988,
+      "step": 2981
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001585259845102629,
+      "loss": 1.1931,
+      "step": 2982
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015850018301472244,
+      "loss": 1.1981,
+      "step": 2983
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015847437559715036,
+      "loss": 1.1553,
+      "step": 2984
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001584485622601592,
+      "loss": 1.1924,
+      "step": 2985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015842274300636208,
+      "loss": 1.1841,
+      "step": 2986
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015839691783837268,
+      "loss": 1.1943,
+      "step": 2987
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015837108675880528,
+      "loss": 1.134,
+      "step": 2988
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015834524977027483,
+      "loss": 1.2281,
+      "step": 2989
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015831940687539685,
+      "loss": 1.1658,
+      "step": 2990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001582935580767874,
+      "loss": 1.1689,
+      "step": 2991
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015826770337706316,
+      "loss": 1.1684,
+      "step": 2992
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001582418427788415,
+      "loss": 1.1839,
+      "step": 2993
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015821597628474024,
+      "loss": 1.197,
+      "step": 2994
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015819010389737795,
+      "loss": 1.1291,
+      "step": 2995
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015816422561937366,
+      "loss": 1.1496,
+      "step": 2996
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015813834145334704,
+      "loss": 1.1678,
+      "step": 2997
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015811245140191841,
+      "loss": 1.2627,
+      "step": 2998
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015808655546770866,
+      "loss": 1.1535,
+      "step": 2999
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001580606536533392,
+      "loss": 1.1035,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015803474596143215,
+      "loss": 1.1667,
+      "step": 3001
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015800883239461014,
+      "loss": 1.1714,
+      "step": 3002
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015798291295549647,
+      "loss": 1.1668,
+      "step": 3003
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001579569876467149,
+      "loss": 1.2369,
+      "step": 3004
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015793105647088997,
+      "loss": 1.1953,
+      "step": 3005
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015790511943064667,
+      "loss": 1.1933,
+      "step": 3006
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015787917652861067,
+      "loss": 1.176,
+      "step": 3007
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015785322776740813,
+      "loss": 1.183,
+      "step": 3008
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015782727314966595,
+      "loss": 1.1059,
+      "step": 3009
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015780131267801147,
+      "loss": 1.2543,
+      "step": 3010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015777534635507274,
+      "loss": 1.1676,
+      "step": 3011
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001577493741834783,
+      "loss": 1.2112,
+      "step": 3012
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015772339616585744,
+      "loss": 1.189,
+      "step": 3013
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015769741230483986,
+      "loss": 1.176,
+      "step": 3014
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015767142260305595,
+      "loss": 1.2109,
+      "step": 3015
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015764542706313668,
+      "loss": 1.1584,
+      "step": 3016
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015761942568771358,
+      "loss": 1.2118,
+      "step": 3017
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015759341847941882,
+      "loss": 1.2071,
+      "step": 3018
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015756740544088513,
+      "loss": 1.2267,
+      "step": 3019
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015754138657474585,
+      "loss": 1.1518,
+      "step": 3020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015751536188363487,
+      "loss": 1.184,
+      "step": 3021
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015748933137018674,
+      "loss": 1.1831,
+      "step": 3022
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001574632950370365,
+      "loss": 1.2405,
+      "step": 3023
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015743725288681985,
+      "loss": 1.2535,
+      "step": 3024
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015741120492217308,
+      "loss": 1.1601,
+      "step": 3025
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000157385151145733,
+      "loss": 1.1352,
+      "step": 3026
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015735909156013714,
+      "loss": 1.1426,
+      "step": 3027
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015733302616802348,
+      "loss": 1.1898,
+      "step": 3028
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015730695497203067,
+      "loss": 1.1101,
+      "step": 3029
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015728087797479794,
+      "loss": 1.183,
+      "step": 3030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015725479517896506,
+      "loss": 1.1457,
+      "step": 3031
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001572287065871724,
+      "loss": 1.1566,
+      "step": 3032
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015720261220206094,
+      "loss": 1.225,
+      "step": 3033
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001571765120262723,
+      "loss": 1.234,
+      "step": 3034
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015715040606244856,
+      "loss": 1.1963,
+      "step": 3035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015712429431323244,
+      "loss": 1.2262,
+      "step": 3036
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015709817678126738,
+      "loss": 1.242,
+      "step": 3037
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015707205346919716,
+      "loss": 1.2419,
+      "step": 3038
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001570459243796663,
+      "loss": 1.1496,
+      "step": 3039
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015701978951531987,
+      "loss": 1.1333,
+      "step": 3040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015699364887880357,
+      "loss": 1.1738,
+      "step": 3041
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001569675024727636,
+      "loss": 1.1493,
+      "step": 3042
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015694135029984682,
+      "loss": 1.1937,
+      "step": 3043
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015691519236270057,
+      "loss": 1.181,
+      "step": 3044
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015688902866397295,
+      "loss": 1.1756,
+      "step": 3045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015686285920631245,
+      "loss": 1.1326,
+      "step": 3046
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001568366839923683,
+      "loss": 1.1757,
+      "step": 3047
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015681050302479015,
+      "loss": 1.2108,
+      "step": 3048
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001567843163062284,
+      "loss": 1.2385,
+      "step": 3049
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015675812383933395,
+      "loss": 1.2049,
+      "step": 3050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015673192562675825,
+      "loss": 1.1636,
+      "step": 3051
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015670572167115343,
+      "loss": 1.2535,
+      "step": 3052
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015667951197517206,
+      "loss": 1.172,
+      "step": 3053
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015665329654146744,
+      "loss": 1.2462,
+      "step": 3054
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015662707537269341,
+      "loss": 1.1654,
+      "step": 3055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001566008484715043,
+      "loss": 1.2247,
+      "step": 3056
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001565746158405551,
+      "loss": 1.1631,
+      "step": 3057
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001565483774825014,
+      "loss": 1.1733,
+      "step": 3058
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015652213339999928,
+      "loss": 1.179,
+      "step": 3059
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015649588359570546,
+      "loss": 1.1365,
+      "step": 3060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001564696280722773,
+      "loss": 1.1455,
+      "step": 3061
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015644336683237262,
+      "loss": 1.1667,
+      "step": 3062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015641709987864987,
+      "loss": 1.1947,
+      "step": 3063
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001563908272137681,
+      "loss": 1.1522,
+      "step": 3064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001563645488403869,
+      "loss": 1.2075,
+      "step": 3065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015633826476116647,
+      "loss": 1.2095,
+      "step": 3066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015631197497876755,
+      "loss": 1.1136,
+      "step": 3067
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015628567949585152,
+      "loss": 1.185,
+      "step": 3068
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015625937831508025,
+      "loss": 1.1986,
+      "step": 3069
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001562330714391163,
+      "loss": 1.1387,
+      "step": 3070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001562067588706227,
+      "loss": 1.2259,
+      "step": 3071
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015618044061226308,
+      "loss": 1.1376,
+      "step": 3072
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015615411666670167,
+      "loss": 1.1863,
+      "step": 3073
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001561277870366033,
+      "loss": 1.2025,
+      "step": 3074
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015610145172463333,
+      "loss": 1.1208,
+      "step": 3075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001560751107334577,
+      "loss": 1.1871,
+      "step": 3076
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015604876406574296,
+      "loss": 1.1658,
+      "step": 3077
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015602241172415617,
+      "loss": 1.1851,
+      "step": 3078
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015599605371136504,
+      "loss": 1.2183,
+      "step": 3079
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015596969003003777,
+      "loss": 1.1332,
+      "step": 3080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001559433206828432,
+      "loss": 1.2006,
+      "step": 3081
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015591694567245078,
+      "loss": 1.1334,
+      "step": 3082
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015589056500153045,
+      "loss": 1.1924,
+      "step": 3083
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001558641786727527,
+      "loss": 1.187,
+      "step": 3084
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001558377866887887,
+      "loss": 1.1674,
+      "step": 3085
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015581138905231012,
+      "loss": 1.184,
+      "step": 3086
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015578498576598922,
+      "loss": 1.1707,
+      "step": 3087
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015575857683249882,
+      "loss": 1.1962,
+      "step": 3088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015573216225451235,
+      "loss": 1.1309,
+      "step": 3089
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015570574203470377,
+      "loss": 1.2169,
+      "step": 3090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001556793161757476,
+      "loss": 1.125,
+      "step": 3091
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000155652884680319,
+      "loss": 1.2182,
+      "step": 3092
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015562644755109361,
+      "loss": 1.1431,
+      "step": 3093
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001556000047907477,
+      "loss": 1.2254,
+      "step": 3094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001555735564019581,
+      "loss": 1.2362,
+      "step": 3095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015554710238740226,
+      "loss": 1.1194,
+      "step": 3096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001555206427497581,
+      "loss": 1.1117,
+      "step": 3097
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001554941774917041,
+      "loss": 1.1871,
+      "step": 3098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015546770661591948,
+      "loss": 1.1928,
+      "step": 3099
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015544123012508384,
+      "loss": 1.194,
+      "step": 3100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015541474802187742,
+      "loss": 1.1595,
+      "step": 3101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015538826030898103,
+      "loss": 1.221,
+      "step": 3102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015536176698907608,
+      "loss": 1.1687,
+      "step": 3103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001553352680648445,
+      "loss": 1.2116,
+      "step": 3104
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001553087635389688,
+      "loss": 1.1293,
+      "step": 3105
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015528225341413204,
+      "loss": 1.1881,
+      "step": 3106
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015525573769301792,
+      "loss": 1.177,
+      "step": 3107
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001552292163783106,
+      "loss": 1.115,
+      "step": 3108
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015520268947269485,
+      "loss": 1.1249,
+      "step": 3109
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015517615697885608,
+      "loss": 1.1216,
+      "step": 3110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015514961889948016,
+      "loss": 1.1958,
+      "step": 3111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015512307523725353,
+      "loss": 1.1147,
+      "step": 3112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001550965259948633,
+      "loss": 1.2298,
+      "step": 3113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015506997117499703,
+      "loss": 1.2266,
+      "step": 3114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001550434107803429,
+      "loss": 1.1759,
+      "step": 3115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015501684481358967,
+      "loss": 1.1445,
+      "step": 3116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015499027327742663,
+      "loss": 1.1566,
+      "step": 3117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015496369617454362,
+      "loss": 1.145,
+      "step": 3118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001549371135076311,
+      "loss": 1.2037,
+      "step": 3119
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015491052527938002,
+      "loss": 1.1345,
+      "step": 3120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015488393149248194,
+      "loss": 1.1692,
+      "step": 3121
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015485733214962898,
+      "loss": 1.188,
+      "step": 3122
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015483072725351387,
+      "loss": 1.1641,
+      "step": 3123
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015480411680682978,
+      "loss": 1.1193,
+      "step": 3124
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015477750081227055,
+      "loss": 1.1213,
+      "step": 3125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001547508792725305,
+      "loss": 1.1018,
+      "step": 3126
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015472425219030465,
+      "loss": 1.1775,
+      "step": 3127
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015469761956828836,
+      "loss": 1.1887,
+      "step": 3128
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015467098140917776,
+      "loss": 1.1503,
+      "step": 3129
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015464433771566945,
+      "loss": 1.2214,
+      "step": 3130
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015461768849046057,
+      "loss": 1.1762,
+      "step": 3131
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015459103373624889,
+      "loss": 1.1482,
+      "step": 3132
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001545643734557326,
+      "loss": 1.1993,
+      "step": 3133
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015453770765161068,
+      "loss": 1.165,
+      "step": 3134
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015451103632658245,
+      "loss": 1.1627,
+      "step": 3135
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001544843594833479,
+      "loss": 1.1773,
+      "step": 3136
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015445767712460754,
+      "loss": 1.123,
+      "step": 3137
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015443098925306247,
+      "loss": 1.1844,
+      "step": 3138
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001544042958714143,
+      "loss": 1.2025,
+      "step": 3139
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015437759698236532,
+      "loss": 1.1756,
+      "step": 3140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015435089258861813,
+      "loss": 1.1661,
+      "step": 3141
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015432418269287616,
+      "loss": 1.187,
+      "step": 3142
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015429746729784323,
+      "loss": 1.1862,
+      "step": 3143
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001542707464062238,
+      "loss": 1.2041,
+      "step": 3144
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015424402002072281,
+      "loss": 1.2037,
+      "step": 3145
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015421728814404585,
+      "loss": 1.1459,
+      "step": 3146
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015419055077889898,
+      "loss": 1.188,
+      "step": 3147
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015416380792798885,
+      "loss": 1.2102,
+      "step": 3148
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015413705959402266,
+      "loss": 1.1306,
+      "step": 3149
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001541103057797082,
+      "loss": 1.1987,
+      "step": 3150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015408354648775377,
+      "loss": 1.1849,
+      "step": 3151
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001540567817208682,
+      "loss": 1.151,
+      "step": 3152
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000154030011481761,
+      "loss": 1.1374,
+      "step": 3153
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015400323577314205,
+      "loss": 1.2149,
+      "step": 3154
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000153976454597722,
+      "loss": 1.1783,
+      "step": 3155
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015394966795821179,
+      "loss": 1.1556,
+      "step": 3156
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015392287585732316,
+      "loss": 1.2503,
+      "step": 3157
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001538960782977683,
+      "loss": 1.1718,
+      "step": 3158
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001538692752822599,
+      "loss": 1.1412,
+      "step": 3159
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001538424668135113,
+      "loss": 1.2025,
+      "step": 3160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001538156528942363,
+      "loss": 1.1116,
+      "step": 3161
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001537888335271494,
+      "loss": 1.1852,
+      "step": 3162
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015376200871496543,
+      "loss": 1.2546,
+      "step": 3163
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015373517846039992,
+      "loss": 1.2304,
+      "step": 3164
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015370834276616898,
+      "loss": 1.1332,
+      "step": 3165
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001536815016349892,
+      "loss": 1.208,
+      "step": 3166
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001536546550695777,
+      "loss": 1.2142,
+      "step": 3167
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001536278030726522,
+      "loss": 1.1024,
+      "step": 3168
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015360094564693095,
+      "loss": 1.1913,
+      "step": 3169
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015357408279513275,
+      "loss": 1.1215,
+      "step": 3170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015354721451997695,
+      "loss": 1.2093,
+      "step": 3171
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015352034082418353,
+      "loss": 1.0985,
+      "step": 3172
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015349346171047282,
+      "loss": 1.1862,
+      "step": 3173
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015346657718156589,
+      "loss": 1.1987,
+      "step": 3174
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015343968724018428,
+      "loss": 1.1701,
+      "step": 3175
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015341279188905006,
+      "loss": 1.1756,
+      "step": 3176
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001533858911308859,
+      "loss": 1.217,
+      "step": 3177
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000153358984968415,
+      "loss": 1.2175,
+      "step": 3178
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015333207340436104,
+      "loss": 1.2257,
+      "step": 3179
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015330515644144835,
+      "loss": 1.1735,
+      "step": 3180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001532782340824018,
+      "loss": 1.1438,
+      "step": 3181
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015325130632994668,
+      "loss": 1.1935,
+      "step": 3182
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015322437318680897,
+      "loss": 1.1485,
+      "step": 3183
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001531974346557151,
+      "loss": 1.1987,
+      "step": 3184
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015317049073939213,
+      "loss": 1.1631,
+      "step": 3185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001531435414405676,
+      "loss": 1.1857,
+      "step": 3186
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001531165867619696,
+      "loss": 1.1452,
+      "step": 3187
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015308962670632678,
+      "loss": 1.2091,
+      "step": 3188
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015306266127636836,
+      "loss": 1.1677,
+      "step": 3189
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015303569047482406,
+      "loss": 1.2387,
+      "step": 3190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015300871430442415,
+      "loss": 1.244,
+      "step": 3191
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015298173276789948,
+      "loss": 1.2028,
+      "step": 3192
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001529547458679814,
+      "loss": 1.127,
+      "step": 3193
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015292775360740182,
+      "loss": 1.1572,
+      "step": 3194
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001529007559888932,
+      "loss": 1.1512,
+      "step": 3195
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015287375301518852,
+      "loss": 1.1805,
+      "step": 3196
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015284674468902132,
+      "loss": 1.1918,
+      "step": 3197
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001528197310131257,
+      "loss": 1.2037,
+      "step": 3198
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015279271199023634,
+      "loss": 1.2196,
+      "step": 3199
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015276568762308826,
+      "loss": 1.1583,
+      "step": 3200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015273865791441725,
+      "loss": 1.2453,
+      "step": 3201
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015271162286695957,
+      "loss": 1.1916,
+      "step": 3202
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015268458248345198,
+      "loss": 1.1775,
+      "step": 3203
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015265753676663174,
+      "loss": 1.1302,
+      "step": 3204
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015263048571923678,
+      "loss": 1.152,
+      "step": 3205
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015260342934400557,
+      "loss": 1.1431,
+      "step": 3206
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015257636764367695,
+      "loss": 1.1593,
+      "step": 3207
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015254930062099046,
+      "loss": 1.2033,
+      "step": 3208
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015252222827868606,
+      "loss": 1.1713,
+      "step": 3209
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001524951506195044,
+      "loss": 1.1709,
+      "step": 3210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015246806764618645,
+      "loss": 1.15,
+      "step": 3211
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015244097936147397,
+      "loss": 1.1925,
+      "step": 3212
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001524138857681091,
+      "loss": 1.1778,
+      "step": 3213
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015238678686883454,
+      "loss": 1.2152,
+      "step": 3214
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015235968266639352,
+      "loss": 1.1914,
+      "step": 3215
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015233257316352985,
+      "loss": 1.173,
+      "step": 3216
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015230545836298782,
+      "loss": 1.1802,
+      "step": 3217
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015227833826751233,
+      "loss": 1.1566,
+      "step": 3218
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015225121287984876,
+      "loss": 1.1911,
+      "step": 3219
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015222408220274302,
+      "loss": 1.1534,
+      "step": 3220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015219694623894162,
+      "loss": 1.1539,
+      "step": 3221
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015216980499119153,
+      "loss": 1.162,
+      "step": 3222
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001521426584622403,
+      "loss": 1.2015,
+      "step": 3223
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015211550665483595,
+      "loss": 1.1184,
+      "step": 3224
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015208834957172713,
+      "loss": 1.2109,
+      "step": 3225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015206118721566297,
+      "loss": 1.155,
+      "step": 3226
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015203401958939318,
+      "loss": 1.1301,
+      "step": 3227
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015200684669566792,
+      "loss": 1.1505,
+      "step": 3228
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015197966853723794,
+      "loss": 1.1908,
+      "step": 3229
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015195248511685447,
+      "loss": 1.1388,
+      "step": 3230
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001519252964372694,
+      "loss": 1.1607,
+      "step": 3231
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015189810250123499,
+      "loss": 1.1557,
+      "step": 3232
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015187090331150415,
+      "loss": 1.1151,
+      "step": 3233
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015184369887083027,
+      "loss": 1.2263,
+      "step": 3234
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015181648918196732,
+      "loss": 1.1867,
+      "step": 3235
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015178927424766968,
+      "loss": 1.1199,
+      "step": 3236
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015176205407069242,
+      "loss": 1.1959,
+      "step": 3237
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015173482865379106,
+      "loss": 1.1639,
+      "step": 3238
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015170759799972158,
+      "loss": 1.2065,
+      "step": 3239
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015168036211124064,
+      "loss": 1.129,
+      "step": 3240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015165312099110537,
+      "loss": 1.135,
+      "step": 3241
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015162587464207335,
+      "loss": 1.1263,
+      "step": 3242
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015159862306690278,
+      "loss": 1.1602,
+      "step": 3243
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001515713662683524,
+      "loss": 1.1858,
+      "step": 3244
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015154410424918143,
+      "loss": 1.1392,
+      "step": 3245
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015151683701214954,
+      "loss": 1.1688,
+      "step": 3246
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015148956456001716,
+      "loss": 1.2049,
+      "step": 3247
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015146228689554502,
+      "loss": 1.1781,
+      "step": 3248
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001514350040214945,
+      "loss": 1.1709,
+      "step": 3249
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015140771594062747,
+      "loss": 1.1649,
+      "step": 3250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015138042265570628,
+      "loss": 1.1583,
+      "step": 3251
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015135312416949393,
+      "loss": 1.2,
+      "step": 3252
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015132582048475382,
+      "loss": 1.2359,
+      "step": 3253
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015129851160424997,
+      "loss": 1.1375,
+      "step": 3254
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015127119753074683,
+      "loss": 1.1859,
+      "step": 3255
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001512438782670095,
+      "loss": 1.1654,
+      "step": 3256
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001512165538158035,
+      "loss": 1.2098,
+      "step": 3257
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015118922417989487,
+      "loss": 1.1525,
+      "step": 3258
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015116188936205026,
+      "loss": 1.1766,
+      "step": 3259
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015113454936503687,
+      "loss": 1.2023,
+      "step": 3260
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001511072041916222,
+      "loss": 1.2308,
+      "step": 3261
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015107985384457456,
+      "loss": 1.2602,
+      "step": 3262
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015105249832666258,
+      "loss": 1.1603,
+      "step": 3263
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015102513764065552,
+      "loss": 1.1773,
+      "step": 3264
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015099777178932312,
+      "loss": 1.1929,
+      "step": 3265
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015097040077543565,
+      "loss": 1.1417,
+      "step": 3266
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015094302460176394,
+      "loss": 1.1941,
+      "step": 3267
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001509156432710792,
+      "loss": 1.2076,
+      "step": 3268
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015088825678615344,
+      "loss": 1.1597,
+      "step": 3269
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015086086514975886,
+      "loss": 1.209,
+      "step": 3270
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015083346836466843,
+      "loss": 1.1393,
+      "step": 3271
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015080606643365553,
+      "loss": 1.1726,
+      "step": 3272
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015077865935949407,
+      "loss": 1.1471,
+      "step": 3273
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015075124714495854,
+      "loss": 1.1815,
+      "step": 3274
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015072382979282388,
+      "loss": 1.2294,
+      "step": 3275
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015069640730586554,
+      "loss": 1.1625,
+      "step": 3276
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001506689796868596,
+      "loss": 1.1866,
+      "step": 3277
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001506415469385825,
+      "loss": 1.0672,
+      "step": 3278
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015061410906381135,
+      "loss": 1.2526,
+      "step": 3279
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001505866660653237,
+      "loss": 1.1834,
+      "step": 3280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015055921794589764,
+      "loss": 1.1636,
+      "step": 3281
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001505317647083117,
+      "loss": 1.1602,
+      "step": 3282
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015050430635534507,
+      "loss": 1.16,
+      "step": 3283
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015047684288977737,
+      "loss": 1.1906,
+      "step": 3284
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015044937431438877,
+      "loss": 1.1805,
+      "step": 3285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015042190063195994,
+      "loss": 1.1611,
+      "step": 3286
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015039442184527202,
+      "loss": 1.131,
+      "step": 3287
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015036693795710674,
+      "loss": 1.1561,
+      "step": 3288
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015033944897024635,
+      "loss": 1.1953,
+      "step": 3289
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015031195488747357,
+      "loss": 1.1808,
+      "step": 3290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015028445571157163,
+      "loss": 1.133,
+      "step": 3291
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015025695144532433,
+      "loss": 1.2206,
+      "step": 3292
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015022944209151592,
+      "loss": 1.2069,
+      "step": 3293
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015020192765293126,
+      "loss": 1.1141,
+      "step": 3294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001501744081323556,
+      "loss": 1.1425,
+      "step": 3295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015014688353257478,
+      "loss": 1.119,
+      "step": 3296
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015011935385637515,
+      "loss": 1.1548,
+      "step": 3297
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015009181910654364,
+      "loss": 1.0645,
+      "step": 3298
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001500642792858675,
+      "loss": 1.2154,
+      "step": 3299
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015003673439713467,
+      "loss": 1.2334,
+      "step": 3300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015000918444313356,
+      "loss": 1.263,
+      "step": 3301
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014998162942665304,
+      "loss": 1.1811,
+      "step": 3302
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014995406935048255,
+      "loss": 1.0961,
+      "step": 3303
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014992650421741203,
+      "loss": 1.2394,
+      "step": 3304
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014989893403023194,
+      "loss": 1.252,
+      "step": 3305
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001498713587917332,
+      "loss": 1.2193,
+      "step": 3306
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014984377850470735,
+      "loss": 1.1627,
+      "step": 3307
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014981619317194623,
+      "loss": 1.1989,
+      "step": 3308
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014978860279624248,
+      "loss": 1.2065,
+      "step": 3309
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014976100738038902,
+      "loss": 1.2226,
+      "step": 3310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001497334069271794,
+      "loss": 1.1996,
+      "step": 3311
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014970580143940763,
+      "loss": 1.1451,
+      "step": 3312
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001496781909198682,
+      "loss": 1.1555,
+      "step": 3313
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001496505753713562,
+      "loss": 1.1704,
+      "step": 3314
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014962295479666716,
+      "loss": 1.1626,
+      "step": 3315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014959532919859718,
+      "loss": 1.1533,
+      "step": 3316
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014956769857994275,
+      "loss": 1.2034,
+      "step": 3317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014954006294350098,
+      "loss": 1.1949,
+      "step": 3318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014951242229206948,
+      "loss": 1.1341,
+      "step": 3319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001494847766284463,
+      "loss": 1.1211,
+      "step": 3320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014945712595543003,
+      "loss": 1.2202,
+      "step": 3321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014942947027581985,
+      "loss": 1.1767,
+      "step": 3322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001494018095924153,
+      "loss": 1.2091,
+      "step": 3323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014937414390801655,
+      "loss": 1.1641,
+      "step": 3324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014934647322542417,
+      "loss": 1.0891,
+      "step": 3325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014931879754743931,
+      "loss": 1.1474,
+      "step": 3326
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014929111687686362,
+      "loss": 1.1271,
+      "step": 3327
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001492634312164992,
+      "loss": 1.2557,
+      "step": 3328
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001492357405691488,
+      "loss": 1.225,
+      "step": 3329
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014920804493761544,
+      "loss": 1.199,
+      "step": 3330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014918034432470287,
+      "loss": 1.1424,
+      "step": 3331
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014915263873321524,
+      "loss": 1.1837,
+      "step": 3332
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014912492816595717,
+      "loss": 1.1942,
+      "step": 3333
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014909721262573383,
+      "loss": 1.1451,
+      "step": 3334
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001490694921153509,
+      "loss": 1.1212,
+      "step": 3335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014904176663761465,
+      "loss": 1.1553,
+      "step": 3336
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001490140361953316,
+      "loss": 1.1915,
+      "step": 3337
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014898630079130903,
+      "loss": 1.1098,
+      "step": 3338
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014895856042835464,
+      "loss": 1.1669,
+      "step": 3339
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014893081510927654,
+      "loss": 1.184,
+      "step": 3340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014890306483688346,
+      "loss": 1.1199,
+      "step": 3341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014887530961398452,
+      "loss": 1.1772,
+      "step": 3342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014884754944338951,
+      "loss": 1.1304,
+      "step": 3343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014881978432790862,
+      "loss": 1.2672,
+      "step": 3344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014879201427035245,
+      "loss": 1.1623,
+      "step": 3345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014876423927353224,
+      "loss": 1.13,
+      "step": 3346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014873645934025966,
+      "loss": 1.1946,
+      "step": 3347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014870867447334694,
+      "loss": 1.2276,
+      "step": 3348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014868088467560674,
+      "loss": 1.2109,
+      "step": 3349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014865308994985226,
+      "loss": 1.2082,
+      "step": 3350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014862529029889717,
+      "loss": 1.1005,
+      "step": 3351
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014859748572555568,
+      "loss": 1.1447,
+      "step": 3352
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014856967623264248,
+      "loss": 1.1673,
+      "step": 3353
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014854186182297271,
+      "loss": 1.1442,
+      "step": 3354
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014851404249936208,
+      "loss": 1.1357,
+      "step": 3355
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014848621826462675,
+      "loss": 1.1336,
+      "step": 3356
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014845838912158347,
+      "loss": 1.1424,
+      "step": 3357
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001484305550730493,
+      "loss": 1.1529,
+      "step": 3358
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014840271612184198,
+      "loss": 1.2203,
+      "step": 3359
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014837487227077964,
+      "loss": 1.2131,
+      "step": 3360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014834702352268096,
+      "loss": 1.1447,
+      "step": 3361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014831916988036512,
+      "loss": 1.1667,
+      "step": 3362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014829131134665171,
+      "loss": 1.176,
+      "step": 3363
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014826344792436092,
+      "loss": 1.1759,
+      "step": 3364
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001482355796163134,
+      "loss": 1.1016,
+      "step": 3365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014820770642533023,
+      "loss": 1.1452,
+      "step": 3366
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001481798283542331,
+      "loss": 1.1379,
+      "step": 3367
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014815194540584412,
+      "loss": 1.2044,
+      "step": 3368
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001481240575829859,
+      "loss": 1.1584,
+      "step": 3369
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014809616488848154,
+      "loss": 1.1566,
+      "step": 3370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001480682673251547,
+      "loss": 1.1711,
+      "step": 3371
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014804036489582937,
+      "loss": 1.16,
+      "step": 3372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014801245760333027,
+      "loss": 1.1719,
+      "step": 3373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014798454545048243,
+      "loss": 1.124,
+      "step": 3374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014795662844011137,
+      "loss": 1.1604,
+      "step": 3375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001479287065750432,
+      "loss": 1.1815,
+      "step": 3376
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014790077985810452,
+      "loss": 1.2108,
+      "step": 3377
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014787284829212235,
+      "loss": 1.1695,
+      "step": 3378
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014784491187992417,
+      "loss": 1.1592,
+      "step": 3379
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001478169706243381,
+      "loss": 1.1892,
+      "step": 3380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014778902452819258,
+      "loss": 1.1425,
+      "step": 3381
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001477610735943167,
+      "loss": 1.2075,
+      "step": 3382
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014773311782553992,
+      "loss": 1.1822,
+      "step": 3383
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014770515722469223,
+      "loss": 1.1257,
+      "step": 3384
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014767719179460413,
+      "loss": 1.1545,
+      "step": 3385
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014764922153810654,
+      "loss": 1.1607,
+      "step": 3386
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000147621246458031,
+      "loss": 1.1268,
+      "step": 3387
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014759326655720936,
+      "loss": 1.1186,
+      "step": 3388
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001475652818384741,
+      "loss": 1.1817,
+      "step": 3389
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014753729230465815,
+      "loss": 1.1921,
+      "step": 3390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001475092979585949,
+      "loss": 1.1463,
+      "step": 3391
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014748129880311827,
+      "loss": 1.2224,
+      "step": 3392
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014745329484106262,
+      "loss": 1.1284,
+      "step": 3393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014742528607526284,
+      "loss": 1.2072,
+      "step": 3394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014739727250855425,
+      "loss": 1.1566,
+      "step": 3395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014736925414377268,
+      "loss": 1.1557,
+      "step": 3396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001473412309837545,
+      "loss": 1.181,
+      "step": 3397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001473132030313365,
+      "loss": 1.1624,
+      "step": 3398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014728517028935603,
+      "loss": 1.1448,
+      "step": 3399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001472571327606508,
+      "loss": 1.1903,
+      "step": 3400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001472290904480591,
+      "loss": 1.1909,
+      "step": 3401
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001472010433544197,
+      "loss": 1.1885,
+      "step": 3402
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014717299148257185,
+      "loss": 1.2143,
+      "step": 3403
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001471449348353552,
+      "loss": 1.1537,
+      "step": 3404
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014711687341561,
+      "loss": 1.1663,
+      "step": 3405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014708880722617693,
+      "loss": 1.2102,
+      "step": 3406
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014706073626989715,
+      "loss": 1.0845,
+      "step": 3407
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014703266054961233,
+      "loss": 1.1155,
+      "step": 3408
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014700458006816457,
+      "loss": 1.1862,
+      "step": 3409
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014697649482839652,
+      "loss": 1.151,
+      "step": 3410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014694840483315126,
+      "loss": 1.1722,
+      "step": 3411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014692031008527238,
+      "loss": 1.2159,
+      "step": 3412
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014689221058760392,
+      "loss": 1.2032,
+      "step": 3413
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014686410634299043,
+      "loss": 1.1993,
+      "step": 3414
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014683599735427694,
+      "loss": 1.1939,
+      "step": 3415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014680788362430894,
+      "loss": 1.2208,
+      "step": 3416
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001467797651559324,
+      "loss": 1.2097,
+      "step": 3417
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001467516419519938,
+      "loss": 1.1293,
+      "step": 3418
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014672351401534003,
+      "loss": 1.1496,
+      "step": 3419
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014669538134881862,
+      "loss": 1.1807,
+      "step": 3420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014666724395527735,
+      "loss": 1.2759,
+      "step": 3421
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014663910183756467,
+      "loss": 1.1498,
+      "step": 3422
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001466109549985294,
+      "loss": 1.1283,
+      "step": 3423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014658280344102085,
+      "loss": 1.1736,
+      "step": 3424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001465546471678889,
+      "loss": 1.2114,
+      "step": 3425
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014652648618198382,
+      "loss": 1.1631,
+      "step": 3426
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001464983204861563,
+      "loss": 1.082,
+      "step": 3427
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001464701500832577,
+      "loss": 1.1694,
+      "step": 3428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014644197497613962,
+      "loss": 1.1839,
+      "step": 3429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014641379516765435,
+      "loss": 1.1856,
+      "step": 3430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014638561066065444,
+      "loss": 1.2269,
+      "step": 3431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014635742145799321,
+      "loss": 1.1898,
+      "step": 3432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001463292275625241,
+      "loss": 1.1861,
+      "step": 3433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014630102897710134,
+      "loss": 1.11,
+      "step": 3434
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014627282570457944,
+      "loss": 1.1994,
+      "step": 3435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014624461774781345,
+      "loss": 1.1434,
+      "step": 3436
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014621640510965894,
+      "loss": 1.1824,
+      "step": 3437
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001461881877929718,
+      "loss": 1.2349,
+      "step": 3438
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014615996580060852,
+      "loss": 1.2142,
+      "step": 3439
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014613173913542615,
+      "loss": 1.1785,
+      "step": 3440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014610350780028204,
+      "loss": 1.1697,
+      "step": 3441
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014607527179803402,
+      "loss": 1.1941,
+      "step": 3442
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001460470311315405,
+      "loss": 1.1055,
+      "step": 3443
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001460187858036603,
+      "loss": 1.2143,
+      "step": 3444
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014599053581725277,
+      "loss": 1.1069,
+      "step": 3445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014596228117517762,
+      "loss": 1.1765,
+      "step": 3446
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001459340218802951,
+      "loss": 1.1675,
+      "step": 3447
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014590575793546596,
+      "loss": 1.1502,
+      "step": 3448
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014587748934355134,
+      "loss": 1.2222,
+      "step": 3449
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014584921610741298,
+      "loss": 1.2179,
+      "step": 3450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001458209382299129,
+      "loss": 1.2409,
+      "step": 3451
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014579265571391375,
+      "loss": 1.1525,
+      "step": 3452
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014576436856227867,
+      "loss": 1.1066,
+      "step": 3453
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014573607677787108,
+      "loss": 1.1806,
+      "step": 3454
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014570778036355504,
+      "loss": 1.1646,
+      "step": 3455
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014567947932219504,
+      "loss": 1.1544,
+      "step": 3456
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014565117365665598,
+      "loss": 1.1225,
+      "step": 3457
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014562286336980329,
+      "loss": 1.1592,
+      "step": 3458
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014559454846450282,
+      "loss": 1.1989,
+      "step": 3459
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014556622894362097,
+      "loss": 1.151,
+      "step": 3460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014553790481002454,
+      "loss": 1.216,
+      "step": 3461
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001455095760665808,
+      "loss": 1.129,
+      "step": 3462
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014548124271615746,
+      "loss": 1.1433,
+      "step": 3463
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001454529047616228,
+      "loss": 1.111,
+      "step": 3464
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014542456220584544,
+      "loss": 1.1455,
+      "step": 3465
+    },
+    {
+      "epoch": 1.39,
+      "eval_loss": 0.8314098715782166,
+      "eval_runtime": 501.6406,
+      "eval_samples_per_second": 51.631,
+      "eval_steps_per_second": 25.815,
+      "step": 3465
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014539621505169458,
+      "loss": 1.1702,
+      "step": 3466
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001453678633020398,
+      "loss": 1.1881,
+      "step": 3467
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014533950695975114,
+      "loss": 1.1977,
+      "step": 3468
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014531114602769923,
+      "loss": 1.1946,
+      "step": 3469
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014528278050875498,
+      "loss": 1.179,
+      "step": 3470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014525441040578996,
+      "loss": 1.1924,
+      "step": 3471
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000145226035721676,
+      "loss": 1.1752,
+      "step": 3472
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014519765645928554,
+      "loss": 1.2243,
+      "step": 3473
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014516927262149144,
+      "loss": 1.1788,
+      "step": 3474
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001451408842111671,
+      "loss": 1.1919,
+      "step": 3475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014511249123118618,
+      "loss": 1.1864,
+      "step": 3476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000145084093684423,
+      "loss": 1.1662,
+      "step": 3477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014505569157375226,
+      "loss": 1.1538,
+      "step": 3478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014502728490204913,
+      "loss": 1.2109,
+      "step": 3479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014499887367218922,
+      "loss": 1.2016,
+      "step": 3480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001449704578870487,
+      "loss": 1.1654,
+      "step": 3481
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014494203754950407,
+      "loss": 1.1486,
+      "step": 3482
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014491361266243238,
+      "loss": 1.2408,
+      "step": 3483
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014488518322871105,
+      "loss": 1.1518,
+      "step": 3484
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014485674925121807,
+      "loss": 1.1102,
+      "step": 3485
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014482831073283183,
+      "loss": 1.1612,
+      "step": 3486
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001447998676764312,
+      "loss": 1.2042,
+      "step": 3487
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001447714200848955,
+      "loss": 1.1422,
+      "step": 3488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001447429679611045,
+      "loss": 1.1388,
+      "step": 3489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001447145113079384,
+      "loss": 1.2237,
+      "step": 3490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014468605012827796,
+      "loss": 1.2026,
+      "step": 3491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014465758442500432,
+      "loss": 1.1693,
+      "step": 3492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014462911420099904,
+      "loss": 1.2028,
+      "step": 3493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001446006394591442,
+      "loss": 1.2248,
+      "step": 3494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001445721602023224,
+      "loss": 1.2063,
+      "step": 3495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014454367643341657,
+      "loss": 1.1196,
+      "step": 3496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014451518815531013,
+      "loss": 1.1756,
+      "step": 3497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014448669537088702,
+      "loss": 1.1735,
+      "step": 3498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014445819808303156,
+      "loss": 1.1655,
+      "step": 3499
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014442969629462855,
+      "loss": 1.2127,
+      "step": 3500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001444011900085633,
+      "loss": 1.2151,
+      "step": 3501
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014437267922772152,
+      "loss": 1.1422,
+      "step": 3502
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014434416395498934,
+      "loss": 1.1824,
+      "step": 3503
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014431564419325346,
+      "loss": 1.1964,
+      "step": 3504
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014428711994540088,
+      "loss": 1.1516,
+      "step": 3505
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001442585912143192,
+      "loss": 1.202,
+      "step": 3506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014423005800289637,
+      "loss": 1.1945,
+      "step": 3507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001442015203140209,
+      "loss": 1.0865,
+      "step": 3508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001441729781505816,
+      "loss": 1.2001,
+      "step": 3509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014414443151546794,
+      "loss": 1.151,
+      "step": 3510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014411588041156957,
+      "loss": 1.1372,
+      "step": 3511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001440873248417769,
+      "loss": 1.2145,
+      "step": 3512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014405876480898052,
+      "loss": 1.0934,
+      "step": 3513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014403020031607166,
+      "loss": 1.1747,
+      "step": 3514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001440016313659419,
+      "loss": 1.1969,
+      "step": 3515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001439730579614833,
+      "loss": 1.1771,
+      "step": 3516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014394448010558844,
+      "loss": 1.2282,
+      "step": 3517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001439158978011502,
+      "loss": 1.0674,
+      "step": 3518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000143887311051062,
+      "loss": 1.2132,
+      "step": 3519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014385871985821778,
+      "loss": 1.1583,
+      "step": 3520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014383012422551178,
+      "loss": 1.1984,
+      "step": 3521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014380152415583878,
+      "loss": 1.1768,
+      "step": 3522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014377291965209403,
+      "loss": 1.1788,
+      "step": 3523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014374431071717317,
+      "loss": 1.151,
+      "step": 3524
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001437156973539723,
+      "loss": 1.1923,
+      "step": 3525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014368707956538798,
+      "loss": 1.2472,
+      "step": 3526
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014365845735431725,
+      "loss": 1.1854,
+      "step": 3527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014362983072365747,
+      "loss": 1.1603,
+      "step": 3528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001436011996763067,
+      "loss": 1.1176,
+      "step": 3529
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014357256421516316,
+      "loss": 1.0801,
+      "step": 3530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001435439243431257,
+      "loss": 1.195,
+      "step": 3531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014351528006309352,
+      "loss": 1.1153,
+      "step": 3532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014348663137796636,
+      "loss": 1.2159,
+      "step": 3533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014345797829064434,
+      "loss": 1.1407,
+      "step": 3534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014342932080402802,
+      "loss": 1.1579,
+      "step": 3535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014340065892101843,
+      "loss": 1.1671,
+      "step": 3536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001433719926445171,
+      "loss": 1.2091,
+      "step": 3537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014334332197742585,
+      "loss": 1.2182,
+      "step": 3538
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014331464692264712,
+      "loss": 1.156,
+      "step": 3539
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014328596748308366,
+      "loss": 1.2179,
+      "step": 3540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001432572836616387,
+      "loss": 1.1128,
+      "step": 3541
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014322859546121603,
+      "loss": 1.1603,
+      "step": 3542
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001431999028847197,
+      "loss": 1.1189,
+      "step": 3543
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014317120593505434,
+      "loss": 1.1878,
+      "step": 3544
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001431425046151249,
+      "loss": 1.1581,
+      "step": 3545
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014311379892783694,
+      "loss": 1.1924,
+      "step": 3546
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014308508887609628,
+      "loss": 1.1781,
+      "step": 3547
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014305637446280928,
+      "loss": 1.2702,
+      "step": 3548
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014302765569088277,
+      "loss": 1.1512,
+      "step": 3549
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000142998932563224,
+      "loss": 1.1659,
+      "step": 3550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014297020508274055,
+      "loss": 1.1809,
+      "step": 3551
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014294147325234058,
+      "loss": 1.2117,
+      "step": 3552
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014291273707493264,
+      "loss": 1.2088,
+      "step": 3553
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014288399655342575,
+      "loss": 1.1271,
+      "step": 3554
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001428552516907293,
+      "loss": 1.1776,
+      "step": 3555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014282650248975318,
+      "loss": 1.1838,
+      "step": 3556
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014279774895340768,
+      "loss": 1.1744,
+      "step": 3557
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014276899108460359,
+      "loss": 1.2158,
+      "step": 3558
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014274022888625208,
+      "loss": 1.2126,
+      "step": 3559
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014271146236126474,
+      "loss": 1.0822,
+      "step": 3560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014268269151255367,
+      "loss": 1.1719,
+      "step": 3561
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014265391634303134,
+      "loss": 1.2346,
+      "step": 3562
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014262513685561076,
+      "loss": 1.1598,
+      "step": 3563
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014259635305320526,
+      "loss": 1.1557,
+      "step": 3564
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014256756493872863,
+      "loss": 1.1536,
+      "step": 3565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014253877251509513,
+      "loss": 1.1624,
+      "step": 3566
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014250997578521947,
+      "loss": 1.1327,
+      "step": 3567
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014248117475201678,
+      "loss": 1.2067,
+      "step": 3568
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014245236941840258,
+      "loss": 1.1925,
+      "step": 3569
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014242355978729286,
+      "loss": 1.1611,
+      "step": 3570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001423947458616041,
+      "loss": 1.1541,
+      "step": 3571
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001423659276442531,
+      "loss": 1.152,
+      "step": 3572
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001423371051381572,
+      "loss": 1.1558,
+      "step": 3573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001423082783462341,
+      "loss": 1.1315,
+      "step": 3574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000142279447271402,
+      "loss": 1.2285,
+      "step": 3575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014225061191657946,
+      "loss": 1.1731,
+      "step": 3576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014222177228468555,
+      "loss": 1.1839,
+      "step": 3577
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014219292837863972,
+      "loss": 1.1782,
+      "step": 3578
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014216408020136182,
+      "loss": 1.1152,
+      "step": 3579
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014213522775577227,
+      "loss": 1.179,
+      "step": 3580
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014210637104479176,
+      "loss": 1.2377,
+      "step": 3581
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014207751007134153,
+      "loss": 1.1523,
+      "step": 3582
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014204864483834316,
+      "loss": 1.2145,
+      "step": 3583
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014201977534871872,
+      "loss": 1.1468,
+      "step": 3584
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014199090160539073,
+      "loss": 1.201,
+      "step": 3585
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014196202361128207,
+      "loss": 1.1947,
+      "step": 3586
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001419331413693161,
+      "loss": 1.2071,
+      "step": 3587
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014190425488241662,
+      "loss": 1.1967,
+      "step": 3588
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001418753641535078,
+      "loss": 1.1252,
+      "step": 3589
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001418464691855143,
+      "loss": 1.2118,
+      "step": 3590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014181756998136118,
+      "loss": 1.1731,
+      "step": 3591
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014178866654397394,
+      "loss": 1.1996,
+      "step": 3592
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001417597588762785,
+      "loss": 1.1944,
+      "step": 3593
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001417308469812012,
+      "loss": 1.2025,
+      "step": 3594
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014170193086166882,
+      "loss": 1.1439,
+      "step": 3595
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001416730105206086,
+      "loss": 1.1955,
+      "step": 3596
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014164408596094815,
+      "loss": 1.1695,
+      "step": 3597
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014161515718561554,
+      "loss": 1.2042,
+      "step": 3598
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014158622419753925,
+      "loss": 1.1716,
+      "step": 3599
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001415572869996482,
+      "loss": 1.1393,
+      "step": 3600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014152834559487173,
+      "loss": 1.1772,
+      "step": 3601
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001414993999861396,
+      "loss": 1.1483,
+      "step": 3602
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000141470450176382,
+      "loss": 1.1872,
+      "step": 3603
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014144149616852956,
+      "loss": 1.1944,
+      "step": 3604
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014141253796551336,
+      "loss": 1.2261,
+      "step": 3605
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014138357557026481,
+      "loss": 1.1847,
+      "step": 3606
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001413546089857158,
+      "loss": 1.2544,
+      "step": 3607
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014132563821479867,
+      "loss": 1.1586,
+      "step": 3608
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014129666326044617,
+      "loss": 1.1623,
+      "step": 3609
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014126768412559145,
+      "loss": 1.2074,
+      "step": 3610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001412387008131681,
+      "loss": 1.2329,
+      "step": 3611
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014120971332611012,
+      "loss": 1.1672,
+      "step": 3612
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014118072166735196,
+      "loss": 1.1779,
+      "step": 3613
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014115172583982846,
+      "loss": 1.0947,
+      "step": 3614
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001411227258464749,
+      "loss": 1.1823,
+      "step": 3615
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014109372169022696,
+      "loss": 1.2216,
+      "step": 3616
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001410647133740208,
+      "loss": 1.1465,
+      "step": 3617
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001410357009007929,
+      "loss": 1.1527,
+      "step": 3618
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001410066842734803,
+      "loss": 1.1768,
+      "step": 3619
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001409776634950203,
+      "loss": 1.1269,
+      "step": 3620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001409486385683508,
+      "loss": 1.1576,
+      "step": 3621
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014091960949640995,
+      "loss": 1.1626,
+      "step": 3622
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014089057628213636,
+      "loss": 1.2162,
+      "step": 3623
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014086153892846917,
+      "loss": 1.1409,
+      "step": 3624
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014083249743834783,
+      "loss": 1.106,
+      "step": 3625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014080345181471224,
+      "loss": 1.1419,
+      "step": 3626
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001407744020605027,
+      "loss": 1.1454,
+      "step": 3627
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014074534817865998,
+      "loss": 1.1481,
+      "step": 3628
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001407162901721252,
+      "loss": 1.1873,
+      "step": 3629
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014068722804383996,
+      "loss": 1.1763,
+      "step": 3630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014065816179674622,
+      "loss": 1.188,
+      "step": 3631
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014062909143378638,
+      "loss": 1.1865,
+      "step": 3632
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001406000169579033,
+      "loss": 1.1722,
+      "step": 3633
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001405709383720402,
+      "loss": 1.1923,
+      "step": 3634
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014054185567914074,
+      "loss": 1.1793,
+      "step": 3635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014051276888214895,
+      "loss": 1.0902,
+      "step": 3636
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014048367798400938,
+      "loss": 1.222,
+      "step": 3637
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014045458298766686,
+      "loss": 1.2857,
+      "step": 3638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014042548389606676,
+      "loss": 1.2341,
+      "step": 3639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001403963807121548,
+      "loss": 1.214,
+      "step": 3640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001403672734388771,
+      "loss": 1.1667,
+      "step": 3641
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014033816207918023,
+      "loss": 1.157,
+      "step": 3642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014030904663601122,
+      "loss": 1.1898,
+      "step": 3643
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014027992711231733,
+      "loss": 1.1755,
+      "step": 3644
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014025080351104646,
+      "loss": 1.1636,
+      "step": 3645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001402216758351468,
+      "loss": 1.1474,
+      "step": 3646
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014019254408756697,
+      "loss": 1.1354,
+      "step": 3647
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014016340827125597,
+      "loss": 1.184,
+      "step": 3648
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014013426838916333,
+      "loss": 1.2204,
+      "step": 3649
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001401051244442388,
+      "loss": 1.1101,
+      "step": 3650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014007597643943273,
+      "loss": 1.2011,
+      "step": 3651
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014004682437769578,
+      "loss": 1.2072,
+      "step": 3652
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014001766826197905,
+      "loss": 1.1948,
+      "step": 3653
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013998850809523403,
+      "loss": 1.1826,
+      "step": 3654
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013995934388041264,
+      "loss": 1.1518,
+      "step": 3655
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013993017562046717,
+      "loss": 1.1877,
+      "step": 3656
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001399010033183504,
+      "loss": 1.2165,
+      "step": 3657
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013987182697701545,
+      "loss": 1.1478,
+      "step": 3658
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013984264659941588,
+      "loss": 1.2045,
+      "step": 3659
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013981346218850565,
+      "loss": 1.1448,
+      "step": 3660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001397842737472391,
+      "loss": 1.1587,
+      "step": 3661
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000139755081278571,
+      "loss": 1.1712,
+      "step": 3662
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013972588478545662,
+      "loss": 1.1463,
+      "step": 3663
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013969668427085146,
+      "loss": 1.1629,
+      "step": 3664
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013966747973771147,
+      "loss": 1.2893,
+      "step": 3665
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001396382711889932,
+      "loss": 1.2042,
+      "step": 3666
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013960905862765336,
+      "loss": 1.0654,
+      "step": 3667
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013957984205664923,
+      "loss": 1.2253,
+      "step": 3668
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013955062147893836,
+      "loss": 1.1844,
+      "step": 3669
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001395213968974788,
+      "loss": 1.1914,
+      "step": 3670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000139492168315229,
+      "loss": 1.1616,
+      "step": 3671
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013946293573514784,
+      "loss": 1.1565,
+      "step": 3672
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013943369916019448,
+      "loss": 1.2035,
+      "step": 3673
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001394044585933286,
+      "loss": 1.1158,
+      "step": 3674
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013937521403751028,
+      "loss": 1.2151,
+      "step": 3675
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013934596549569994,
+      "loss": 1.1034,
+      "step": 3676
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001393167129708585,
+      "loss": 1.1593,
+      "step": 3677
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001392874564659471,
+      "loss": 1.1887,
+      "step": 3678
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001392581959839275,
+      "loss": 1.1639,
+      "step": 3679
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013922893152776178,
+      "loss": 1.1463,
+      "step": 3680
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013919966310041234,
+      "loss": 1.1917,
+      "step": 3681
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001391703907048421,
+      "loss": 1.1674,
+      "step": 3682
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001391411143440143,
+      "loss": 1.2125,
+      "step": 3683
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013911183402089266,
+      "loss": 1.1373,
+      "step": 3684
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013908254973844127,
+      "loss": 1.1708,
+      "step": 3685
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001390532614996245,
+      "loss": 1.1581,
+      "step": 3686
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013902396930740734,
+      "loss": 1.1496,
+      "step": 3687
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000138994673164755,
+      "loss": 1.1839,
+      "step": 3688
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013896537307463323,
+      "loss": 1.1772,
+      "step": 3689
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013893606904000802,
+      "loss": 1.2098,
+      "step": 3690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001389067610638459,
+      "loss": 1.1506,
+      "step": 3691
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013887744914911374,
+      "loss": 1.1539,
+      "step": 3692
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013884813329877876,
+      "loss": 1.1298,
+      "step": 3693
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013881881351580875,
+      "loss": 1.1935,
+      "step": 3694
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013878948980317166,
+      "loss": 1.1749,
+      "step": 3695
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000138760162163836,
+      "loss": 1.1515,
+      "step": 3696
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013873083060077067,
+      "loss": 1.1728,
+      "step": 3697
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013870149511694492,
+      "loss": 1.202,
+      "step": 3698
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013867215571532836,
+      "loss": 1.1512,
+      "step": 3699
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013864281239889112,
+      "loss": 1.1987,
+      "step": 3700
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001386134651706036,
+      "loss": 1.1655,
+      "step": 3701
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001385841140334367,
+      "loss": 1.0855,
+      "step": 3702
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001385547589903616,
+      "loss": 1.1983,
+      "step": 3703
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013852540004434996,
+      "loss": 1.1469,
+      "step": 3704
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013849603719837385,
+      "loss": 1.1267,
+      "step": 3705
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013846667045540567,
+      "loss": 1.1948,
+      "step": 3706
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013843729981841824,
+      "loss": 1.1775,
+      "step": 3707
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001384079252903848,
+      "loss": 1.1891,
+      "step": 3708
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013837854687427897,
+      "loss": 1.1445,
+      "step": 3709
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013834916457307472,
+      "loss": 1.1296,
+      "step": 3710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013831977838974649,
+      "loss": 1.1237,
+      "step": 3711
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013829038832726903,
+      "loss": 1.1529,
+      "step": 3712
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001382609943886175,
+      "loss": 1.1633,
+      "step": 3713
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013823159657676759,
+      "loss": 1.1648,
+      "step": 3714
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001382021948946952,
+      "loss": 1.1317,
+      "step": 3715
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013817278934537668,
+      "loss": 1.19,
+      "step": 3716
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001381433799317888,
+      "loss": 1.1317,
+      "step": 3717
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001381139666569087,
+      "loss": 1.177,
+      "step": 3718
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013808454952371393,
+      "loss": 1.2252,
+      "step": 3719
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013805512853518235,
+      "loss": 1.1549,
+      "step": 3720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013802570369429238,
+      "loss": 1.1376,
+      "step": 3721
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013799627500402266,
+      "loss": 1.1646,
+      "step": 3722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001379668424673523,
+      "loss": 1.2044,
+      "step": 3723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013793740608726083,
+      "loss": 1.1803,
+      "step": 3724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013790796586672803,
+      "loss": 1.1676,
+      "step": 3725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001378785218087342,
+      "loss": 1.1748,
+      "step": 3726
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013784907391626002,
+      "loss": 1.1304,
+      "step": 3727
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013781962219228653,
+      "loss": 1.179,
+      "step": 3728
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001377901666397951,
+      "loss": 1.2083,
+      "step": 3729
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001377607072617676,
+      "loss": 1.1481,
+      "step": 3730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013773124406118626,
+      "loss": 1.1337,
+      "step": 3731
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001377017770410336,
+      "loss": 1.2501,
+      "step": 3732
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001376723062042926,
+      "loss": 1.1782,
+      "step": 3733
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013764283155394663,
+      "loss": 1.1344,
+      "step": 3734
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001376133530929795,
+      "loss": 1.1902,
+      "step": 3735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013758387082437527,
+      "loss": 1.1502,
+      "step": 3736
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001375543847511185,
+      "loss": 1.1955,
+      "step": 3737
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013752489487619404,
+      "loss": 1.1853,
+      "step": 3738
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013749540120258725,
+      "loss": 1.1296,
+      "step": 3739
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001374659037332838,
+      "loss": 1.2114,
+      "step": 3740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013743640247126967,
+      "loss": 1.2185,
+      "step": 3741
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013740689741953134,
+      "loss": 1.1901,
+      "step": 3742
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013737738858105567,
+      "loss": 1.156,
+      "step": 3743
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013734787595882987,
+      "loss": 1.243,
+      "step": 3744
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013731835955584152,
+      "loss": 1.1719,
+      "step": 3745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013728883937507856,
+      "loss": 1.131,
+      "step": 3746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013725931541952934,
+      "loss": 1.2097,
+      "step": 3747
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013722978769218264,
+      "loss": 1.1739,
+      "step": 3748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001372002561960276,
+      "loss": 1.1168,
+      "step": 3749
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013717072093405368,
+      "loss": 1.1815,
+      "step": 3750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013714118190925079,
+      "loss": 1.1952,
+      "step": 3751
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013711163912460918,
+      "loss": 1.1936,
+      "step": 3752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013708209258311952,
+      "loss": 1.2188,
+      "step": 3753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013705254228777277,
+      "loss": 1.2,
+      "step": 3754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013702298824156038,
+      "loss": 1.1544,
+      "step": 3755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013699343044747418,
+      "loss": 1.2026,
+      "step": 3756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013696386890850625,
+      "loss": 1.2084,
+      "step": 3757
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001369343036276492,
+      "loss": 1.1729,
+      "step": 3758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001369047346078959,
+      "loss": 1.1588,
+      "step": 3759
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013687516185223968,
+      "loss": 1.1302,
+      "step": 3760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013684558536367422,
+      "loss": 1.1242,
+      "step": 3761
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013681600514519357,
+      "loss": 1.1253,
+      "step": 3762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013678642119979218,
+      "loss": 1.0809,
+      "step": 3763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013675683353046487,
+      "loss": 1.213,
+      "step": 3764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001367272421402068,
+      "loss": 1.1669,
+      "step": 3765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013669764703201355,
+      "loss": 1.1448,
+      "step": 3766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013666804820888103,
+      "loss": 1.1764,
+      "step": 3767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001366384456738056,
+      "loss": 1.1649,
+      "step": 3768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000136608839429784,
+      "loss": 1.1843,
+      "step": 3769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013657922947981317,
+      "loss": 1.1772,
+      "step": 3770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013654961582689062,
+      "loss": 1.1538,
+      "step": 3771
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001365199984740142,
+      "loss": 1.1744,
+      "step": 3772
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001364903774241821,
+      "loss": 1.2408,
+      "step": 3773
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013646075268039284,
+      "loss": 1.2315,
+      "step": 3774
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013643112424564538,
+      "loss": 1.1873,
+      "step": 3775
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013640149212293908,
+      "loss": 1.1531,
+      "step": 3776
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013637185631527356,
+      "loss": 1.1477,
+      "step": 3777
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013634221682564895,
+      "loss": 1.1695,
+      "step": 3778
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013631257365706562,
+      "loss": 1.1548,
+      "step": 3779
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013628292681252443,
+      "loss": 1.1946,
+      "step": 3780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013625327629502652,
+      "loss": 1.2174,
+      "step": 3781
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013622362210757348,
+      "loss": 1.1819,
+      "step": 3782
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001361939642531672,
+      "loss": 1.1673,
+      "step": 3783
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013616430273481,
+      "loss": 1.2054,
+      "step": 3784
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013613463755550448,
+      "loss": 1.1423,
+      "step": 3785
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001361049687182538,
+      "loss": 1.1674,
+      "step": 3786
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013607529622606125,
+      "loss": 1.2084,
+      "step": 3787
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013604562008193066,
+      "loss": 1.2292,
+      "step": 3788
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013601594028886613,
+      "loss": 1.1738,
+      "step": 3789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013598625684987229,
+      "loss": 1.1633,
+      "step": 3790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013595656976795386,
+      "loss": 1.1775,
+      "step": 3791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001359268790461162,
+      "loss": 1.1594,
+      "step": 3792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013589718468736493,
+      "loss": 1.1336,
+      "step": 3793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013586748669470604,
+      "loss": 1.2558,
+      "step": 3794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013583778507114582,
+      "loss": 1.0672,
+      "step": 3795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013580807981969102,
+      "loss": 1.1242,
+      "step": 3796
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013577837094334876,
+      "loss": 1.139,
+      "step": 3797
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013574865844512654,
+      "loss": 1.2391,
+      "step": 3798
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001357189423280321,
+      "loss": 1.2161,
+      "step": 3799
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013568922259507367,
+      "loss": 1.1213,
+      "step": 3800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013565949924925982,
+      "loss": 1.1307,
+      "step": 3801
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013562977229359944,
+      "loss": 1.1744,
+      "step": 3802
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013560004173110185,
+      "loss": 1.1264,
+      "step": 3803
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013557030756477668,
+      "loss": 1.1776,
+      "step": 3804
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013554056979763396,
+      "loss": 1.1236,
+      "step": 3805
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001355108284326841,
+      "loss": 1.1274,
+      "step": 3806
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013548108347293783,
+      "loss": 1.1347,
+      "step": 3807
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013545133492140626,
+      "loss": 1.188,
+      "step": 3808
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013542158278110082,
+      "loss": 1.154,
+      "step": 3809
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013539182705503338,
+      "loss": 1.1966,
+      "step": 3810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001353620677462162,
+      "loss": 1.2043,
+      "step": 3811
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013533230485766176,
+      "loss": 1.0848,
+      "step": 3812
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013530253839238304,
+      "loss": 1.1635,
+      "step": 3813
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001352727683533933,
+      "loss": 1.1655,
+      "step": 3814
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013524299474370626,
+      "loss": 1.1518,
+      "step": 3815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001352132175663358,
+      "loss": 1.2135,
+      "step": 3816
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001351834368242964,
+      "loss": 1.1072,
+      "step": 3817
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013515365252060273,
+      "loss": 1.1633,
+      "step": 3818
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013512386465826996,
+      "loss": 1.1668,
+      "step": 3819
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001350940732403135,
+      "loss": 1.1973,
+      "step": 3820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013506427826974913,
+      "loss": 1.1701,
+      "step": 3821
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013503447974959307,
+      "loss": 1.2569,
+      "step": 3822
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013500467768286188,
+      "loss": 1.1356,
+      "step": 3823
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013497487207257236,
+      "loss": 1.1818,
+      "step": 3824
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013494506292174185,
+      "loss": 1.1509,
+      "step": 3825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001349152502333879,
+      "loss": 1.1768,
+      "step": 3826
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013488543401052856,
+      "loss": 1.1113,
+      "step": 3827
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001348556142561821,
+      "loss": 1.1835,
+      "step": 3828
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013482579097336716,
+      "loss": 1.2021,
+      "step": 3829
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013479596416510282,
+      "loss": 1.1795,
+      "step": 3830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001347661338344085,
+      "loss": 1.2185,
+      "step": 3831
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013473629998430396,
+      "loss": 1.1624,
+      "step": 3832
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013470646261780928,
+      "loss": 1.1886,
+      "step": 3833
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013467662173794492,
+      "loss": 1.1707,
+      "step": 3834
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013464677734773171,
+      "loss": 1.1721,
+      "step": 3835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013461692945019086,
+      "loss": 1.1694,
+      "step": 3836
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013458707804834386,
+      "loss": 1.1444,
+      "step": 3837
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013455722314521261,
+      "loss": 1.1313,
+      "step": 3838
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013452736474381936,
+      "loss": 1.1618,
+      "step": 3839
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001344975028471867,
+      "loss": 1.1057,
+      "step": 3840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001344676374583376,
+      "loss": 1.1178,
+      "step": 3841
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001344377685802953,
+      "loss": 1.2211,
+      "step": 3842
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001344078962160835,
+      "loss": 1.2403,
+      "step": 3843
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013437802036872623,
+      "loss": 1.178,
+      "step": 3844
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013434814104124784,
+      "loss": 1.1577,
+      "step": 3845
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000134318258236673,
+      "loss": 1.1304,
+      "step": 3846
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013428837195802684,
+      "loss": 1.1539,
+      "step": 3847
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013425848220833473,
+      "loss": 1.1923,
+      "step": 3848
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013422858899062249,
+      "loss": 1.1636,
+      "step": 3849
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013419869230791616,
+      "loss": 1.1994,
+      "step": 3850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013416879216324222,
+      "loss": 1.1366,
+      "step": 3851
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013413888855962758,
+      "loss": 1.2084,
+      "step": 3852
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013410898150009937,
+      "loss": 1.1953,
+      "step": 3853
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013407907098768508,
+      "loss": 1.1581,
+      "step": 3854
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001340491570254126,
+      "loss": 1.2304,
+      "step": 3855
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013401923961631014,
+      "loss": 1.1974,
+      "step": 3856
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013398931876340632,
+      "loss": 1.2148,
+      "step": 3857
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013395939446973,
+      "loss": 1.1246,
+      "step": 3858
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013392946673831046,
+      "loss": 1.2236,
+      "step": 3859
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001338995355721773,
+      "loss": 1.1849,
+      "step": 3860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013386960097436056,
+      "loss": 1.1521,
+      "step": 3861
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013383966294789045,
+      "loss": 1.2431,
+      "step": 3862
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013380972149579766,
+      "loss": 1.1346,
+      "step": 3863
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001337797766211132,
+      "loss": 1.2138,
+      "step": 3864
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013374982832686842,
+      "loss": 1.1525,
+      "step": 3865
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000133719876616095,
+      "loss": 1.1355,
+      "step": 3866
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000133689921491825,
+      "loss": 1.124,
+      "step": 3867
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001336599629570908,
+      "loss": 1.1775,
+      "step": 3868
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013363000101492513,
+      "loss": 1.128,
+      "step": 3869
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001336000356683611,
+      "loss": 1.2109,
+      "step": 3870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013357006692043207,
+      "loss": 1.1931,
+      "step": 3871
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013354009477417185,
+      "loss": 1.2129,
+      "step": 3872
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013351011923261455,
+      "loss": 1.1848,
+      "step": 3873
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001334801402987946,
+      "loss": 1.2042,
+      "step": 3874
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013345015797574681,
+      "loss": 1.1595,
+      "step": 3875
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013342017226650635,
+      "loss": 1.1238,
+      "step": 3876
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013339018317410863,
+      "loss": 1.1258,
+      "step": 3877
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013336019070158954,
+      "loss": 1.1886,
+      "step": 3878
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001333301948519852,
+      "loss": 1.1968,
+      "step": 3879
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013330019562833218,
+      "loss": 1.2284,
+      "step": 3880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001332701930336673,
+      "loss": 1.1907,
+      "step": 3881
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013324018707102771,
+      "loss": 1.1359,
+      "step": 3882
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013321017774345104,
+      "loss": 1.1533,
+      "step": 3883
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013318016505397507,
+      "loss": 1.1982,
+      "step": 3884
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013315014900563803,
+      "loss": 1.2349,
+      "step": 3885
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013312012960147853,
+      "loss": 1.2287,
+      "step": 3886
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013309010684453543,
+      "loss": 1.1472,
+      "step": 3887
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013306008073784794,
+      "loss": 1.1834,
+      "step": 3888
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013303005128445568,
+      "loss": 1.1218,
+      "step": 3889
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013300001848739852,
+      "loss": 1.1811,
+      "step": 3890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013296998234971677,
+      "loss": 1.1649,
+      "step": 3891
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001329399428744509,
+      "loss": 1.1845,
+      "step": 3892
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013290990006464197,
+      "loss": 1.1798,
+      "step": 3893
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013287985392333116,
+      "loss": 1.1907,
+      "step": 3894
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013284980445356012,
+      "loss": 1.1832,
+      "step": 3895
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013281975165837072,
+      "loss": 1.1785,
+      "step": 3896
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001327896955408053,
+      "loss": 1.1808,
+      "step": 3897
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013275963610390648,
+      "loss": 1.1496,
+      "step": 3898
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013272957335071713,
+      "loss": 1.1633,
+      "step": 3899
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001326995072842806,
+      "loss": 1.1275,
+      "step": 3900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001326694379076405,
+      "loss": 1.1353,
+      "step": 3901
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013263936522384071,
+      "loss": 1.2171,
+      "step": 3902
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001326092892359256,
+      "loss": 1.2602,
+      "step": 3903
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001325792099469398,
+      "loss": 1.2029,
+      "step": 3904
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013254912735992822,
+      "loss": 1.1672,
+      "step": 3905
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013251904147793613,
+      "loss": 1.1534,
+      "step": 3906
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013248895230400925,
+      "loss": 1.0839,
+      "step": 3907
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013245885984119343,
+      "loss": 1.2171,
+      "step": 3908
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013242876409253498,
+      "loss": 1.1761,
+      "step": 3909
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013239866506108058,
+      "loss": 1.0989,
+      "step": 3910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013236856274987716,
+      "loss": 1.1521,
+      "step": 3911
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000132338457161972,
+      "loss": 1.1087,
+      "step": 3912
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001323083483004127,
+      "loss": 1.1389,
+      "step": 3913
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013227823616824723,
+      "loss": 1.1567,
+      "step": 3914
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013224812076852386,
+      "loss": 1.1781,
+      "step": 3915
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013221800210429122,
+      "loss": 1.1441,
+      "step": 3916
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013218788017859827,
+      "loss": 1.2046,
+      "step": 3917
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001321577549944942,
+      "loss": 1.1729,
+      "step": 3918
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001321276265550287,
+      "loss": 1.1416,
+      "step": 3919
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013209749486325166,
+      "loss": 1.1801,
+      "step": 3920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013206735992221335,
+      "loss": 1.2146,
+      "step": 3921
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013203722173496434,
+      "loss": 1.1733,
+      "step": 3922
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001320070803045556,
+      "loss": 1.1211,
+      "step": 3923
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013197693563403836,
+      "loss": 1.1934,
+      "step": 3924
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013194678772646416,
+      "loss": 1.1667,
+      "step": 3925
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013191663658488492,
+      "loss": 1.1788,
+      "step": 3926
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013188648221235284,
+      "loss": 1.1572,
+      "step": 3927
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013185632461192056,
+      "loss": 1.1766,
+      "step": 3928
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001318261637866409,
+      "loss": 1.2486,
+      "step": 3929
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001317959997395671,
+      "loss": 1.2281,
+      "step": 3930
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013176583247375266,
+      "loss": 1.2399,
+      "step": 3931
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013173566199225145,
+      "loss": 1.1556,
+      "step": 3932
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013170548829811765,
+      "loss": 1.2477,
+      "step": 3933
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013167531139440585,
+      "loss": 1.2198,
+      "step": 3934
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001316451312841708,
+      "loss": 1.1545,
+      "step": 3935
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001316149479704677,
+      "loss": 1.209,
+      "step": 3936
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013158476145635207,
+      "loss": 1.1741,
+      "step": 3937
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001315545717448796,
+      "loss": 1.1493,
+      "step": 3938
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013152437883910657,
+      "loss": 1.1632,
+      "step": 3939
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013149418274208932,
+      "loss": 1.1576,
+      "step": 3940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013146398345688476,
+      "loss": 1.1894,
+      "step": 3941
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013143378098654987,
+      "loss": 1.1849,
+      "step": 3942
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013140357533414213,
+      "loss": 1.1918,
+      "step": 3943
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013137336650271926,
+      "loss": 1.1183,
+      "step": 3944
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001313431544953394,
+      "loss": 1.1421,
+      "step": 3945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001313129393150609,
+      "loss": 1.2495,
+      "step": 3946
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001312827209649424,
+      "loss": 1.2112,
+      "step": 3947
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013125249944804303,
+      "loss": 1.1807,
+      "step": 3948
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013122227476742215,
+      "loss": 1.1642,
+      "step": 3949
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013119204692613935,
+      "loss": 1.1332,
+      "step": 3950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013116181592725472,
+      "loss": 1.1457,
+      "step": 3951
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013113158177382848,
+      "loss": 1.1977,
+      "step": 3952
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013110134446892135,
+      "loss": 1.1479,
+      "step": 3953
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013107110401559423,
+      "loss": 1.1386,
+      "step": 3954
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001310408604169084,
+      "loss": 1.1622,
+      "step": 3955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013101061367592543,
+      "loss": 1.1439,
+      "step": 3956
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001309803637957073,
+      "loss": 1.1831,
+      "step": 3957
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013095011077931615,
+      "loss": 1.1851,
+      "step": 3958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013091985462981457,
+      "loss": 1.1616,
+      "step": 3959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013088959535026542,
+      "loss": 1.1516,
+      "step": 3960
+    },
+    {
+      "epoch": 1.59,
+      "eval_loss": 0.8274641036987305,
+      "eval_runtime": 502.1403,
+      "eval_samples_per_second": 51.579,
+      "eval_steps_per_second": 25.79,
+      "step": 3960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013085933294373184,
+      "loss": 1.114,
+      "step": 3961
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013082906741327738,
+      "loss": 1.1832,
+      "step": 3962
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001307987987619658,
+      "loss": 1.246,
+      "step": 3963
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013076852699286126,
+      "loss": 1.1901,
+      "step": 3964
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001307382521090282,
+      "loss": 1.2107,
+      "step": 3965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013070797411353133,
+      "loss": 1.1475,
+      "step": 3966
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013067769300943578,
+      "loss": 1.1616,
+      "step": 3967
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013064740879980687,
+      "loss": 1.204,
+      "step": 3968
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013061712148771038,
+      "loss": 1.0646,
+      "step": 3969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013058683107621227,
+      "loss": 1.1619,
+      "step": 3970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001305565375683789,
+      "loss": 1.2415,
+      "step": 3971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001305262409672769,
+      "loss": 1.1458,
+      "step": 3972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001304959412759732,
+      "loss": 1.1365,
+      "step": 3973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001304656384975351,
+      "loss": 1.0984,
+      "step": 3974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013043533263503018,
+      "loss": 1.2134,
+      "step": 3975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001304050236915263,
+      "loss": 1.1716,
+      "step": 3976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001303747116700917,
+      "loss": 1.2107,
+      "step": 3977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013034439657379486,
+      "loss": 1.1404,
+      "step": 3978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001303140784057047,
+      "loss": 1.1077,
+      "step": 3979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013028375716889022,
+      "loss": 1.148,
+      "step": 3980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013025343286642094,
+      "loss": 1.1457,
+      "step": 3981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013022310550136663,
+      "loss": 1.2004,
+      "step": 3982
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013019277507679737,
+      "loss": 1.2195,
+      "step": 3983
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013016244159578348,
+      "loss": 1.2418,
+      "step": 3984
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001301321050613957,
+      "loss": 1.1245,
+      "step": 3985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000130101765476705,
+      "loss": 1.1998,
+      "step": 3986
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001300714228447827,
+      "loss": 1.2106,
+      "step": 3987
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013004107716870042,
+      "loss": 1.2093,
+      "step": 3988
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013001072845153006,
+      "loss": 1.1905,
+      "step": 3989
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012998037669634388,
+      "loss": 1.1556,
+      "step": 3990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012995002190621438,
+      "loss": 1.2288,
+      "step": 3991
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012991966408421447,
+      "loss": 1.2107,
+      "step": 3992
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012988930323341725,
+      "loss": 1.2466,
+      "step": 3993
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012985893935689617,
+      "loss": 1.1442,
+      "step": 3994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012982857245772504,
+      "loss": 1.2012,
+      "step": 3995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012979820253897786,
+      "loss": 1.1113,
+      "step": 3996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012976782960372912,
+      "loss": 1.2086,
+      "step": 3997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012973745365505338,
+      "loss": 1.1619,
+      "step": 3998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012970707469602572,
+      "loss": 1.1584,
+      "step": 3999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012967669272972141,
+      "loss": 1.161,
+      "step": 4000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000129646307759216,
+      "loss": 1.2158,
+      "step": 4001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001296159197875854,
+      "loss": 1.1741,
+      "step": 4002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001295855288179059,
+      "loss": 1.1161,
+      "step": 4003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001295551348532539,
+      "loss": 1.1602,
+      "step": 4004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001295247378967063,
+      "loss": 1.1715,
+      "step": 4005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001294943379513401,
+      "loss": 1.1147,
+      "step": 4006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012946393502023286,
+      "loss": 1.1481,
+      "step": 4007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001294335291064622,
+      "loss": 1.1597,
+      "step": 4008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012940312021310617,
+      "loss": 1.195,
+      "step": 4009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012937270834324308,
+      "loss": 1.2635,
+      "step": 4010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012934229349995157,
+      "loss": 1.1435,
+      "step": 4011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012931187568631058,
+      "loss": 1.1527,
+      "step": 4012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001292814549053993,
+      "loss": 1.2303,
+      "step": 4013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012925103116029726,
+      "loss": 1.1057,
+      "step": 4014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001292206044540843,
+      "loss": 1.1926,
+      "step": 4015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012919017478984056,
+      "loss": 1.2143,
+      "step": 4016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012915974217064645,
+      "loss": 1.1338,
+      "step": 4017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012912930659958265,
+      "loss": 1.1536,
+      "step": 4018
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012909886807973024,
+      "loss": 1.0822,
+      "step": 4019
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012906842661417054,
+      "loss": 1.155,
+      "step": 4020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012903798220598517,
+      "loss": 1.1863,
+      "step": 4021
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000129007534858256,
+      "loss": 1.1262,
+      "step": 4022
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001289770845740653,
+      "loss": 1.1483,
+      "step": 4023
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012894663135649555,
+      "loss": 1.1975,
+      "step": 4024
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012891617520862958,
+      "loss": 1.1083,
+      "step": 4025
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012888571613355046,
+      "loss": 1.1075,
+      "step": 4026
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012885525413434165,
+      "loss": 1.1394,
+      "step": 4027
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001288247892140868,
+      "loss": 1.1955,
+      "step": 4028
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012879432137586995,
+      "loss": 1.2209,
+      "step": 4029
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012876385062277534,
+      "loss": 1.1234,
+      "step": 4030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012873337695788755,
+      "loss": 1.087,
+      "step": 4031
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012870290038429153,
+      "loss": 1.113,
+      "step": 4032
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012867242090507238,
+      "loss": 1.1656,
+      "step": 4033
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012864193852331563,
+      "loss": 1.1804,
+      "step": 4034
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000128611453242107,
+      "loss": 1.2178,
+      "step": 4035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001285809650645325,
+      "loss": 1.1469,
+      "step": 4036
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012855047399367857,
+      "loss": 1.1836,
+      "step": 4037
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012851998003263182,
+      "loss": 1.1639,
+      "step": 4038
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012848948318447919,
+      "loss": 1.1723,
+      "step": 4039
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012845898345230787,
+      "loss": 1.1599,
+      "step": 4040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012842848083920543,
+      "loss": 1.1766,
+      "step": 4041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012839797534825963,
+      "loss": 1.163,
+      "step": 4042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012836746698255858,
+      "loss": 1.1783,
+      "step": 4043
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001283369557451907,
+      "loss": 1.1139,
+      "step": 4044
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012830644163924464,
+      "loss": 1.1611,
+      "step": 4045
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012827592466780943,
+      "loss": 1.1973,
+      "step": 4046
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012824540483397426,
+      "loss": 1.233,
+      "step": 4047
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012821488214082876,
+      "loss": 1.1339,
+      "step": 4048
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012818435659146268,
+      "loss": 1.1862,
+      "step": 4049
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001281538281889662,
+      "loss": 1.1179,
+      "step": 4050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001281232969364298,
+      "loss": 1.1751,
+      "step": 4051
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001280927628369441,
+      "loss": 1.1599,
+      "step": 4052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001280622258936001,
+      "loss": 1.1275,
+      "step": 4053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012803168610948915,
+      "loss": 1.1827,
+      "step": 4054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012800114348770277,
+      "loss": 1.2047,
+      "step": 4055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012797059803133288,
+      "loss": 1.1823,
+      "step": 4056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012794004974347152,
+      "loss": 1.1646,
+      "step": 4057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001279094986272112,
+      "loss": 1.1726,
+      "step": 4058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012787894468564465,
+      "loss": 1.2122,
+      "step": 4059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012784838792186484,
+      "loss": 1.1592,
+      "step": 4060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012781782833896506,
+      "loss": 1.1452,
+      "step": 4061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012778726594003893,
+      "loss": 1.1681,
+      "step": 4062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012775670072818026,
+      "loss": 1.0696,
+      "step": 4063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012772613270648324,
+      "loss": 1.0958,
+      "step": 4064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012769556187804227,
+      "loss": 1.1654,
+      "step": 4065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012766498824595205,
+      "loss": 1.2186,
+      "step": 4066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012763441181330763,
+      "loss": 1.2499,
+      "step": 4067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001276038325832043,
+      "loss": 1.1782,
+      "step": 4068
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012757325055873754,
+      "loss": 1.1631,
+      "step": 4069
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012754266574300325,
+      "loss": 1.1762,
+      "step": 4070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001275120781390976,
+      "loss": 1.2091,
+      "step": 4071
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012748148775011697,
+      "loss": 1.1819,
+      "step": 4072
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012745089457915802,
+      "loss": 1.1406,
+      "step": 4073
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012742029862931778,
+      "loss": 1.1827,
+      "step": 4074
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012738969990369348,
+      "loss": 1.1852,
+      "step": 4075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001273590984053827,
+      "loss": 1.1765,
+      "step": 4076
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012732849413748317,
+      "loss": 1.1865,
+      "step": 4077
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012729788710309306,
+      "loss": 1.1662,
+      "step": 4078
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012726727730531074,
+      "loss": 1.2061,
+      "step": 4079
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001272366647472349,
+      "loss": 1.2255,
+      "step": 4080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012720604943196442,
+      "loss": 1.1709,
+      "step": 4081
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012717543136259857,
+      "loss": 1.1753,
+      "step": 4082
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001271448105422368,
+      "loss": 1.1847,
+      "step": 4083
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012711418697397895,
+      "loss": 1.1914,
+      "step": 4084
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012708356066092497,
+      "loss": 1.1949,
+      "step": 4085
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012705293160617532,
+      "loss": 1.1832,
+      "step": 4086
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001270222998128305,
+      "loss": 1.1599,
+      "step": 4087
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001269916652839915,
+      "loss": 1.1341,
+      "step": 4088
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012696102802275939,
+      "loss": 1.1469,
+      "step": 4089
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012693038803223564,
+      "loss": 1.2048,
+      "step": 4090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012689974531552192,
+      "loss": 1.2194,
+      "step": 4091
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012686909987572037,
+      "loss": 1.1288,
+      "step": 4092
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012683845171593313,
+      "loss": 1.196,
+      "step": 4093
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001268078008392628,
+      "loss": 1.1727,
+      "step": 4094
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012677714724881214,
+      "loss": 1.1191,
+      "step": 4095
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001267464909476843,
+      "loss": 1.0917,
+      "step": 4096
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012671583193898264,
+      "loss": 1.1948,
+      "step": 4097
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012668517022581074,
+      "loss": 1.1399,
+      "step": 4098
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001266545058112726,
+      "loss": 1.1771,
+      "step": 4099
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012662383869847238,
+      "loss": 1.2045,
+      "step": 4100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012659316889051457,
+      "loss": 1.1663,
+      "step": 4101
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012656249639050383,
+      "loss": 1.1798,
+      "step": 4102
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001265318212015452,
+      "loss": 1.1807,
+      "step": 4103
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000126501143326744,
+      "loss": 1.2179,
+      "step": 4104
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012647046276920574,
+      "loss": 1.1397,
+      "step": 4105
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012643977953203625,
+      "loss": 1.1006,
+      "step": 4106
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012640909361834164,
+      "loss": 1.1423,
+      "step": 4107
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012637840503122824,
+      "loss": 1.1585,
+      "step": 4108
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012634771377380277,
+      "loss": 1.1562,
+      "step": 4109
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000126317019849172,
+      "loss": 1.1807,
+      "step": 4110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012628632326044324,
+      "loss": 1.153,
+      "step": 4111
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012625562401072385,
+      "loss": 1.1147,
+      "step": 4112
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001262249221031216,
+      "loss": 1.1856,
+      "step": 4113
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012619421754074444,
+      "loss": 1.1436,
+      "step": 4114
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012616351032670062,
+      "loss": 1.1378,
+      "step": 4115
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001261328004640987,
+      "loss": 1.1971,
+      "step": 4116
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001261020879560474,
+      "loss": 1.1576,
+      "step": 4117
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001260713728056559,
+      "loss": 1.2477,
+      "step": 4118
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012604065501603336,
+      "loss": 1.1332,
+      "step": 4119
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012600993459028952,
+      "loss": 1.2221,
+      "step": 4120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012597921153153415,
+      "loss": 1.1609,
+      "step": 4121
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012594848584287745,
+      "loss": 1.2387,
+      "step": 4122
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012591775752742975,
+      "loss": 1.2114,
+      "step": 4123
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001258870265883017,
+      "loss": 1.1907,
+      "step": 4124
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012585629302860424,
+      "loss": 1.1967,
+      "step": 4125
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001258255568514486,
+      "loss": 1.1741,
+      "step": 4126
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001257948180599462,
+      "loss": 1.2033,
+      "step": 4127
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012576407665720874,
+      "loss": 1.1961,
+      "step": 4128
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012573333264634824,
+      "loss": 1.1644,
+      "step": 4129
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012570258603047694,
+      "loss": 1.1721,
+      "step": 4130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001256718368127073,
+      "loss": 1.193,
+      "step": 4131
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012564108499615214,
+      "loss": 1.1629,
+      "step": 4132
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001256103305839245,
+      "loss": 1.2299,
+      "step": 4133
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001255795735791377,
+      "loss": 1.1623,
+      "step": 4134
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012554881398490523,
+      "loss": 1.2368,
+      "step": 4135
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012551805180434098,
+      "loss": 1.1485,
+      "step": 4136
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012548728704055902,
+      "loss": 1.1652,
+      "step": 4137
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012545651969667363,
+      "loss": 1.1596,
+      "step": 4138
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012542574977579958,
+      "loss": 1.1277,
+      "step": 4139
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012539497728105155,
+      "loss": 1.1516,
+      "step": 4140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012536420221554478,
+      "loss": 1.1507,
+      "step": 4141
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012533342458239465,
+      "loss": 1.1421,
+      "step": 4142
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012530264438471682,
+      "loss": 1.1783,
+      "step": 4143
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012527186162562717,
+      "loss": 1.1706,
+      "step": 4144
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012524107630824185,
+      "loss": 1.1336,
+      "step": 4145
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001252102884356773,
+      "loss": 1.1901,
+      "step": 4146
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001251794980110503,
+      "loss": 1.2021,
+      "step": 4147
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012514870503747767,
+      "loss": 1.1856,
+      "step": 4148
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012511790951807666,
+      "loss": 1.1385,
+      "step": 4149
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012508711145596475,
+      "loss": 1.1712,
+      "step": 4150
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012505631085425964,
+      "loss": 1.2107,
+      "step": 4151
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001250255077160793,
+      "loss": 1.1638,
+      "step": 4152
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012499470204454196,
+      "loss": 1.1603,
+      "step": 4153
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012496389384276612,
+      "loss": 1.1112,
+      "step": 4154
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012493308311387055,
+      "loss": 1.1867,
+      "step": 4155
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012490226986097423,
+      "loss": 1.1898,
+      "step": 4156
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012487145408719637,
+      "loss": 1.164,
+      "step": 4157
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012484063579565652,
+      "loss": 1.1834,
+      "step": 4158
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012480981498947448,
+      "loss": 1.1537,
+      "step": 4159
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012477899167177023,
+      "loss": 1.2348,
+      "step": 4160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012474816584566404,
+      "loss": 1.1775,
+      "step": 4161
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012471733751427645,
+      "loss": 1.1187,
+      "step": 4162
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012468650668072826,
+      "loss": 1.1628,
+      "step": 4163
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001246556733481405,
+      "loss": 1.1772,
+      "step": 4164
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012462483751963442,
+      "loss": 1.1776,
+      "step": 4165
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012459399919833162,
+      "loss": 1.1655,
+      "step": 4166
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012456315838735383,
+      "loss": 1.1908,
+      "step": 4167
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012453231508982316,
+      "loss": 1.1853,
+      "step": 4168
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012450146930886185,
+      "loss": 1.1646,
+      "step": 4169
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001244706210475925,
+      "loss": 1.1777,
+      "step": 4170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012443977030913788,
+      "loss": 1.1972,
+      "step": 4171
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001244089170966211,
+      "loss": 1.0957,
+      "step": 4172
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012437806141316537,
+      "loss": 1.1567,
+      "step": 4173
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012434720326189423,
+      "loss": 1.1962,
+      "step": 4174
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001243163426459316,
+      "loss": 1.1264,
+      "step": 4175
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012428547956840148,
+      "loss": 1.1457,
+      "step": 4176
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001242546140324281,
+      "loss": 1.1574,
+      "step": 4177
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012422374604113606,
+      "loss": 1.1082,
+      "step": 4178
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012419287559765018,
+      "loss": 1.1974,
+      "step": 4179
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001241620027050955,
+      "loss": 1.1085,
+      "step": 4180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012413112736659726,
+      "loss": 1.1061,
+      "step": 4181
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012410024958528105,
+      "loss": 1.2347,
+      "step": 4182
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012406936936427268,
+      "loss": 1.1729,
+      "step": 4183
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012403848670669814,
+      "loss": 1.2115,
+      "step": 4184
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001240076016156837,
+      "loss": 1.1301,
+      "step": 4185
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012397671409435597,
+      "loss": 1.1733,
+      "step": 4186
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012394582414584164,
+      "loss": 1.1096,
+      "step": 4187
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001239149317732677,
+      "loss": 1.0802,
+      "step": 4188
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012388403697976156,
+      "loss": 1.2061,
+      "step": 4189
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012385313976845063,
+      "loss": 1.2413,
+      "step": 4190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012382224014246266,
+      "loss": 1.1639,
+      "step": 4191
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012379133810492567,
+      "loss": 1.1244,
+      "step": 4192
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001237604336589679,
+      "loss": 1.1422,
+      "step": 4193
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012372952680771789,
+      "loss": 1.1339,
+      "step": 4194
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012369861755430426,
+      "loss": 1.2067,
+      "step": 4195
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012366770590185606,
+      "loss": 1.1836,
+      "step": 4196
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012363679185350255,
+      "loss": 1.2289,
+      "step": 4197
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012360587541237306,
+      "loss": 1.165,
+      "step": 4198
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001235749565815974,
+      "loss": 1.1433,
+      "step": 4199
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001235440353643055,
+      "loss": 1.1761,
+      "step": 4200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001235131117636275,
+      "loss": 1.1303,
+      "step": 4201
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012348218578269387,
+      "loss": 1.1795,
+      "step": 4202
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012345125742463525,
+      "loss": 1.1567,
+      "step": 4203
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012342032669258255,
+      "loss": 1.1651,
+      "step": 4204
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012338939358966694,
+      "loss": 1.2013,
+      "step": 4205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012335845811901983,
+      "loss": 1.1483,
+      "step": 4206
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012332752028377278,
+      "loss": 1.2374,
+      "step": 4207
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012329658008705766,
+      "loss": 1.1417,
+      "step": 4208
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012326563753200663,
+      "loss": 1.178,
+      "step": 4209
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012323469262175206,
+      "loss": 1.1796,
+      "step": 4210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012320374535942645,
+      "loss": 1.1702,
+      "step": 4211
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012317279574816268,
+      "loss": 1.2223,
+      "step": 4212
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012314184379109378,
+      "loss": 1.1719,
+      "step": 4213
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012311088949135306,
+      "loss": 1.2116,
+      "step": 4214
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012307993285207405,
+      "loss": 1.141,
+      "step": 4215
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001230489738763905,
+      "loss": 1.208,
+      "step": 4216
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001230180125674365,
+      "loss": 1.1366,
+      "step": 4217
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012298704892834623,
+      "loss": 1.1069,
+      "step": 4218
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012295608296225414,
+      "loss": 1.1563,
+      "step": 4219
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012292511467229498,
+      "loss": 1.1854,
+      "step": 4220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012289414406160373,
+      "loss": 1.1766,
+      "step": 4221
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012286317113331555,
+      "loss": 1.1632,
+      "step": 4222
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012283219589056583,
+      "loss": 1.1501,
+      "step": 4223
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012280121833649026,
+      "loss": 1.1273,
+      "step": 4224
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012277023847422473,
+      "loss": 1.1699,
+      "step": 4225
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012273925630690535,
+      "loss": 1.1171,
+      "step": 4226
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001227082718376685,
+      "loss": 1.1399,
+      "step": 4227
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012267728506965074,
+      "loss": 1.1216,
+      "step": 4228
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012264629600598884,
+      "loss": 1.1819,
+      "step": 4229
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012261530464981998,
+      "loss": 1.2214,
+      "step": 4230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012258431100428137,
+      "loss": 1.0803,
+      "step": 4231
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012255331507251053,
+      "loss": 1.2347,
+      "step": 4232
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001225223168576452,
+      "loss": 1.1679,
+      "step": 4233
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001224913163628234,
+      "loss": 1.1461,
+      "step": 4234
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001224603135911833,
+      "loss": 1.1822,
+      "step": 4235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012242930854586335,
+      "loss": 1.1442,
+      "step": 4236
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012239830123000225,
+      "loss": 1.2147,
+      "step": 4237
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012236729164673887,
+      "loss": 1.1959,
+      "step": 4238
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012233627979921233,
+      "loss": 1.1714,
+      "step": 4239
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012230526569056207,
+      "loss": 1.2099,
+      "step": 4240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012227424932392757,
+      "loss": 1.2267,
+      "step": 4241
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012224323070244872,
+      "loss": 1.1611,
+      "step": 4242
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012221220982926552,
+      "loss": 1.1603,
+      "step": 4243
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012218118670751828,
+      "loss": 1.1662,
+      "step": 4244
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012215016134034748,
+      "loss": 1.1163,
+      "step": 4245
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012211913373089389,
+      "loss": 1.2462,
+      "step": 4246
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001220881038822984,
+      "loss": 1.1562,
+      "step": 4247
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012205707179770225,
+      "loss": 1.2022,
+      "step": 4248
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012202603748024681,
+      "loss": 1.2211,
+      "step": 4249
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001219950009330737,
+      "loss": 1.1628,
+      "step": 4250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012196396215932484,
+      "loss": 1.1865,
+      "step": 4251
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012193292116214232,
+      "loss": 1.0746,
+      "step": 4252
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012190187794466839,
+      "loss": 1.1453,
+      "step": 4253
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012187083251004558,
+      "loss": 1.1664,
+      "step": 4254
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012183978486141671,
+      "loss": 1.2211,
+      "step": 4255
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012180873500192473,
+      "loss": 1.2496,
+      "step": 4256
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012177768293471286,
+      "loss": 1.2332,
+      "step": 4257
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012174662866292452,
+      "loss": 1.155,
+      "step": 4258
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012171557218970337,
+      "loss": 1.1215,
+      "step": 4259
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001216845135181933,
+      "loss": 1.2205,
+      "step": 4260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012165345265153843,
+      "loss": 1.1024,
+      "step": 4261
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012162238959288301,
+      "loss": 1.144,
+      "step": 4262
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012159132434537159,
+      "loss": 1.1087,
+      "step": 4263
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012156025691214905,
+      "loss": 1.1213,
+      "step": 4264
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012152918729636025,
+      "loss": 1.1412,
+      "step": 4265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012149811550115047,
+      "loss": 1.1592,
+      "step": 4266
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012146704152966508,
+      "loss": 1.1485,
+      "step": 4267
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001214359653850498,
+      "loss": 1.1716,
+      "step": 4268
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012140488707045045,
+      "loss": 1.2632,
+      "step": 4269
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012137380658901311,
+      "loss": 1.1455,
+      "step": 4270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012134272394388415,
+      "loss": 1.0945,
+      "step": 4271
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012131163913821004,
+      "loss": 1.116,
+      "step": 4272
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012128055217513757,
+      "loss": 1.1643,
+      "step": 4273
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012124946305781364,
+      "loss": 1.2343,
+      "step": 4274
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012121837178938547,
+      "loss": 1.1275,
+      "step": 4275
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012118727837300046,
+      "loss": 1.1892,
+      "step": 4276
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012115618281180622,
+      "loss": 1.2117,
+      "step": 4277
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012112508510895062,
+      "loss": 1.1807,
+      "step": 4278
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012109398526758166,
+      "loss": 1.2139,
+      "step": 4279
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012106288329084762,
+      "loss": 1.193,
+      "step": 4280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012103177918189704,
+      "loss": 1.1189,
+      "step": 4281
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012100067294387853,
+      "loss": 1.3081,
+      "step": 4282
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012096956457994108,
+      "loss": 1.1558,
+      "step": 4283
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012093845409323374,
+      "loss": 1.1162,
+      "step": 4284
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012090734148690597,
+      "loss": 1.1197,
+      "step": 4285
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012087622676410722,
+      "loss": 1.1897,
+      "step": 4286
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012084510992798733,
+      "loss": 1.1499,
+      "step": 4287
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012081399098169626,
+      "loss": 1.1305,
+      "step": 4288
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012078286992838421,
+      "loss": 1.2443,
+      "step": 4289
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012075174677120165,
+      "loss": 1.1558,
+      "step": 4290
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012072062151329909,
+      "loss": 1.1803,
+      "step": 4291
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012068949415782749,
+      "loss": 1.149,
+      "step": 4292
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012065836470793784,
+      "loss": 1.1589,
+      "step": 4293
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012062723316678144,
+      "loss": 1.15,
+      "step": 4294
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012059609953750974,
+      "loss": 1.2012,
+      "step": 4295
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012056496382327443,
+      "loss": 1.2431,
+      "step": 4296
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012053382602722742,
+      "loss": 1.1897,
+      "step": 4297
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012050268615252082,
+      "loss": 1.1968,
+      "step": 4298
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012047154420230694,
+      "loss": 1.1837,
+      "step": 4299
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012044040017973832,
+      "loss": 1.1008,
+      "step": 4300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012040925408796769,
+      "loss": 1.183,
+      "step": 4301
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012037810593014803,
+      "loss": 1.1649,
+      "step": 4302
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012034695570943247,
+      "loss": 1.233,
+      "step": 4303
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012031580342897437,
+      "loss": 1.2524,
+      "step": 4304
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012028464909192729,
+      "loss": 1.193,
+      "step": 4305
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012025349270144507,
+      "loss": 1.1645,
+      "step": 4306
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012022233426068173,
+      "loss": 1.1405,
+      "step": 4307
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012019117377279136,
+      "loss": 1.1526,
+      "step": 4308
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012016001124092844,
+      "loss": 1.2098,
+      "step": 4309
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012012884666824756,
+      "loss": 1.0922,
+      "step": 4310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012009768005790356,
+      "loss": 1.2056,
+      "step": 4311
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012006651141305143,
+      "loss": 1.2409,
+      "step": 4312
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012003534073684645,
+      "loss": 1.1008,
+      "step": 4313
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012000416803244404,
+      "loss": 1.1868,
+      "step": 4314
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011997299330299985,
+      "loss": 1.1867,
+      "step": 4315
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001199418165516697,
+      "loss": 1.1092,
+      "step": 4316
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011991063778160967,
+      "loss": 1.1387,
+      "step": 4317
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011987945699597598,
+      "loss": 1.2056,
+      "step": 4318
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011984827419792517,
+      "loss": 1.1573,
+      "step": 4319
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011981708939061384,
+      "loss": 1.2028,
+      "step": 4320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011978590257719889,
+      "loss": 1.1198,
+      "step": 4321
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011975471376083736,
+      "loss": 1.1924,
+      "step": 4322
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011972352294468654,
+      "loss": 1.1651,
+      "step": 4323
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011969233013190393,
+      "loss": 1.1257,
+      "step": 4324
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011966113532564715,
+      "loss": 1.1837,
+      "step": 4325
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011962993852907416,
+      "loss": 1.1574,
+      "step": 4326
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011959873974534301,
+      "loss": 1.1978,
+      "step": 4327
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011956753897761199,
+      "loss": 1.1384,
+      "step": 4328
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011953633622903955,
+      "loss": 1.2271,
+      "step": 4329
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001195051315027844,
+      "loss": 1.145,
+      "step": 4330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011947392480200541,
+      "loss": 1.112,
+      "step": 4331
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011944271612986168,
+      "loss": 1.1542,
+      "step": 4332
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011941150548951252,
+      "loss": 1.143,
+      "step": 4333
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011938029288411736,
+      "loss": 1.1529,
+      "step": 4334
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011934907831683593,
+      "loss": 1.191,
+      "step": 4335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011931786179082811,
+      "loss": 1.114,
+      "step": 4336
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011928664330925394,
+      "loss": 1.1485,
+      "step": 4337
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011925542287527373,
+      "loss": 1.1723,
+      "step": 4338
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011922420049204793,
+      "loss": 1.1815,
+      "step": 4339
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011919297616273724,
+      "loss": 1.165,
+      "step": 4340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011916174989050252,
+      "loss": 1.165,
+      "step": 4341
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011913052167850483,
+      "loss": 1.2363,
+      "step": 4342
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011909929152990545,
+      "loss": 1.1935,
+      "step": 4343
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011906805944786581,
+      "loss": 1.1524,
+      "step": 4344
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011903682543554762,
+      "loss": 1.1702,
+      "step": 4345
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011900558949611263,
+      "loss": 1.1366,
+      "step": 4346
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011897435163272297,
+      "loss": 1.1535,
+      "step": 4347
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011894311184854088,
+      "loss": 1.2059,
+      "step": 4348
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011891187014672877,
+      "loss": 1.1449,
+      "step": 4349
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011888062653044927,
+      "loss": 1.1572,
+      "step": 4350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011884938100286518,
+      "loss": 1.1427,
+      "step": 4351
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011881813356713954,
+      "loss": 1.1497,
+      "step": 4352
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011878688422643556,
+      "loss": 1.1504,
+      "step": 4353
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011875563298391665,
+      "loss": 1.145,
+      "step": 4354
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011872437984274638,
+      "loss": 1.1792,
+      "step": 4355
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011869312480608856,
+      "loss": 1.1564,
+      "step": 4356
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011866186787710718,
+      "loss": 1.2084,
+      "step": 4357
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011863060905896637,
+      "loss": 1.179,
+      "step": 4358
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011859934835483049,
+      "loss": 1.242,
+      "step": 4359
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011856808576786412,
+      "loss": 1.1687,
+      "step": 4360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000118536821301232,
+      "loss": 1.1655,
+      "step": 4361
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011850555495809905,
+      "loss": 1.2737,
+      "step": 4362
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011847428674163042,
+      "loss": 1.1793,
+      "step": 4363
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011844301665499138,
+      "loss": 1.2169,
+      "step": 4364
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011841174470134747,
+      "loss": 1.1756,
+      "step": 4365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011838047088386435,
+      "loss": 1.2047,
+      "step": 4366
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011834919520570791,
+      "loss": 1.2175,
+      "step": 4367
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011831791767004423,
+      "loss": 1.1307,
+      "step": 4368
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011828663828003957,
+      "loss": 1.0902,
+      "step": 4369
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011825535703886038,
+      "loss": 1.1879,
+      "step": 4370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011822407394967325,
+      "loss": 1.1937,
+      "step": 4371
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011819278901564501,
+      "loss": 1.2247,
+      "step": 4372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011816150223994266,
+      "loss": 1.1841,
+      "step": 4373
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011813021362573347,
+      "loss": 1.1604,
+      "step": 4374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011809892317618471,
+      "loss": 1.1073,
+      "step": 4375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011806763089446402,
+      "loss": 1.1733,
+      "step": 4376
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011803633678373909,
+      "loss": 1.1656,
+      "step": 4377
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011800504084717791,
+      "loss": 1.1672,
+      "step": 4378
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011797374308794854,
+      "loss": 1.1726,
+      "step": 4379
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011794244350921933,
+      "loss": 1.1511,
+      "step": 4380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011791114211415873,
+      "loss": 1.1907,
+      "step": 4381
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011787983890593546,
+      "loss": 1.1618,
+      "step": 4382
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011784853388771835,
+      "loss": 1.149,
+      "step": 4383
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011781722706267641,
+      "loss": 1.2037,
+      "step": 4384
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001177859184339789,
+      "loss": 1.1634,
+      "step": 4385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001177546080047952,
+      "loss": 1.1767,
+      "step": 4386
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011772329577829491,
+      "loss": 1.2048,
+      "step": 4387
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001176919817576478,
+      "loss": 1.2376,
+      "step": 4388
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011766066594602381,
+      "loss": 1.177,
+      "step": 4389
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011762934834659307,
+      "loss": 1.1576,
+      "step": 4390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001175980289625259,
+      "loss": 1.1282,
+      "step": 4391
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011756670779699279,
+      "loss": 1.2112,
+      "step": 4392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001175353848531644,
+      "loss": 1.1029,
+      "step": 4393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011750406013421158,
+      "loss": 1.2162,
+      "step": 4394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011747273364330542,
+      "loss": 1.2149,
+      "step": 4395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011744140538361706,
+      "loss": 1.2094,
+      "step": 4396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001174100753583179,
+      "loss": 1.1519,
+      "step": 4397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011737874357057954,
+      "loss": 1.176,
+      "step": 4398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011734741002357374,
+      "loss": 1.2323,
+      "step": 4399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011731607472047239,
+      "loss": 1.1737,
+      "step": 4400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011728473766444756,
+      "loss": 1.1716,
+      "step": 4401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011725339885867162,
+      "loss": 1.1507,
+      "step": 4402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000117222058306317,
+      "loss": 1.1876,
+      "step": 4403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011719071601055629,
+      "loss": 1.2343,
+      "step": 4404
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011715937197456235,
+      "loss": 1.2498,
+      "step": 4405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011712802620150813,
+      "loss": 1.1454,
+      "step": 4406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011709667869456682,
+      "loss": 1.1527,
+      "step": 4407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011706532945691175,
+      "loss": 1.1401,
+      "step": 4408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011703397849171645,
+      "loss": 1.1395,
+      "step": 4409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011700262580215459,
+      "loss": 1.1136,
+      "step": 4410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011697127139140002,
+      "loss": 1.154,
+      "step": 4411
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011693991526262686,
+      "loss": 1.1935,
+      "step": 4412
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011690855741900921,
+      "loss": 1.1405,
+      "step": 4413
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011687719786372151,
+      "loss": 1.1496,
+      "step": 4414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001168458365999383,
+      "loss": 1.1899,
+      "step": 4415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011681447363083438,
+      "loss": 1.1106,
+      "step": 4416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011678310895958456,
+      "loss": 1.124,
+      "step": 4417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011675174258936396,
+      "loss": 1.178,
+      "step": 4418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011672037452334785,
+      "loss": 1.1844,
+      "step": 4419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011668900476471161,
+      "loss": 1.1341,
+      "step": 4420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011665763331663085,
+      "loss": 1.1423,
+      "step": 4421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011662626018228134,
+      "loss": 1.1976,
+      "step": 4422
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011659488536483901,
+      "loss": 1.2152,
+      "step": 4423
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011656350886747995,
+      "loss": 1.1418,
+      "step": 4424
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011653213069338045,
+      "loss": 1.2088,
+      "step": 4425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011650075084571694,
+      "loss": 1.1097,
+      "step": 4426
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011646936932766604,
+      "loss": 1.088,
+      "step": 4427
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011643798614240453,
+      "loss": 1.1692,
+      "step": 4428
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011640660129310936,
+      "loss": 1.2657,
+      "step": 4429
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011637521478295765,
+      "loss": 1.1804,
+      "step": 4430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011634382661512671,
+      "loss": 1.1841,
+      "step": 4431
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011631243679279397,
+      "loss": 1.1766,
+      "step": 4432
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011628104531913708,
+      "loss": 1.1739,
+      "step": 4433
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001162496521973338,
+      "loss": 1.1607,
+      "step": 4434
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011621825743056205,
+      "loss": 1.1847,
+      "step": 4435
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011618686102200002,
+      "loss": 1.1805,
+      "step": 4436
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011615546297482603,
+      "loss": 1.1338,
+      "step": 4437
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011612406329221843,
+      "loss": 1.1642,
+      "step": 4438
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001160926619773559,
+      "loss": 1.2104,
+      "step": 4439
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011606125903341725,
+      "loss": 1.2341,
+      "step": 4440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011602985446358136,
+      "loss": 1.1859,
+      "step": 4441
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011599844827102743,
+      "loss": 1.1787,
+      "step": 4442
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011596704045893467,
+      "loss": 1.1812,
+      "step": 4443
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011593563103048257,
+      "loss": 1.1468,
+      "step": 4444
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011590421998885075,
+      "loss": 1.1845,
+      "step": 4445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001158728073372189,
+      "loss": 1.2458,
+      "step": 4446
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011584139307876702,
+      "loss": 1.1841,
+      "step": 4447
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011580997721667522,
+      "loss": 1.1301,
+      "step": 4448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011577855975412368,
+      "loss": 1.1492,
+      "step": 4449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011574714069429291,
+      "loss": 1.19,
+      "step": 4450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011571572004036345,
+      "loss": 1.1971,
+      "step": 4451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011568429779551604,
+      "loss": 1.1868,
+      "step": 4452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011565287396293159,
+      "loss": 1.1179,
+      "step": 4453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011562144854579118,
+      "loss": 1.2133,
+      "step": 4454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011559002154727603,
+      "loss": 1.1702,
+      "step": 4455
+    },
+    {
+      "epoch": 1.79,
+      "eval_loss": 0.8235860466957092,
+      "eval_runtime": 502.3065,
+      "eval_samples_per_second": 51.562,
+      "eval_steps_per_second": 25.781,
+      "step": 4455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011555859297056748,
+      "loss": 1.0334,
+      "step": 4456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011552716281884714,
+      "loss": 1.2472,
+      "step": 4457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011549573109529672,
+      "loss": 1.1972,
+      "step": 4458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011546429780309804,
+      "loss": 1.0783,
+      "step": 4459
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011543286294543314,
+      "loss": 1.1355,
+      "step": 4460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001154014265254842,
+      "loss": 1.1295,
+      "step": 4461
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011536998854643357,
+      "loss": 1.1912,
+      "step": 4462
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011533854901146373,
+      "loss": 1.1887,
+      "step": 4463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011530710792375736,
+      "loss": 1.176,
+      "step": 4464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011527566528649725,
+      "loss": 1.148,
+      "step": 4465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011524422110286639,
+      "loss": 1.2109,
+      "step": 4466
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001152127753760479,
+      "loss": 1.1256,
+      "step": 4467
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011518132810922502,
+      "loss": 1.1734,
+      "step": 4468
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011514987930558123,
+      "loss": 1.1959,
+      "step": 4469
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011511842896830014,
+      "loss": 1.2015,
+      "step": 4470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011508697710056546,
+      "loss": 1.2069,
+      "step": 4471
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011505552370556112,
+      "loss": 1.172,
+      "step": 4472
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011502406878647116,
+      "loss": 1.1341,
+      "step": 4473
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001149926123464798,
+      "loss": 1.1615,
+      "step": 4474
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011496115438877145,
+      "loss": 1.1801,
+      "step": 4475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011492969491653051,
+      "loss": 1.1416,
+      "step": 4476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001148982339329418,
+      "loss": 1.1444,
+      "step": 4477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011486677144119004,
+      "loss": 1.2001,
+      "step": 4478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011483530744446027,
+      "loss": 1.1927,
+      "step": 4479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011480384194593759,
+      "loss": 1.191,
+      "step": 4480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001147723749488073,
+      "loss": 1.1814,
+      "step": 4481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011474090645625481,
+      "loss": 1.1549,
+      "step": 4482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011470943647146574,
+      "loss": 1.1523,
+      "step": 4483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011467796499762582,
+      "loss": 1.2058,
+      "step": 4484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011464649203792093,
+      "loss": 1.1653,
+      "step": 4485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011461501759553707,
+      "loss": 1.1551,
+      "step": 4486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011458354167366053,
+      "loss": 1.2443,
+      "step": 4487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011455206427547758,
+      "loss": 1.1276,
+      "step": 4488
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001145205854041747,
+      "loss": 1.1339,
+      "step": 4489
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001144891050629385,
+      "loss": 1.157,
+      "step": 4490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011445762325495588,
+      "loss": 1.1817,
+      "step": 4491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011442613998341367,
+      "loss": 1.1453,
+      "step": 4492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011439465525149899,
+      "loss": 1.194,
+      "step": 4493
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011436316906239908,
+      "loss": 1.2473,
+      "step": 4494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011433168141930129,
+      "loss": 1.1663,
+      "step": 4495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001143001923253932,
+      "loss": 1.2541,
+      "step": 4496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011426870178386237,
+      "loss": 1.1392,
+      "step": 4497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011423720979789676,
+      "loss": 1.1811,
+      "step": 4498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011420571637068422,
+      "loss": 1.1552,
+      "step": 4499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011417422150541296,
+      "loss": 1.1992,
+      "step": 4500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011414272520527117,
+      "loss": 1.1548,
+      "step": 4501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011411122747344724,
+      "loss": 1.1566,
+      "step": 4502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011407972831312977,
+      "loss": 1.1376,
+      "step": 4503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011404822772750743,
+      "loss": 1.1809,
+      "step": 4504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011401672571976903,
+      "loss": 1.1993,
+      "step": 4505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011398522229310358,
+      "loss": 1.1582,
+      "step": 4506
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011395371745070018,
+      "loss": 1.1984,
+      "step": 4507
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011392221119574813,
+      "loss": 1.1515,
+      "step": 4508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011389070353143682,
+      "loss": 1.1386,
+      "step": 4509
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011385919446095578,
+      "loss": 1.2109,
+      "step": 4510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011382768398749473,
+      "loss": 1.2174,
+      "step": 4511
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011379617211424352,
+      "loss": 1.1985,
+      "step": 4512
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011376465884439209,
+      "loss": 1.1684,
+      "step": 4513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001137331441811306,
+      "loss": 1.1576,
+      "step": 4514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011370162812764929,
+      "loss": 1.1219,
+      "step": 4515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011367011068713853,
+      "loss": 1.1358,
+      "step": 4516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011363859186278893,
+      "loss": 1.1112,
+      "step": 4517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001136070716577911,
+      "loss": 1.172,
+      "step": 4518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011357555007533593,
+      "loss": 1.1519,
+      "step": 4519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011354402711861434,
+      "loss": 1.1389,
+      "step": 4520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011351250279081744,
+      "loss": 1.1256,
+      "step": 4521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011348097709513648,
+      "loss": 1.1585,
+      "step": 4522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001134494500347628,
+      "loss": 1.1685,
+      "step": 4523
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011341792161288796,
+      "loss": 1.1216,
+      "step": 4524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011338639183270356,
+      "loss": 1.1869,
+      "step": 4525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011335486069740147,
+      "loss": 1.1746,
+      "step": 4526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011332332821017356,
+      "loss": 1.2654,
+      "step": 4527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011329179437421189,
+      "loss": 1.181,
+      "step": 4528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011326025919270869,
+      "loss": 1.1929,
+      "step": 4529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011322872266885631,
+      "loss": 1.1899,
+      "step": 4530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011319718480584716,
+      "loss": 1.1218,
+      "step": 4531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001131656456068739,
+      "loss": 1.2056,
+      "step": 4532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001131341050751293,
+      "loss": 1.1635,
+      "step": 4533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011310256321380616,
+      "loss": 1.1562,
+      "step": 4534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011307102002609758,
+      "loss": 1.1905,
+      "step": 4535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011303947551519665,
+      "loss": 1.1061,
+      "step": 4536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011300792968429665,
+      "loss": 1.2177,
+      "step": 4537
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011297638253659102,
+      "loss": 1.0965,
+      "step": 4538
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011294483407527332,
+      "loss": 1.1503,
+      "step": 4539
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011291328430353721,
+      "loss": 1.1976,
+      "step": 4540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011288173322457649,
+      "loss": 1.1814,
+      "step": 4541
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011285018084158516,
+      "loss": 1.1197,
+      "step": 4542
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011281862715775726,
+      "loss": 1.2561,
+      "step": 4543
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011278707217628699,
+      "loss": 1.1511,
+      "step": 4544
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011275551590036868,
+      "loss": 1.0791,
+      "step": 4545
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001127239583331969,
+      "loss": 1.1751,
+      "step": 4546
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011269239947796615,
+      "loss": 1.1926,
+      "step": 4547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011266083933787119,
+      "loss": 1.2182,
+      "step": 4548
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011262927791610694,
+      "loss": 1.1653,
+      "step": 4549
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011259771521586831,
+      "loss": 1.145,
+      "step": 4550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011256615124035048,
+      "loss": 1.1266,
+      "step": 4551
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011253458599274865,
+      "loss": 1.1446,
+      "step": 4552
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011250301947625826,
+      "loss": 1.2293,
+      "step": 4553
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001124714516940748,
+      "loss": 1.0998,
+      "step": 4554
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011243988264939391,
+      "loss": 1.0953,
+      "step": 4555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011240831234541133,
+      "loss": 1.1731,
+      "step": 4556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011237674078532299,
+      "loss": 1.1727,
+      "step": 4557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011234516797232489,
+      "loss": 1.1842,
+      "step": 4558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011231359390961317,
+      "loss": 1.1853,
+      "step": 4559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011228201860038412,
+      "loss": 1.1084,
+      "step": 4560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011225044204783412,
+      "loss": 1.193,
+      "step": 4561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011221886425515972,
+      "loss": 1.232,
+      "step": 4562
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001121872852255576,
+      "loss": 1.1323,
+      "step": 4563
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011215570496222445,
+      "loss": 1.1303,
+      "step": 4564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011212412346835723,
+      "loss": 1.173,
+      "step": 4565
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011209254074715294,
+      "loss": 1.1219,
+      "step": 4566
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001120609568018088,
+      "loss": 1.1711,
+      "step": 4567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011202937163552197,
+      "loss": 1.1339,
+      "step": 4568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011199778525148992,
+      "loss": 1.1603,
+      "step": 4569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011196619765291017,
+      "loss": 1.0776,
+      "step": 4570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011193460884298035,
+      "loss": 1.091,
+      "step": 4571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011190301882489824,
+      "loss": 1.1186,
+      "step": 4572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011187142760186169,
+      "loss": 1.191,
+      "step": 4573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011183983517706875,
+      "loss": 1.1386,
+      "step": 4574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011180824155371754,
+      "loss": 1.1628,
+      "step": 4575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011177664673500634,
+      "loss": 1.1438,
+      "step": 4576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011174505072413347,
+      "loss": 1.1114,
+      "step": 4577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011171345352429745,
+      "loss": 1.1509,
+      "step": 4578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011168185513869692,
+      "loss": 1.1704,
+      "step": 4579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011165025557053059,
+      "loss": 1.1881,
+      "step": 4580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011161865482299731,
+      "loss": 1.2601,
+      "step": 4581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001115870528992961,
+      "loss": 1.1757,
+      "step": 4582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011155544980262597,
+      "loss": 1.1553,
+      "step": 4583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011152384553618627,
+      "loss": 1.1498,
+      "step": 4584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011149224010317619,
+      "loss": 1.2232,
+      "step": 4585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011146063350679523,
+      "loss": 1.18,
+      "step": 4586
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011142902575024294,
+      "loss": 1.1896,
+      "step": 4587
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011139741683671908,
+      "loss": 1.2063,
+      "step": 4588
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011136580676942336,
+      "loss": 1.1993,
+      "step": 4589
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011133419555155576,
+      "loss": 1.0982,
+      "step": 4590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011130258318631628,
+      "loss": 1.0901,
+      "step": 4591
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001112709696769051,
+      "loss": 1.196,
+      "step": 4592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011123935502652245,
+      "loss": 1.1613,
+      "step": 4593
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011120773923836874,
+      "loss": 1.1379,
+      "step": 4594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011117612231564448,
+      "loss": 1.1847,
+      "step": 4595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011114450426155023,
+      "loss": 1.0737,
+      "step": 4596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011111288507928679,
+      "loss": 1.1285,
+      "step": 4597
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011108126477205491,
+      "loss": 1.1799,
+      "step": 4598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011104964334305563,
+      "loss": 1.2416,
+      "step": 4599
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011101802079548998,
+      "loss": 1.1788,
+      "step": 4600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011098639713255914,
+      "loss": 1.1886,
+      "step": 4601
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011095477235746441,
+      "loss": 1.1542,
+      "step": 4602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001109231464734072,
+      "loss": 1.1631,
+      "step": 4603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011089151948358902,
+      "loss": 1.1972,
+      "step": 4604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011085989139121155,
+      "loss": 1.1652,
+      "step": 4605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011082826219947646,
+      "loss": 1.1662,
+      "step": 4606
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011079663191158563,
+      "loss": 1.1732,
+      "step": 4607
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011076500053074103,
+      "loss": 1.1822,
+      "step": 4608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011073336806014478,
+      "loss": 1.1732,
+      "step": 4609
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011070173450299901,
+      "loss": 1.129,
+      "step": 4610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011067009986250601,
+      "loss": 1.2096,
+      "step": 4611
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011063846414186823,
+      "loss": 1.2011,
+      "step": 4612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011060682734428814,
+      "loss": 1.1521,
+      "step": 4613
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001105751894729684,
+      "loss": 1.1791,
+      "step": 4614
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011054355053111173,
+      "loss": 1.1885,
+      "step": 4615
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011051191052192098,
+      "loss": 1.106,
+      "step": 4616
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011048026944859908,
+      "loss": 1.1529,
+      "step": 4617
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011044862731434913,
+      "loss": 1.0913,
+      "step": 4618
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011041698412237424,
+      "loss": 1.1312,
+      "step": 4619
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001103853398758777,
+      "loss": 1.1524,
+      "step": 4620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001103536945780629,
+      "loss": 1.1471,
+      "step": 4621
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011032204823213333,
+      "loss": 1.1388,
+      "step": 4622
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011029040084129253,
+      "loss": 1.1459,
+      "step": 4623
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011025875240874428,
+      "loss": 1.1589,
+      "step": 4624
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011022710293769232,
+      "loss": 1.1614,
+      "step": 4625
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001101954524313406,
+      "loss": 1.1558,
+      "step": 4626
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011016380089289308,
+      "loss": 1.1933,
+      "step": 4627
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001101321483255539,
+      "loss": 1.1758,
+      "step": 4628
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001101004947325273,
+      "loss": 1.204,
+      "step": 4629
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011006884011701761,
+      "loss": 1.1972,
+      "step": 4630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011003718448222922,
+      "loss": 1.2052,
+      "step": 4631
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011000552783136668,
+      "loss": 1.1378,
+      "step": 4632
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010997387016763463,
+      "loss": 1.185,
+      "step": 4633
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001099422114942378,
+      "loss": 1.1557,
+      "step": 4634
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010991055181438101,
+      "loss": 1.1711,
+      "step": 4635
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010987889113126926,
+      "loss": 1.2056,
+      "step": 4636
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010984722944810752,
+      "loss": 1.163,
+      "step": 4637
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010981556676810099,
+      "loss": 1.1458,
+      "step": 4638
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010978390309445488,
+      "loss": 1.1962,
+      "step": 4639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010975223843037454,
+      "loss": 1.1808,
+      "step": 4640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010972057277906544,
+      "loss": 1.1217,
+      "step": 4641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010968890614373308,
+      "loss": 1.141,
+      "step": 4642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010965723852758317,
+      "loss": 1.1343,
+      "step": 4643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010962556993382138,
+      "loss": 1.1701,
+      "step": 4644
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010959390036565361,
+      "loss": 1.2342,
+      "step": 4645
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010956222982628576,
+      "loss": 1.1073,
+      "step": 4646
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010953055831892393,
+      "loss": 1.189,
+      "step": 4647
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010949888584677419,
+      "loss": 1.1929,
+      "step": 4648
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001094672124130428,
+      "loss": 1.1336,
+      "step": 4649
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010943553802093611,
+      "loss": 1.1066,
+      "step": 4650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010940386267366056,
+      "loss": 1.1026,
+      "step": 4651
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010937218637442266,
+      "loss": 1.182,
+      "step": 4652
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000109340509126429,
+      "loss": 1.1843,
+      "step": 4653
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010930883093288633,
+      "loss": 1.1228,
+      "step": 4654
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010927715179700146,
+      "loss": 1.128,
+      "step": 4655
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010924547172198132,
+      "loss": 1.1771,
+      "step": 4656
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010921379071103288,
+      "loss": 1.1388,
+      "step": 4657
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010918210876736325,
+      "loss": 1.21,
+      "step": 4658
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010915042589417967,
+      "loss": 1.1078,
+      "step": 4659
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010911874209468939,
+      "loss": 1.1709,
+      "step": 4660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010908705737209978,
+      "loss": 1.1365,
+      "step": 4661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010905537172961831,
+      "loss": 1.1135,
+      "step": 4662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010902368517045258,
+      "loss": 1.1425,
+      "step": 4663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010899199769781027,
+      "loss": 1.0656,
+      "step": 4664
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001089603093148991,
+      "loss": 1.1651,
+      "step": 4665
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010892862002492692,
+      "loss": 1.1326,
+      "step": 4666
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010889692983110165,
+      "loss": 1.1928,
+      "step": 4667
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010886523873663137,
+      "loss": 1.1287,
+      "step": 4668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010883354674472414,
+      "loss": 1.2157,
+      "step": 4669
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010880185385858821,
+      "loss": 1.0909,
+      "step": 4670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010877016008143188,
+      "loss": 1.1962,
+      "step": 4671
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010873846541646358,
+      "loss": 1.1587,
+      "step": 4672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001087067698668917,
+      "loss": 1.1617,
+      "step": 4673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010867507343592487,
+      "loss": 1.1742,
+      "step": 4674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010864337612677176,
+      "loss": 1.1934,
+      "step": 4675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010861167794264109,
+      "loss": 1.0953,
+      "step": 4676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010857997888674172,
+      "loss": 1.1389,
+      "step": 4677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010854827896228255,
+      "loss": 1.1861,
+      "step": 4678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010851657817247263,
+      "loss": 1.1726,
+      "step": 4679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010848487652052104,
+      "loss": 1.1632,
+      "step": 4680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010845317400963702,
+      "loss": 1.235,
+      "step": 4681
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010842147064302974,
+      "loss": 1.1542,
+      "step": 4682
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010838976642390864,
+      "loss": 1.1397,
+      "step": 4683
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010835806135548316,
+      "loss": 1.2575,
+      "step": 4684
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010832635544096286,
+      "loss": 1.1956,
+      "step": 4685
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010829464868355733,
+      "loss": 1.2271,
+      "step": 4686
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010826294108647626,
+      "loss": 1.1527,
+      "step": 4687
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010823123265292945,
+      "loss": 1.2253,
+      "step": 4688
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001081995233861268,
+      "loss": 1.1523,
+      "step": 4689
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010816781328927826,
+      "loss": 1.1209,
+      "step": 4690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010813610236559386,
+      "loss": 1.1324,
+      "step": 4691
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010810439061828375,
+      "loss": 1.1741,
+      "step": 4692
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010807267805055814,
+      "loss": 1.1393,
+      "step": 4693
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001080409646656273,
+      "loss": 1.1677,
+      "step": 4694
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010800925046670163,
+      "loss": 1.1929,
+      "step": 4695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010797753545699158,
+      "loss": 1.1691,
+      "step": 4696
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001079458196397077,
+      "loss": 1.1723,
+      "step": 4697
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010791410301806062,
+      "loss": 1.16,
+      "step": 4698
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010788238559526102,
+      "loss": 1.1612,
+      "step": 4699
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001078506673745197,
+      "loss": 1.1846,
+      "step": 4700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010781894835904753,
+      "loss": 1.203,
+      "step": 4701
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010778722855205547,
+      "loss": 1.1524,
+      "step": 4702
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010775550795675451,
+      "loss": 1.1916,
+      "step": 4703
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001077237865763558,
+      "loss": 1.1749,
+      "step": 4704
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001076920644140705,
+      "loss": 1.2064,
+      "step": 4705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010766034147310989,
+      "loss": 1.1217,
+      "step": 4706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010762861775668529,
+      "loss": 1.151,
+      "step": 4707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010759689326800815,
+      "loss": 1.1348,
+      "step": 4708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010756516801028997,
+      "loss": 1.1602,
+      "step": 4709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001075334419867423,
+      "loss": 1.2023,
+      "step": 4710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010750171520057683,
+      "loss": 1.198,
+      "step": 4711
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010746998765500529,
+      "loss": 1.1939,
+      "step": 4712
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010743825935323948,
+      "loss": 1.1923,
+      "step": 4713
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010740653029849133,
+      "loss": 1.1925,
+      "step": 4714
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010737480049397271,
+      "loss": 1.2467,
+      "step": 4715
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010734306994289573,
+      "loss": 1.2516,
+      "step": 4716
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010731133864847247,
+      "loss": 1.1652,
+      "step": 4717
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010727960661391519,
+      "loss": 1.1843,
+      "step": 4718
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010724787384243606,
+      "loss": 1.1466,
+      "step": 4719
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010721614033724749,
+      "loss": 1.2262,
+      "step": 4720
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010718440610156187,
+      "loss": 1.0852,
+      "step": 4721
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010715267113859166,
+      "loss": 1.156,
+      "step": 4722
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010712093545154949,
+      "loss": 1.17,
+      "step": 4723
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010708919904364791,
+      "loss": 1.1548,
+      "step": 4724
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001070574619180997,
+      "loss": 1.1818,
+      "step": 4725
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010702572407811761,
+      "loss": 1.0997,
+      "step": 4726
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010699398552691454,
+      "loss": 1.2037,
+      "step": 4727
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001069622462677033,
+      "loss": 1.2155,
+      "step": 4728
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010693050630369699,
+      "loss": 1.2315,
+      "step": 4729
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010689876563810865,
+      "loss": 1.1712,
+      "step": 4730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001068670242741514,
+      "loss": 1.205,
+      "step": 4731
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010683528221503846,
+      "loss": 1.1141,
+      "step": 4732
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010680353946398316,
+      "loss": 1.1738,
+      "step": 4733
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010677179602419876,
+      "loss": 1.1894,
+      "step": 4734
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010674005189889878,
+      "loss": 1.1148,
+      "step": 4735
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001067083070912966,
+      "loss": 1.1584,
+      "step": 4736
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010667656160460584,
+      "loss": 1.1298,
+      "step": 4737
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010664481544204012,
+      "loss": 1.1531,
+      "step": 4738
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010661306860681317,
+      "loss": 1.1636,
+      "step": 4739
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001065813211021387,
+      "loss": 1.1838,
+      "step": 4740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010654957293123056,
+      "loss": 1.1911,
+      "step": 4741
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010651782409730264,
+      "loss": 1.2058,
+      "step": 4742
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010648607460356892,
+      "loss": 1.1695,
+      "step": 4743
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010645432445324345,
+      "loss": 1.1731,
+      "step": 4744
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010642257364954027,
+      "loss": 1.1748,
+      "step": 4745
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001063908221956736,
+      "loss": 1.1627,
+      "step": 4746
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010635907009485766,
+      "loss": 1.23,
+      "step": 4747
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010632731735030676,
+      "loss": 1.1031,
+      "step": 4748
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010629556396523521,
+      "loss": 1.2402,
+      "step": 4749
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001062638099428575,
+      "loss": 1.1852,
+      "step": 4750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010623205528638807,
+      "loss": 1.1567,
+      "step": 4751
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010620029999904148,
+      "loss": 1.1484,
+      "step": 4752
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010616854408403239,
+      "loss": 1.2339,
+      "step": 4753
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010613678754457545,
+      "loss": 1.1386,
+      "step": 4754
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010610503038388542,
+      "loss": 1.1234,
+      "step": 4755
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001060732726051771,
+      "loss": 1.2129,
+      "step": 4756
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010604151421166539,
+      "loss": 1.1988,
+      "step": 4757
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010600975520656515,
+      "loss": 1.1584,
+      "step": 4758
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010597799559309145,
+      "loss": 1.1561,
+      "step": 4759
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010594623537445934,
+      "loss": 1.1981,
+      "step": 4760
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001059144745538839,
+      "loss": 1.1283,
+      "step": 4761
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010588271313458033,
+      "loss": 1.1038,
+      "step": 4762
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010585095111976386,
+      "loss": 1.1255,
+      "step": 4763
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010581918851264982,
+      "loss": 1.1215,
+      "step": 4764
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010578742531645353,
+      "loss": 1.2145,
+      "step": 4765
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010575566153439045,
+      "loss": 1.1896,
+      "step": 4766
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010572389716967601,
+      "loss": 1.0955,
+      "step": 4767
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010569213222552581,
+      "loss": 1.1072,
+      "step": 4768
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001056603667051554,
+      "loss": 1.1447,
+      "step": 4769
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010562860061178046,
+      "loss": 1.1318,
+      "step": 4770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010559683394861668,
+      "loss": 1.0809,
+      "step": 4771
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010556506671887981,
+      "loss": 1.141,
+      "step": 4772
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010553329892578576,
+      "loss": 1.1619,
+      "step": 4773
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010550153057255035,
+      "loss": 1.1143,
+      "step": 4774
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010546976166238953,
+      "loss": 1.1193,
+      "step": 4775
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010543799219851934,
+      "loss": 1.244,
+      "step": 4776
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010540622218415577,
+      "loss": 1.217,
+      "step": 4777
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010537445162251498,
+      "loss": 1.1434,
+      "step": 4778
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010534268051681309,
+      "loss": 1.1677,
+      "step": 4779
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010531090887026635,
+      "loss": 1.1044,
+      "step": 4780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010527913668609107,
+      "loss": 1.1093,
+      "step": 4781
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010524736396750352,
+      "loss": 1.1314,
+      "step": 4782
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010521559071772009,
+      "loss": 1.2454,
+      "step": 4783
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010518381693995725,
+      "loss": 1.1569,
+      "step": 4784
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010515204263743145,
+      "loss": 1.1495,
+      "step": 4785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010512026781335928,
+      "loss": 1.1247,
+      "step": 4786
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010508849247095733,
+      "loss": 1.1519,
+      "step": 4787
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010505671661344223,
+      "loss": 1.162,
+      "step": 4788
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010502494024403067,
+      "loss": 1.1907,
+      "step": 4789
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010499316336593944,
+      "loss": 1.2138,
+      "step": 4790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010496138598238532,
+      "loss": 1.0945,
+      "step": 4791
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010492960809658517,
+      "loss": 1.1061,
+      "step": 4792
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010489782971175585,
+      "loss": 1.1473,
+      "step": 4793
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010486605083111445,
+      "loss": 1.1965,
+      "step": 4794
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010483427145787786,
+      "loss": 1.1573,
+      "step": 4795
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010480249159526316,
+      "loss": 1.1662,
+      "step": 4796
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010477071124648748,
+      "loss": 1.1243,
+      "step": 4797
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010473893041476795,
+      "loss": 1.1694,
+      "step": 4798
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010470714910332182,
+      "loss": 1.092,
+      "step": 4799
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010467536731536627,
+      "loss": 1.1753,
+      "step": 4800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010464358505411864,
+      "loss": 1.1675,
+      "step": 4801
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010461180232279635,
+      "loss": 1.1742,
+      "step": 4802
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010458001912461668,
+      "loss": 1.1481,
+      "step": 4803
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010454823546279714,
+      "loss": 1.1386,
+      "step": 4804
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010451645134055521,
+      "loss": 1.2371,
+      "step": 4805
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010448466676110841,
+      "loss": 1.1492,
+      "step": 4806
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010445288172767438,
+      "loss": 1.1695,
+      "step": 4807
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010442109624347067,
+      "loss": 1.0888,
+      "step": 4808
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010438931031171501,
+      "loss": 1.1255,
+      "step": 4809
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010435752393562513,
+      "loss": 1.1634,
+      "step": 4810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010432573711841878,
+      "loss": 1.1555,
+      "step": 4811
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010429394986331375,
+      "loss": 1.1906,
+      "step": 4812
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010426216217352789,
+      "loss": 1.1716,
+      "step": 4813
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010423037405227915,
+      "loss": 1.1059,
+      "step": 4814
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010419858550278545,
+      "loss": 1.1233,
+      "step": 4815
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010416679652826478,
+      "loss": 1.1738,
+      "step": 4816
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010413500713193513,
+      "loss": 1.1349,
+      "step": 4817
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001041032173170146,
+      "loss": 1.1119,
+      "step": 4818
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010407142708672134,
+      "loss": 1.2107,
+      "step": 4819
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010403963644427346,
+      "loss": 1.134,
+      "step": 4820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010400784539288915,
+      "loss": 1.1317,
+      "step": 4821
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010397605393578669,
+      "loss": 1.1083,
+      "step": 4822
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010394426207618436,
+      "loss": 1.1855,
+      "step": 4823
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010391246981730045,
+      "loss": 1.1681,
+      "step": 4824
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010388067716235335,
+      "loss": 1.1776,
+      "step": 4825
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010384888411456141,
+      "loss": 1.1345,
+      "step": 4826
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010381709067714313,
+      "loss": 1.2041,
+      "step": 4827
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010378529685331698,
+      "loss": 1.1639,
+      "step": 4828
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010375350264630146,
+      "loss": 1.1344,
+      "step": 4829
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010372170805931514,
+      "loss": 1.1561,
+      "step": 4830
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010368991309557662,
+      "loss": 1.1627,
+      "step": 4831
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010365811775830456,
+      "loss": 1.1306,
+      "step": 4832
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010362632205071761,
+      "loss": 1.1897,
+      "step": 4833
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010359452597603441,
+      "loss": 1.137,
+      "step": 4834
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010356272953747383,
+      "loss": 1.1693,
+      "step": 4835
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001035309327382546,
+      "loss": 1.1456,
+      "step": 4836
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010349913558159554,
+      "loss": 1.1219,
+      "step": 4837
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010346733807071551,
+      "loss": 1.1709,
+      "step": 4838
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010343554020883341,
+      "loss": 1.2165,
+      "step": 4839
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010340374199916816,
+      "loss": 1.1723,
+      "step": 4840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010337194344493872,
+      "loss": 1.2116,
+      "step": 4841
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010334014454936409,
+      "loss": 1.1788,
+      "step": 4842
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010330834531566333,
+      "loss": 1.1652,
+      "step": 4843
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010327654574705548,
+      "loss": 1.123,
+      "step": 4844
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010324474584675967,
+      "loss": 1.1627,
+      "step": 4845
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010321294561799501,
+      "loss": 1.1519,
+      "step": 4846
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010318114506398067,
+      "loss": 1.1574,
+      "step": 4847
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010314934418793585,
+      "loss": 1.1397,
+      "step": 4848
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010311754299307979,
+      "loss": 1.1611,
+      "step": 4849
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010308574148263177,
+      "loss": 1.1763,
+      "step": 4850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010305393965981107,
+      "loss": 1.1806,
+      "step": 4851
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010302213752783702,
+      "loss": 1.1124,
+      "step": 4852
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010299033508992902,
+      "loss": 1.1593,
+      "step": 4853
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010295853234930639,
+      "loss": 1.1895,
+      "step": 4854
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010292672930918859,
+      "loss": 1.2276,
+      "step": 4855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010289492597279508,
+      "loss": 1.1995,
+      "step": 4856
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010286312234334538,
+      "loss": 1.1157,
+      "step": 4857
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010283131842405894,
+      "loss": 1.1842,
+      "step": 4858
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010279951421815532,
+      "loss": 1.1898,
+      "step": 4859
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001027677097288541,
+      "loss": 1.2379,
+      "step": 4860
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010273590495937486,
+      "loss": 1.1735,
+      "step": 4861
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010270409991293726,
+      "loss": 1.1743,
+      "step": 4862
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010267229459276093,
+      "loss": 1.1546,
+      "step": 4863
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010264048900206557,
+      "loss": 1.186,
+      "step": 4864
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010260868314407091,
+      "loss": 1.1763,
+      "step": 4865
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010257687702199668,
+      "loss": 1.1056,
+      "step": 4866
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010254507063906258,
+      "loss": 1.1859,
+      "step": 4867
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010251326399848846,
+      "loss": 1.09,
+      "step": 4868
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010248145710349414,
+      "loss": 1.1991,
+      "step": 4869
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010244964995729947,
+      "loss": 1.2171,
+      "step": 4870
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010241784256312428,
+      "loss": 1.1403,
+      "step": 4871
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001023860349241885,
+      "loss": 1.1632,
+      "step": 4872
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010235422704371203,
+      "loss": 1.235,
+      "step": 4873
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010232241892491483,
+      "loss": 1.1479,
+      "step": 4874
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010229061057101683,
+      "loss": 1.209,
+      "step": 4875
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010225880198523808,
+      "loss": 1.2152,
+      "step": 4876
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010222699317079854,
+      "loss": 1.1579,
+      "step": 4877
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010219518413091828,
+      "loss": 1.1682,
+      "step": 4878
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010216337486881735,
+      "loss": 1.1692,
+      "step": 4879
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010213156538771584,
+      "loss": 1.1524,
+      "step": 4880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010209975569083386,
+      "loss": 1.1465,
+      "step": 4881
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010206794578139149,
+      "loss": 1.179,
+      "step": 4882
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010203613566260895,
+      "loss": 1.1434,
+      "step": 4883
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010200432533770634,
+      "loss": 1.1875,
+      "step": 4884
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010197251480990391,
+      "loss": 1.2079,
+      "step": 4885
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010194070408242184,
+      "loss": 1.1772,
+      "step": 4886
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001019088931584804,
+      "loss": 1.1487,
+      "step": 4887
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010187708204129978,
+      "loss": 1.1809,
+      "step": 4888
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010184527073410026,
+      "loss": 1.1985,
+      "step": 4889
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001018134592401022,
+      "loss": 1.2113,
+      "step": 4890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010178164756252584,
+      "loss": 1.1842,
+      "step": 4891
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010174983570459154,
+      "loss": 1.1708,
+      "step": 4892
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010171802366951961,
+      "loss": 1.1957,
+      "step": 4893
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010168621146053046,
+      "loss": 1.149,
+      "step": 4894
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010165439908084445,
+      "loss": 1.2186,
+      "step": 4895
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010162258653368196,
+      "loss": 1.2037,
+      "step": 4896
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010159077382226341,
+      "loss": 1.0833,
+      "step": 4897
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010155896094980927,
+      "loss": 1.1992,
+      "step": 4898
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010152714791953998,
+      "loss": 1.2734,
+      "step": 4899
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010149533473467599,
+      "loss": 1.2147,
+      "step": 4900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010146352139843776,
+      "loss": 1.2122,
+      "step": 4901
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001014317079140458,
+      "loss": 1.0903,
+      "step": 4902
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010139989428472063,
+      "loss": 1.1735,
+      "step": 4903
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010136808051368278,
+      "loss": 1.096,
+      "step": 4904
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010133626660415278,
+      "loss": 1.0908,
+      "step": 4905
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010130445255935119,
+      "loss": 1.1679,
+      "step": 4906
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010127263838249856,
+      "loss": 1.161,
+      "step": 4907
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010124082407681552,
+      "loss": 1.203,
+      "step": 4908
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010120900964552261,
+      "loss": 1.1303,
+      "step": 4909
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001011771950918404,
+      "loss": 1.2246,
+      "step": 4910
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001011453804189896,
+      "loss": 1.1705,
+      "step": 4911
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010111356563019083,
+      "loss": 1.1637,
+      "step": 4912
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010108175072866468,
+      "loss": 1.1606,
+      "step": 4913
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010104993571763184,
+      "loss": 1.206,
+      "step": 4914
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010101812060031295,
+      "loss": 1.124,
+      "step": 4915
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010098630537992873,
+      "loss": 1.1312,
+      "step": 4916
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001009544900596998,
+      "loss": 1.162,
+      "step": 4917
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001009226746428469,
+      "loss": 1.1421,
+      "step": 4918
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010089085913259073,
+      "loss": 1.1587,
+      "step": 4919
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010085904353215204,
+      "loss": 1.1588,
+      "step": 4920
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010082722784475149,
+      "loss": 1.17,
+      "step": 4921
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010079541207360985,
+      "loss": 1.1872,
+      "step": 4922
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010076359622194784,
+      "loss": 1.144,
+      "step": 4923
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010073178029298624,
+      "loss": 1.1788,
+      "step": 4924
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010069996428994582,
+      "loss": 1.1542,
+      "step": 4925
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010066814821604728,
+      "loss": 1.1234,
+      "step": 4926
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010063633207451145,
+      "loss": 1.228,
+      "step": 4927
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010060451586855908,
+      "loss": 1.1485,
+      "step": 4928
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010057269960141101,
+      "loss": 1.1341,
+      "step": 4929
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010054088327628795,
+      "loss": 1.1974,
+      "step": 4930
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010050906689641073,
+      "loss": 1.1711,
+      "step": 4931
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010047725046500019,
+      "loss": 1.1806,
+      "step": 4932
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010044543398527712,
+      "loss": 1.2063,
+      "step": 4933
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010041361746046232,
+      "loss": 1.1358,
+      "step": 4934
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010038180089377658,
+      "loss": 1.1647,
+      "step": 4935
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010034998428844078,
+      "loss": 1.1174,
+      "step": 4936
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010031816764767571,
+      "loss": 1.2264,
+      "step": 4937
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010028635097470224,
+      "loss": 1.1777,
+      "step": 4938
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010025453427274117,
+      "loss": 1.1323,
+      "step": 4939
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010022271754501333,
+      "loss": 1.0498,
+      "step": 4940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010019090079473958,
+      "loss": 1.2014,
+      "step": 4941
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010015908402514076,
+      "loss": 1.1,
+      "step": 4942
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001001272672394377,
+      "loss": 1.1835,
+      "step": 4943
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010009545044085123,
+      "loss": 1.1729,
+      "step": 4944
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010006363363260225,
+      "loss": 1.1576,
+      "step": 4945
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010003181681791156,
+      "loss": 1.1426,
+      "step": 4946
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001,
+      "loss": 1.1452,
+      "step": 4947
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.996818318208846e-05,
+      "loss": 1.1399,
+      "step": 4948
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.993636636739778e-05,
+      "loss": 1.1441,
+      "step": 4949
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.99045495591488e-05,
+      "loss": 1.2532,
+      "step": 4950
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.8188599944114685,
+      "eval_runtime": 503.913,
+      "eval_samples_per_second": 51.398,
+      "eval_steps_per_second": 25.699,
+      "step": 4950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.987273276056235e-05,
+      "loss": 1.2062,
+      "step": 4951
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.984091597485928e-05,
+      "loss": 1.1609,
+      "step": 4952
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.980909920526045e-05,
+      "loss": 1.1107,
+      "step": 4953
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.977728245498668e-05,
+      "loss": 1.1477,
+      "step": 4954
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.974546572725883e-05,
+      "loss": 1.1704,
+      "step": 4955
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.971364902529776e-05,
+      "loss": 1.1663,
+      "step": 4956
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.968183235232427e-05,
+      "loss": 1.1957,
+      "step": 4957
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965001571155923e-05,
+      "loss": 1.206,
+      "step": 4958
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.961819910622343e-05,
+      "loss": 1.1902,
+      "step": 4959
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.958638253953772e-05,
+      "loss": 1.1557,
+      "step": 4960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.95545660147229e-05,
+      "loss": 1.1323,
+      "step": 4961
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.952274953499984e-05,
+      "loss": 1.1455,
+      "step": 4962
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.94909331035893e-05,
+      "loss": 1.1703,
+      "step": 4963
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.945911672371209e-05,
+      "loss": 1.1778,
+      "step": 4964
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.942730039858903e-05,
+      "loss": 1.1721,
+      "step": 4965
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.939548413144093e-05,
+      "loss": 1.0917,
+      "step": 4966
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.936366792548859e-05,
+      "loss": 1.1812,
+      "step": 4967
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.933185178395271e-05,
+      "loss": 1.1722,
+      "step": 4968
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.930003571005419e-05,
+      "loss": 1.2138,
+      "step": 4969
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.926821970701377e-05,
+      "loss": 1.1349,
+      "step": 4970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.923640377805217e-05,
+      "loss": 1.1644,
+      "step": 4971
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.920458792639017e-05,
+      "loss": 1.2009,
+      "step": 4972
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.917277215524853e-05,
+      "loss": 1.2042,
+      "step": 4973
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.914095646784797e-05,
+      "loss": 1.1502,
+      "step": 4974
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.910914086740929e-05,
+      "loss": 1.1145,
+      "step": 4975
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.907732535715313e-05,
+      "loss": 1.1595,
+      "step": 4976
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.904550994030023e-05,
+      "loss": 1.2335,
+      "step": 4977
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.901369462007132e-05,
+      "loss": 1.1899,
+      "step": 4978
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.898187939968707e-05,
+      "loss": 1.1557,
+      "step": 4979
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.89500642823682e-05,
+      "loss": 1.1278,
+      "step": 4980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.891824927133536e-05,
+      "loss": 1.2124,
+      "step": 4981
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.888643436980919e-05,
+      "loss": 1.2213,
+      "step": 4982
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.88546195810104e-05,
+      "loss": 1.1043,
+      "step": 4983
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.88228049081596e-05,
+      "loss": 1.1974,
+      "step": 4984
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.879099035447743e-05,
+      "loss": 1.1075,
+      "step": 4985
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.87591759231845e-05,
+      "loss": 1.2136,
+      "step": 4986
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.872736161750145e-05,
+      "loss": 1.1008,
+      "step": 4987
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.869554744064883e-05,
+      "loss": 1.1715,
+      "step": 4988
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.866373339584724e-05,
+      "loss": 1.1706,
+      "step": 4989
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.863191948631724e-05,
+      "loss": 1.1609,
+      "step": 4990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.860010571527938e-05,
+      "loss": 1.1866,
+      "step": 4991
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.856829208595424e-05,
+      "loss": 1.1788,
+      "step": 4992
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.853647860156229e-05,
+      "loss": 1.1904,
+      "step": 4993
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.850466526532406e-05,
+      "loss": 1.1556,
+      "step": 4994
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.847285208046005e-05,
+      "loss": 1.1453,
+      "step": 4995
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.844103905019074e-05,
+      "loss": 1.1516,
+      "step": 4996
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.840922617773659e-05,
+      "loss": 1.0712,
+      "step": 4997
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.837741346631805e-05,
+      "loss": 1.0972,
+      "step": 4998
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.834560091915558e-05,
+      "loss": 1.0995,
+      "step": 4999
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.831378853946958e-05,
+      "loss": 1.1481,
+      "step": 5000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.828197633048041e-05,
+      "loss": 1.1246,
+      "step": 5001
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.82501642954085e-05,
+      "loss": 1.1227,
+      "step": 5002
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.821835243747418e-05,
+      "loss": 1.0886,
+      "step": 5003
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.818654075989784e-05,
+      "loss": 1.153,
+      "step": 5004
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.815472926589975e-05,
+      "loss": 1.0791,
+      "step": 5005
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.812291795870026e-05,
+      "loss": 1.1261,
+      "step": 5006
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.809110684151965e-05,
+      "loss": 1.1035,
+      "step": 5007
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.805929591757818e-05,
+      "loss": 1.0751,
+      "step": 5008
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.802748519009608e-05,
+      "loss": 1.1315,
+      "step": 5009
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.799567466229364e-05,
+      "loss": 1.1078,
+      "step": 5010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.796386433739106e-05,
+      "loss": 1.1171,
+      "step": 5011
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.793205421860852e-05,
+      "loss": 1.0869,
+      "step": 5012
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.790024430916618e-05,
+      "loss": 1.1217,
+      "step": 5013
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.786843461228418e-05,
+      "loss": 1.1504,
+      "step": 5014
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.783662513118266e-05,
+      "loss": 1.1311,
+      "step": 5015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.780481586908173e-05,
+      "loss": 1.0378,
+      "step": 5016
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.777300682920148e-05,
+      "loss": 1.149,
+      "step": 5017
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.774119801476195e-05,
+      "loss": 1.1712,
+      "step": 5018
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.770938942898318e-05,
+      "loss": 1.1351,
+      "step": 5019
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.76775810750852e-05,
+      "loss": 1.1398,
+      "step": 5020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.7645772956288e-05,
+      "loss": 1.1398,
+      "step": 5021
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.761396507581154e-05,
+      "loss": 1.129,
+      "step": 5022
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.758215743687571e-05,
+      "loss": 1.1084,
+      "step": 5023
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.755035004270053e-05,
+      "loss": 1.1065,
+      "step": 5024
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.751854289650586e-05,
+      "loss": 1.0635,
+      "step": 5025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.748673600151155e-05,
+      "loss": 1.1372,
+      "step": 5026
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.745492936093744e-05,
+      "loss": 1.1012,
+      "step": 5027
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.742312297800336e-05,
+      "loss": 1.1532,
+      "step": 5028
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.739131685592911e-05,
+      "loss": 1.0771,
+      "step": 5029
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.735951099793444e-05,
+      "loss": 1.1022,
+      "step": 5030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.732770540723908e-05,
+      "loss": 1.1239,
+      "step": 5031
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.729590008706277e-05,
+      "loss": 1.1383,
+      "step": 5032
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.726409504062516e-05,
+      "loss": 1.053,
+      "step": 5033
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.723229027114595e-05,
+      "loss": 1.0704,
+      "step": 5034
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.720048578184473e-05,
+      "loss": 1.0948,
+      "step": 5035
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.716868157594111e-05,
+      "loss": 1.1034,
+      "step": 5036
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.713687765665464e-05,
+      "loss": 1.175,
+      "step": 5037
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.710507402720491e-05,
+      "loss": 1.1028,
+      "step": 5038
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.707327069081141e-05,
+      "loss": 1.1584,
+      "step": 5039
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.704146765069362e-05,
+      "loss": 1.1051,
+      "step": 5040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.7009664910071e-05,
+      "loss": 1.104,
+      "step": 5041
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.6977862472163e-05,
+      "loss": 1.0822,
+      "step": 5042
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.694606034018896e-05,
+      "loss": 1.1455,
+      "step": 5043
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.691425851736827e-05,
+      "loss": 1.1361,
+      "step": 5044
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.688245700692023e-05,
+      "loss": 1.1047,
+      "step": 5045
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.685065581206417e-05,
+      "loss": 1.0492,
+      "step": 5046
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.681885493601938e-05,
+      "loss": 1.1819,
+      "step": 5047
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.678705438200504e-05,
+      "loss": 1.0659,
+      "step": 5048
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.675525415324036e-05,
+      "loss": 1.1415,
+      "step": 5049
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.672345425294453e-05,
+      "loss": 1.099,
+      "step": 5050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.669165468433668e-05,
+      "loss": 1.1113,
+      "step": 5051
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.66598554506359e-05,
+      "loss": 1.1908,
+      "step": 5052
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.662805655506129e-05,
+      "loss": 1.17,
+      "step": 5053
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.659625800083187e-05,
+      "loss": 1.1613,
+      "step": 5054
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.656445979116661e-05,
+      "loss": 1.1027,
+      "step": 5055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.653266192928451e-05,
+      "loss": 1.1444,
+      "step": 5056
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.650086441840448e-05,
+      "loss": 1.0709,
+      "step": 5057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.646906726174542e-05,
+      "loss": 1.1267,
+      "step": 5058
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.64372704625262e-05,
+      "loss": 1.1291,
+      "step": 5059
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.64054740239656e-05,
+      "loss": 1.1336,
+      "step": 5060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.637367794928245e-05,
+      "loss": 1.136,
+      "step": 5061
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.634188224169548e-05,
+      "loss": 1.0705,
+      "step": 5062
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.631008690442339e-05,
+      "loss": 1.1488,
+      "step": 5063
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.627829194068485e-05,
+      "loss": 1.1265,
+      "step": 5064
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.624649735369854e-05,
+      "loss": 1.0729,
+      "step": 5065
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.621470314668302e-05,
+      "loss": 1.1878,
+      "step": 5066
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.618290932285689e-05,
+      "loss": 1.0505,
+      "step": 5067
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.615111588543861e-05,
+      "loss": 1.1381,
+      "step": 5068
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.611932283764669e-05,
+      "loss": 1.1484,
+      "step": 5069
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.608753018269957e-05,
+      "loss": 1.1168,
+      "step": 5070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.605573792381568e-05,
+      "loss": 1.0912,
+      "step": 5071
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.602394606421334e-05,
+      "loss": 1.1527,
+      "step": 5072
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.599215460711086e-05,
+      "loss": 1.1094,
+      "step": 5073
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.596036355572658e-05,
+      "loss": 1.1719,
+      "step": 5074
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.592857291327869e-05,
+      "loss": 1.0909,
+      "step": 5075
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.589678268298543e-05,
+      "loss": 1.1536,
+      "step": 5076
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.586499286806492e-05,
+      "loss": 1.0166,
+      "step": 5077
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.583320347173523e-05,
+      "loss": 1.1142,
+      "step": 5078
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.580141449721456e-05,
+      "loss": 1.1133,
+      "step": 5079
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.576962594772086e-05,
+      "loss": 1.1103,
+      "step": 5080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.573783782647211e-05,
+      "loss": 1.1377,
+      "step": 5081
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.570605013668626e-05,
+      "loss": 1.051,
+      "step": 5082
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.567426288158123e-05,
+      "loss": 1.1488,
+      "step": 5083
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.564247606437488e-05,
+      "loss": 1.1406,
+      "step": 5084
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.5610689688285e-05,
+      "loss": 1.068,
+      "step": 5085
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.557890375652934e-05,
+      "loss": 1.1093,
+      "step": 5086
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.554711827232565e-05,
+      "loss": 1.137,
+      "step": 5087
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.55153332388916e-05,
+      "loss": 1.119,
+      "step": 5088
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.548354865944483e-05,
+      "loss": 1.1032,
+      "step": 5089
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.54517645372029e-05,
+      "loss": 1.0367,
+      "step": 5090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.541998087538335e-05,
+      "loss": 1.1583,
+      "step": 5091
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.538819767720368e-05,
+      "loss": 1.1164,
+      "step": 5092
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.535641494588134e-05,
+      "loss": 1.1585,
+      "step": 5093
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.532463268463374e-05,
+      "loss": 1.0972,
+      "step": 5094
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.529285089667821e-05,
+      "loss": 1.0598,
+      "step": 5095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.526106958523208e-05,
+      "loss": 1.0196,
+      "step": 5096
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.522928875351256e-05,
+      "loss": 1.1329,
+      "step": 5097
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.519750840473687e-05,
+      "loss": 1.1107,
+      "step": 5098
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.516572854212217e-05,
+      "loss": 1.1074,
+      "step": 5099
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.513394916888558e-05,
+      "loss": 1.1196,
+      "step": 5100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.510217028824416e-05,
+      "loss": 1.134,
+      "step": 5101
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.507039190341488e-05,
+      "loss": 1.079,
+      "step": 5102
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.503861401761473e-05,
+      "loss": 1.0876,
+      "step": 5103
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.500683663406059e-05,
+      "loss": 1.026,
+      "step": 5104
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.497505975596936e-05,
+      "loss": 1.0687,
+      "step": 5105
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.494328338655778e-05,
+      "loss": 1.0714,
+      "step": 5106
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.491150752904268e-05,
+      "loss": 1.1233,
+      "step": 5107
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.48797321866407e-05,
+      "loss": 1.1252,
+      "step": 5108
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.484795736256856e-05,
+      "loss": 1.061,
+      "step": 5109
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.481618306004277e-05,
+      "loss": 1.1115,
+      "step": 5110
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.478440928227993e-05,
+      "loss": 1.104,
+      "step": 5111
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.475263603249652e-05,
+      "loss": 1.0884,
+      "step": 5112
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.472086331390896e-05,
+      "loss": 1.0891,
+      "step": 5113
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.468909112973367e-05,
+      "loss": 1.1702,
+      "step": 5114
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.465731948318695e-05,
+      "loss": 1.1474,
+      "step": 5115
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.462554837748506e-05,
+      "loss": 1.0724,
+      "step": 5116
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.459377781584427e-05,
+      "loss": 1.0816,
+      "step": 5117
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.456200780148073e-05,
+      "loss": 1.095,
+      "step": 5118
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.453023833761045e-05,
+      "loss": 1.1668,
+      "step": 5119
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.449846942744965e-05,
+      "loss": 1.0482,
+      "step": 5120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.446670107421425e-05,
+      "loss": 1.1124,
+      "step": 5121
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.443493328112019e-05,
+      "loss": 1.1859,
+      "step": 5122
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.440316605138335e-05,
+      "loss": 1.1001,
+      "step": 5123
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.437139938821957e-05,
+      "loss": 1.1206,
+      "step": 5124
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.43396332948446e-05,
+      "loss": 1.1399,
+      "step": 5125
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.430786777447423e-05,
+      "loss": 1.1342,
+      "step": 5126
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.4276102830324e-05,
+      "loss": 1.073,
+      "step": 5127
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.424433846560959e-05,
+      "loss": 1.0261,
+      "step": 5128
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.421257468354649e-05,
+      "loss": 1.1418,
+      "step": 5129
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.418081148735021e-05,
+      "loss": 1.0788,
+      "step": 5130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.414904888023617e-05,
+      "loss": 1.0695,
+      "step": 5131
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.41172868654197e-05,
+      "loss": 1.0781,
+      "step": 5132
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.408552544611609e-05,
+      "loss": 1.1125,
+      "step": 5133
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.405376462554067e-05,
+      "loss": 1.0866,
+      "step": 5134
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.402200440690854e-05,
+      "loss": 1.0627,
+      "step": 5135
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.399024479343484e-05,
+      "loss": 1.043,
+      "step": 5136
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.395848578833463e-05,
+      "loss": 1.1133,
+      "step": 5137
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.39267273948229e-05,
+      "loss": 1.1349,
+      "step": 5138
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.389496961611459e-05,
+      "loss": 1.071,
+      "step": 5139
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.386321245542457e-05,
+      "loss": 1.1844,
+      "step": 5140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.383145591596762e-05,
+      "loss": 1.0921,
+      "step": 5141
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.379970000095854e-05,
+      "loss": 1.1063,
+      "step": 5142
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.376794471361197e-05,
+      "loss": 1.1881,
+      "step": 5143
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.373619005714256e-05,
+      "loss": 1.1049,
+      "step": 5144
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.370443603476482e-05,
+      "loss": 1.1486,
+      "step": 5145
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.367268264969329e-05,
+      "loss": 1.1258,
+      "step": 5146
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.364092990514234e-05,
+      "loss": 1.1054,
+      "step": 5147
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.36091778043264e-05,
+      "loss": 1.1487,
+      "step": 5148
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.357742635045973e-05,
+      "loss": 1.0931,
+      "step": 5149
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.354567554675657e-05,
+      "loss": 1.0557,
+      "step": 5150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.35139253964311e-05,
+      "loss": 1.0676,
+      "step": 5151
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.34821759026974e-05,
+      "loss": 1.17,
+      "step": 5152
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.345042706876948e-05,
+      "loss": 1.0969,
+      "step": 5153
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.341867889786133e-05,
+      "loss": 1.0606,
+      "step": 5154
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.338693139318686e-05,
+      "loss": 1.0907,
+      "step": 5155
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.33551845579599e-05,
+      "loss": 1.0634,
+      "step": 5156
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.332343839539419e-05,
+      "loss": 1.0576,
+      "step": 5157
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.329169290870345e-05,
+      "loss": 1.0931,
+      "step": 5158
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.325994810110129e-05,
+      "loss": 1.1279,
+      "step": 5159
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.322820397580125e-05,
+      "loss": 1.0811,
+      "step": 5160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.319646053601685e-05,
+      "loss": 1.1047,
+      "step": 5161
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.316471778496152e-05,
+      "loss": 1.1383,
+      "step": 5162
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.313297572584862e-05,
+      "loss": 1.1537,
+      "step": 5163
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.310123436189137e-05,
+      "loss": 1.0901,
+      "step": 5164
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.306949369630302e-05,
+      "loss": 1.1626,
+      "step": 5165
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.303775373229672e-05,
+      "loss": 1.1051,
+      "step": 5166
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.30060144730855e-05,
+      "loss": 1.1538,
+      "step": 5167
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.297427592188241e-05,
+      "loss": 1.1119,
+      "step": 5168
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.294253808190032e-05,
+      "loss": 1.1008,
+      "step": 5169
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.291080095635211e-05,
+      "loss": 1.1513,
+      "step": 5170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.287906454845054e-05,
+      "loss": 1.1291,
+      "step": 5171
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.284732886140836e-05,
+      "loss": 1.0666,
+      "step": 5172
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.281559389843818e-05,
+      "loss": 1.0896,
+      "step": 5173
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.278385966275252e-05,
+      "loss": 1.0908,
+      "step": 5174
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.275212615756394e-05,
+      "loss": 1.1239,
+      "step": 5175
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.272039338608484e-05,
+      "loss": 1.176,
+      "step": 5176
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.268866135152753e-05,
+      "loss": 1.1484,
+      "step": 5177
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.265693005710429e-05,
+      "loss": 1.1036,
+      "step": 5178
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.26251995060273e-05,
+      "loss": 1.0729,
+      "step": 5179
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.259346970150871e-05,
+      "loss": 1.1063,
+      "step": 5180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.256174064676053e-05,
+      "loss": 1.0968,
+      "step": 5181
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.253001234499472e-05,
+      "loss": 1.0953,
+      "step": 5182
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.249828479942317e-05,
+      "loss": 1.0988,
+      "step": 5183
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.24665580132577e-05,
+      "loss": 1.1839,
+      "step": 5184
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.243483198971008e-05,
+      "loss": 1.1812,
+      "step": 5185
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.240310673199189e-05,
+      "loss": 1.0797,
+      "step": 5186
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.237138224331475e-05,
+      "loss": 1.1096,
+      "step": 5187
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.233965852689012e-05,
+      "loss": 1.0943,
+      "step": 5188
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.230793558592951e-05,
+      "loss": 1.0741,
+      "step": 5189
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.22762134236442e-05,
+      "loss": 1.0407,
+      "step": 5190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.224449204324548e-05,
+      "loss": 1.138,
+      "step": 5191
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.221277144794454e-05,
+      "loss": 1.1566,
+      "step": 5192
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.218105164095249e-05,
+      "loss": 1.1582,
+      "step": 5193
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.214933262548032e-05,
+      "loss": 1.1499,
+      "step": 5194
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.211761440473901e-05,
+      "loss": 1.143,
+      "step": 5195
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.208589698193942e-05,
+      "loss": 1.0797,
+      "step": 5196
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.205418036029232e-05,
+      "loss": 1.0967,
+      "step": 5197
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.202246454300845e-05,
+      "loss": 1.1425,
+      "step": 5198
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.199074953329842e-05,
+      "loss": 1.091,
+      "step": 5199
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.195903533437273e-05,
+      "loss": 1.1471,
+      "step": 5200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.192732194944191e-05,
+      "loss": 1.0992,
+      "step": 5201
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.189560938171626e-05,
+      "loss": 1.0876,
+      "step": 5202
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.186389763440615e-05,
+      "loss": 1.1495,
+      "step": 5203
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.183218671072175e-05,
+      "loss": 1.1986,
+      "step": 5204
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.180047661387323e-05,
+      "loss": 1.1295,
+      "step": 5205
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.176876734707058e-05,
+      "loss": 1.1247,
+      "step": 5206
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.173705891352378e-05,
+      "loss": 1.131,
+      "step": 5207
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.170535131644271e-05,
+      "loss": 1.1294,
+      "step": 5208
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.167364455903715e-05,
+      "loss": 1.1185,
+      "step": 5209
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.164193864451686e-05,
+      "loss": 1.1447,
+      "step": 5210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.161023357609137e-05,
+      "loss": 1.1882,
+      "step": 5211
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.157852935697028e-05,
+      "loss": 1.0715,
+      "step": 5212
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.154682599036304e-05,
+      "loss": 1.1738,
+      "step": 5213
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.151512347947897e-05,
+      "loss": 1.1412,
+      "step": 5214
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.148342182752742e-05,
+      "loss": 1.125,
+      "step": 5215
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.145172103771744e-05,
+      "loss": 1.0867,
+      "step": 5216
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.142002111325829e-05,
+      "loss": 1.0783,
+      "step": 5217
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.138832205735893e-05,
+      "loss": 1.0483,
+      "step": 5218
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.135662387322826e-05,
+      "loss": 1.0471,
+      "step": 5219
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.132492656407513e-05,
+      "loss": 1.0621,
+      "step": 5220
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.129323013310832e-05,
+      "loss": 1.0232,
+      "step": 5221
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.126153458353646e-05,
+      "loss": 1.0983,
+      "step": 5222
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.122983991856814e-05,
+      "loss": 1.1434,
+      "step": 5223
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.119814614141181e-05,
+      "loss": 1.15,
+      "step": 5224
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.116645325527588e-05,
+      "loss": 1.1361,
+      "step": 5225
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.113476126336867e-05,
+      "loss": 1.0495,
+      "step": 5226
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.110307016889836e-05,
+      "loss": 1.1182,
+      "step": 5227
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.107137997507313e-05,
+      "loss": 1.2304,
+      "step": 5228
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.10396906851009e-05,
+      "loss": 1.0955,
+      "step": 5229
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.100800230218974e-05,
+      "loss": 1.0904,
+      "step": 5230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.09763148295474e-05,
+      "loss": 1.0901,
+      "step": 5231
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.09446282703817e-05,
+      "loss": 1.1149,
+      "step": 5232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.091294262790024e-05,
+      "loss": 1.1573,
+      "step": 5233
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.088125790531062e-05,
+      "loss": 1.1052,
+      "step": 5234
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.084957410582036e-05,
+      "loss": 1.1132,
+      "step": 5235
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.081789123263676e-05,
+      "loss": 1.1528,
+      "step": 5236
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.078620928896713e-05,
+      "loss": 1.0627,
+      "step": 5237
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.075452827801871e-05,
+      "loss": 1.0922,
+      "step": 5238
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.072284820299855e-05,
+      "loss": 1.1223,
+      "step": 5239
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.069116906711372e-05,
+      "loss": 1.1178,
+      "step": 5240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.065949087357105e-05,
+      "loss": 1.1123,
+      "step": 5241
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.062781362557739e-05,
+      "loss": 1.122,
+      "step": 5242
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.059613732633945e-05,
+      "loss": 1.0655,
+      "step": 5243
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.056446197906387e-05,
+      "loss": 1.0036,
+      "step": 5244
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.053278758695719e-05,
+      "loss": 1.0317,
+      "step": 5245
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.05011141532258e-05,
+      "loss": 1.1383,
+      "step": 5246
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.046944168107608e-05,
+      "loss": 1.1007,
+      "step": 5247
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.043777017371425e-05,
+      "loss": 1.1215,
+      "step": 5248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.04060996343464e-05,
+      "loss": 1.1239,
+      "step": 5249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.037443006617865e-05,
+      "loss": 1.0753,
+      "step": 5250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.034276147241686e-05,
+      "loss": 1.1027,
+      "step": 5251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.031109385626695e-05,
+      "loss": 1.1477,
+      "step": 5252
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.027942722093461e-05,
+      "loss": 1.1616,
+      "step": 5253
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.024776156962548e-05,
+      "loss": 1.1071,
+      "step": 5254
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.021609690554516e-05,
+      "loss": 1.1242,
+      "step": 5255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.018443323189905e-05,
+      "loss": 1.1016,
+      "step": 5256
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.015277055189249e-05,
+      "loss": 1.1166,
+      "step": 5257
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.012110886873075e-05,
+      "loss": 1.1137,
+      "step": 5258
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.008944818561897e-05,
+      "loss": 1.1295,
+      "step": 5259
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.005778850576223e-05,
+      "loss": 1.1201,
+      "step": 5260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.00261298323654e-05,
+      "loss": 1.0411,
+      "step": 5261
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.999447216863334e-05,
+      "loss": 1.1661,
+      "step": 5262
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.996281551777079e-05,
+      "loss": 1.0742,
+      "step": 5263
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.993115988298241e-05,
+      "loss": 1.1219,
+      "step": 5264
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.989950526747272e-05,
+      "loss": 1.0628,
+      "step": 5265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.986785167444612e-05,
+      "loss": 1.0695,
+      "step": 5266
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.983619910710695e-05,
+      "loss": 1.1409,
+      "step": 5267
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.980454756865945e-05,
+      "loss": 1.1111,
+      "step": 5268
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.97728970623077e-05,
+      "loss": 1.0516,
+      "step": 5269
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.974124759125577e-05,
+      "loss": 1.1203,
+      "step": 5270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.970959915870746e-05,
+      "loss": 1.1006,
+      "step": 5271
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.967795176786671e-05,
+      "loss": 1.1406,
+      "step": 5272
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.964630542193713e-05,
+      "loss": 1.1391,
+      "step": 5273
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.961466012412232e-05,
+      "loss": 1.0794,
+      "step": 5274
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.958301587762579e-05,
+      "loss": 1.1171,
+      "step": 5275
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.955137268565089e-05,
+      "loss": 1.1374,
+      "step": 5276
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.951973055140094e-05,
+      "loss": 1.1346,
+      "step": 5277
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.948808947807905e-05,
+      "loss": 1.1221,
+      "step": 5278
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.945644946888829e-05,
+      "loss": 1.1639,
+      "step": 5279
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.942481052703162e-05,
+      "loss": 1.0929,
+      "step": 5280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.939317265571187e-05,
+      "loss": 1.0539,
+      "step": 5281
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.936153585813183e-05,
+      "loss": 1.1508,
+      "step": 5282
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.932990013749403e-05,
+      "loss": 1.1157,
+      "step": 5283
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.9298265497001e-05,
+      "loss": 1.0863,
+      "step": 5284
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.926663193985524e-05,
+      "loss": 1.109,
+      "step": 5285
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.923499946925896e-05,
+      "loss": 1.1043,
+      "step": 5286
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.920336808841437e-05,
+      "loss": 1.1469,
+      "step": 5287
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.917173780052356e-05,
+      "loss": 1.1169,
+      "step": 5288
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.914010860878847e-05,
+      "loss": 1.2035,
+      "step": 5289
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.910848051641099e-05,
+      "loss": 1.1615,
+      "step": 5290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.907685352659281e-05,
+      "loss": 1.1283,
+      "step": 5291
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.904522764253561e-05,
+      "loss": 1.1249,
+      "step": 5292
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.901360286744089e-05,
+      "loss": 1.1288,
+      "step": 5293
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.898197920451005e-05,
+      "loss": 1.1257,
+      "step": 5294
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.895035665694441e-05,
+      "loss": 1.0526,
+      "step": 5295
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.891873522794511e-05,
+      "loss": 1.1416,
+      "step": 5296
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.888711492071326e-05,
+      "loss": 1.0637,
+      "step": 5297
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.885549573844978e-05,
+      "loss": 1.1095,
+      "step": 5298
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.882387768435553e-05,
+      "loss": 1.0747,
+      "step": 5299
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.879226076163125e-05,
+      "loss": 1.1813,
+      "step": 5300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.876064497347755e-05,
+      "loss": 1.1003,
+      "step": 5301
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.872903032309493e-05,
+      "loss": 1.1599,
+      "step": 5302
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.869741681368373e-05,
+      "loss": 1.0829,
+      "step": 5303
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.866580444844425e-05,
+      "loss": 1.1546,
+      "step": 5304
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.863419323057665e-05,
+      "loss": 1.1134,
+      "step": 5305
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.860258316328093e-05,
+      "loss": 1.1321,
+      "step": 5306
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.857097424975708e-05,
+      "loss": 1.0615,
+      "step": 5307
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.853936649320482e-05,
+      "loss": 1.0478,
+      "step": 5308
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.850775989682388e-05,
+      "loss": 1.1917,
+      "step": 5309
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.847615446381378e-05,
+      "loss": 1.1062,
+      "step": 5310
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.844455019737404e-05,
+      "loss": 1.125,
+      "step": 5311
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.841294710070391e-05,
+      "loss": 1.1659,
+      "step": 5312
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.838134517700268e-05,
+      "loss": 1.1022,
+      "step": 5313
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.834974442946942e-05,
+      "loss": 1.1384,
+      "step": 5314
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.831814486130309e-05,
+      "loss": 1.0813,
+      "step": 5315
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.828654647570255e-05,
+      "loss": 1.0534,
+      "step": 5316
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.825494927586654e-05,
+      "loss": 1.0467,
+      "step": 5317
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.822335326499369e-05,
+      "loss": 1.1321,
+      "step": 5318
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.819175844628249e-05,
+      "loss": 1.1395,
+      "step": 5319
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.816016482293127e-05,
+      "loss": 1.0624,
+      "step": 5320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.812857239813835e-05,
+      "loss": 1.1388,
+      "step": 5321
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.80969811751018e-05,
+      "loss": 1.071,
+      "step": 5322
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.806539115701968e-05,
+      "loss": 1.0731,
+      "step": 5323
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.803380234708988e-05,
+      "loss": 1.1776,
+      "step": 5324
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.800221474851011e-05,
+      "loss": 1.1189,
+      "step": 5325
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.797062836447803e-05,
+      "loss": 1.1303,
+      "step": 5326
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.793904319819124e-05,
+      "loss": 1.0299,
+      "step": 5327
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.790745925284705e-05,
+      "loss": 1.0726,
+      "step": 5328
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.787587653164278e-05,
+      "loss": 1.074,
+      "step": 5329
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.784429503777556e-05,
+      "loss": 1.1158,
+      "step": 5330
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.781271477444241e-05,
+      "loss": 1.1226,
+      "step": 5331
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.778113574484029e-05,
+      "loss": 1.1209,
+      "step": 5332
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.774955795216589e-05,
+      "loss": 1.1463,
+      "step": 5333
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.77179813996159e-05,
+      "loss": 1.1153,
+      "step": 5334
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.768640609038685e-05,
+      "loss": 1.0982,
+      "step": 5335
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.765483202767514e-05,
+      "loss": 1.1537,
+      "step": 5336
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.762325921467705e-05,
+      "loss": 1.0987,
+      "step": 5337
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.75916876545887e-05,
+      "loss": 1.1526,
+      "step": 5338
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.75601173506061e-05,
+      "loss": 1.0822,
+      "step": 5339
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.75285483059252e-05,
+      "loss": 1.1564,
+      "step": 5340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.749698052374174e-05,
+      "loss": 1.1361,
+      "step": 5341
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.746541400725135e-05,
+      "loss": 1.1252,
+      "step": 5342
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.743384875964953e-05,
+      "loss": 1.0648,
+      "step": 5343
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.740228478413173e-05,
+      "loss": 1.1643,
+      "step": 5344
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.73707220838931e-05,
+      "loss": 1.15,
+      "step": 5345
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.733916066212882e-05,
+      "loss": 1.1135,
+      "step": 5346
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.730760052203387e-05,
+      "loss": 1.0672,
+      "step": 5347
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.727604166680311e-05,
+      "loss": 1.1333,
+      "step": 5348
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.724448409963133e-05,
+      "loss": 1.1055,
+      "step": 5349
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.721292782371306e-05,
+      "loss": 1.1269,
+      "step": 5350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.718137284224279e-05,
+      "loss": 1.0889,
+      "step": 5351
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.714981915841488e-05,
+      "loss": 1.1125,
+      "step": 5352
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.71182667754235e-05,
+      "loss": 1.1005,
+      "step": 5353
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.70867156964628e-05,
+      "loss": 1.1151,
+      "step": 5354
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.705516592472669e-05,
+      "loss": 1.0888,
+      "step": 5355
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.702361746340897e-05,
+      "loss": 1.1017,
+      "step": 5356
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.699207031570338e-05,
+      "loss": 1.1139,
+      "step": 5357
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.696052448480339e-05,
+      "loss": 1.0671,
+      "step": 5358
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.692897997390245e-05,
+      "loss": 1.1234,
+      "step": 5359
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.689743678619385e-05,
+      "loss": 1.1567,
+      "step": 5360
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.686589492487072e-05,
+      "loss": 1.1196,
+      "step": 5361
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.683435439312612e-05,
+      "loss": 1.1107,
+      "step": 5362
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.680281519415288e-05,
+      "loss": 1.0554,
+      "step": 5363
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.677127733114374e-05,
+      "loss": 1.1094,
+      "step": 5364
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.673974080729134e-05,
+      "loss": 1.105,
+      "step": 5365
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.670820562578816e-05,
+      "loss": 1.1168,
+      "step": 5366
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.667667178982645e-05,
+      "loss": 1.0793,
+      "step": 5367
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.664513930259854e-05,
+      "loss": 1.1072,
+      "step": 5368
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.661360816729645e-05,
+      "loss": 1.1216,
+      "step": 5369
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.658207838711207e-05,
+      "loss": 1.1881,
+      "step": 5370
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.655054996523721e-05,
+      "loss": 1.1579,
+      "step": 5371
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.651902290486354e-05,
+      "loss": 1.0341,
+      "step": 5372
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.648749720918257e-05,
+      "loss": 1.0612,
+      "step": 5373
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.64559728813857e-05,
+      "loss": 1.0669,
+      "step": 5374
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.642444992466409e-05,
+      "loss": 1.1144,
+      "step": 5375
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.639292834220891e-05,
+      "loss": 1.1321,
+      "step": 5376
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.63614081372111e-05,
+      "loss": 1.146,
+      "step": 5377
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.632988931286149e-05,
+      "loss": 1.0662,
+      "step": 5378
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.629837187235076e-05,
+      "loss": 1.0993,
+      "step": 5379
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.626685581886946e-05,
+      "loss": 1.1128,
+      "step": 5380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.62353411556079e-05,
+      "loss": 1.1325,
+      "step": 5381
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.62038278857565e-05,
+      "loss": 1.0554,
+      "step": 5382
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.617231601250528e-05,
+      "loss": 1.1274,
+      "step": 5383
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.614080553904424e-05,
+      "loss": 1.0387,
+      "step": 5384
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.610929646856321e-05,
+      "loss": 1.0484,
+      "step": 5385
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.60777888042519e-05,
+      "loss": 1.1424,
+      "step": 5386
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.604628254929984e-05,
+      "loss": 1.1567,
+      "step": 5387
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.601477770689646e-05,
+      "loss": 1.1197,
+      "step": 5388
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.5983274280231e-05,
+      "loss": 1.1084,
+      "step": 5389
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.595177227249261e-05,
+      "loss": 1.1715,
+      "step": 5390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.592027168687027e-05,
+      "loss": 1.1606,
+      "step": 5391
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.588877252655278e-05,
+      "loss": 1.1583,
+      "step": 5392
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.585727479472887e-05,
+      "loss": 1.1084,
+      "step": 5393
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.582577849458704e-05,
+      "loss": 1.1353,
+      "step": 5394
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.579428362931577e-05,
+      "loss": 1.0666,
+      "step": 5395
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.576279020210325e-05,
+      "loss": 1.0622,
+      "step": 5396
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.573129821613761e-05,
+      "loss": 1.1102,
+      "step": 5397
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.569980767460684e-05,
+      "loss": 1.1088,
+      "step": 5398
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.566831858069872e-05,
+      "loss": 1.148,
+      "step": 5399
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.563683093760096e-05,
+      "loss": 1.1309,
+      "step": 5400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.560534474850103e-05,
+      "loss": 1.0937,
+      "step": 5401
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.557386001658635e-05,
+      "loss": 1.1273,
+      "step": 5402
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.554237674504415e-05,
+      "loss": 1.065,
+      "step": 5403
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.551089493706152e-05,
+      "loss": 1.112,
+      "step": 5404
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.547941459582536e-05,
+      "loss": 1.1267,
+      "step": 5405
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.544793572452248e-05,
+      "loss": 1.1029,
+      "step": 5406
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.541645832633951e-05,
+      "loss": 1.0825,
+      "step": 5407
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.538498240446291e-05,
+      "loss": 1.0757,
+      "step": 5408
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.535350796207908e-05,
+      "loss": 1.0655,
+      "step": 5409
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.532203500237419e-05,
+      "loss": 1.1473,
+      "step": 5410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.529056352853427e-05,
+      "loss": 1.0566,
+      "step": 5411
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.525909354374521e-05,
+      "loss": 1.0876,
+      "step": 5412
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.522762505119273e-05,
+      "loss": 1.1229,
+      "step": 5413
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.519615805406242e-05,
+      "loss": 1.157,
+      "step": 5414
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.516469255553974e-05,
+      "loss": 1.0343,
+      "step": 5415
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.513322855880998e-05,
+      "loss": 1.1089,
+      "step": 5416
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.510176606705824e-05,
+      "loss": 1.1611,
+      "step": 5417
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.50703050834695e-05,
+      "loss": 1.1513,
+      "step": 5418
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.50388456112286e-05,
+      "loss": 1.1218,
+      "step": 5419
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.500738765352023e-05,
+      "loss": 1.1052,
+      "step": 5420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.497593121352889e-05,
+      "loss": 1.073,
+      "step": 5421
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.494447629443889e-05,
+      "loss": 1.1402,
+      "step": 5422
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.491302289943454e-05,
+      "loss": 1.1423,
+      "step": 5423
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.488157103169989e-05,
+      "loss": 1.1317,
+      "step": 5424
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.485012069441879e-05,
+      "loss": 1.0742,
+      "step": 5425
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.4818671890775e-05,
+      "loss": 1.1668,
+      "step": 5426
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.478722462395214e-05,
+      "loss": 1.1309,
+      "step": 5427
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.475577889713364e-05,
+      "loss": 1.1627,
+      "step": 5428
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.472433471350279e-05,
+      "loss": 1.1637,
+      "step": 5429
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.469289207624266e-05,
+      "loss": 1.0694,
+      "step": 5430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.466145098853628e-05,
+      "loss": 1.1072,
+      "step": 5431
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.463001145356646e-05,
+      "loss": 1.0626,
+      "step": 5432
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.459857347451585e-05,
+      "loss": 1.0943,
+      "step": 5433
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.45671370545669e-05,
+      "loss": 1.1258,
+      "step": 5434
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.4535702196902e-05,
+      "loss": 1.1961,
+      "step": 5435
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.450426890470329e-05,
+      "loss": 1.1851,
+      "step": 5436
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.447283718115285e-05,
+      "loss": 1.0903,
+      "step": 5437
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.444140702943251e-05,
+      "loss": 1.1095,
+      "step": 5438
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.4409978452724e-05,
+      "loss": 1.1056,
+      "step": 5439
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.437855145420883e-05,
+      "loss": 1.0546,
+      "step": 5440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.434712603706843e-05,
+      "loss": 1.0846,
+      "step": 5441
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.4315702204484e-05,
+      "loss": 1.0667,
+      "step": 5442
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.428427995963658e-05,
+      "loss": 1.1072,
+      "step": 5443
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.425285930570711e-05,
+      "loss": 1.113,
+      "step": 5444
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.422144024587633e-05,
+      "loss": 1.1068,
+      "step": 5445
+    },
+    {
+      "epoch": 2.18,
+      "eval_loss": 0.8212035894393921,
+      "eval_runtime": 504.249,
+      "eval_samples_per_second": 51.364,
+      "eval_steps_per_second": 25.682,
+      "step": 5445
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.419002278332483e-05,
+      "loss": 1.1197,
+      "step": 5446
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.415860692123301e-05,
+      "loss": 1.0386,
+      "step": 5447
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.412719266278113e-05,
+      "loss": 1.0703,
+      "step": 5448
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.409578001114929e-05,
+      "loss": 1.153,
+      "step": 5449
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.406436896951743e-05,
+      "loss": 1.0993,
+      "step": 5450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.403295954106531e-05,
+      "loss": 1.0631,
+      "step": 5451
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.400155172897258e-05,
+      "loss": 1.1189,
+      "step": 5452
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.397014553641865e-05,
+      "loss": 1.0644,
+      "step": 5453
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.393874096658279e-05,
+      "loss": 1.2069,
+      "step": 5454
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.390733802264412e-05,
+      "loss": 1.1697,
+      "step": 5455
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.387593670778159e-05,
+      "loss": 1.1272,
+      "step": 5456
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.384453702517401e-05,
+      "loss": 1.1476,
+      "step": 5457
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.3813138978e-05,
+      "loss": 1.1152,
+      "step": 5458
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.378174256943798e-05,
+      "loss": 1.1399,
+      "step": 5459
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.375034780266627e-05,
+      "loss": 1.0416,
+      "step": 5460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.371895468086297e-05,
+      "loss": 1.1033,
+      "step": 5461
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.368756320720605e-05,
+      "loss": 1.2044,
+      "step": 5462
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.365617338487329e-05,
+      "loss": 1.0899,
+      "step": 5463
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.362478521704233e-05,
+      "loss": 1.1725,
+      "step": 5464
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.359339870689063e-05,
+      "loss": 1.1023,
+      "step": 5465
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.35620138575955e-05,
+      "loss": 1.1026,
+      "step": 5466
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.353063067233399e-05,
+      "loss": 1.1581,
+      "step": 5467
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.349924915428308e-05,
+      "loss": 1.1028,
+      "step": 5468
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.346786930661959e-05,
+      "loss": 1.095,
+      "step": 5469
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.343649113252008e-05,
+      "loss": 1.1569,
+      "step": 5470
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.340511463516103e-05,
+      "loss": 1.0386,
+      "step": 5471
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.337373981771869e-05,
+      "loss": 1.1574,
+      "step": 5472
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.334236668336918e-05,
+      "loss": 1.1102,
+      "step": 5473
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.331099523528841e-05,
+      "loss": 1.1275,
+      "step": 5474
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.327962547665217e-05,
+      "loss": 1.1029,
+      "step": 5475
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.324825741063606e-05,
+      "loss": 1.1204,
+      "step": 5476
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.321689104041544e-05,
+      "loss": 1.1411,
+      "step": 5477
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.318552636916565e-05,
+      "loss": 1.0941,
+      "step": 5478
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.31541634000617e-05,
+      "loss": 1.1577,
+      "step": 5479
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.31228021362785e-05,
+      "loss": 1.0279,
+      "step": 5480
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.309144258099081e-05,
+      "loss": 1.0981,
+      "step": 5481
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.306008473737317e-05,
+      "loss": 1.1165,
+      "step": 5482
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.302872860859999e-05,
+      "loss": 1.1166,
+      "step": 5483
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.299737419784543e-05,
+      "loss": 1.1031,
+      "step": 5484
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.296602150828358e-05,
+      "loss": 1.1129,
+      "step": 5485
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.293467054308826e-05,
+      "loss": 1.1105,
+      "step": 5486
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.29033213054332e-05,
+      "loss": 1.0704,
+      "step": 5487
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.28719737984919e-05,
+      "loss": 1.1271,
+      "step": 5488
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.284062802543771e-05,
+      "loss": 1.1548,
+      "step": 5489
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.280928398944375e-05,
+      "loss": 1.0797,
+      "step": 5490
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.277794169368303e-05,
+      "loss": 1.1565,
+      "step": 5491
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.274660114132839e-05,
+      "loss": 1.0516,
+      "step": 5492
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.271526233555243e-05,
+      "loss": 1.1193,
+      "step": 5493
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.268392527952765e-05,
+      "loss": 1.1483,
+      "step": 5494
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.265258997642628e-05,
+      "loss": 1.1018,
+      "step": 5495
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.262125642942048e-05,
+      "loss": 1.1422,
+      "step": 5496
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.258992464168211e-05,
+      "loss": 1.1149,
+      "step": 5497
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.255859461638298e-05,
+      "loss": 1.1393,
+      "step": 5498
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.252726635669461e-05,
+      "loss": 1.0872,
+      "step": 5499
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.249593986578844e-05,
+      "loss": 1.1253,
+      "step": 5500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.246461514683564e-05,
+      "loss": 1.1292,
+      "step": 5501
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.243329220300725e-05,
+      "loss": 1.0812,
+      "step": 5502
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.240197103747413e-05,
+      "loss": 1.1469,
+      "step": 5503
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.237065165340693e-05,
+      "loss": 1.1056,
+      "step": 5504
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.23393340539762e-05,
+      "loss": 1.1201,
+      "step": 5505
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.230801824235221e-05,
+      "loss": 1.1531,
+      "step": 5506
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.227670422170509e-05,
+      "loss": 1.131,
+      "step": 5507
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.224539199520482e-05,
+      "loss": 1.0821,
+      "step": 5508
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.221408156602113e-05,
+      "loss": 1.1068,
+      "step": 5509
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.218277293732361e-05,
+      "loss": 1.1388,
+      "step": 5510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.215146611228169e-05,
+      "loss": 1.1249,
+      "step": 5511
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.212016109406455e-05,
+      "loss": 1.0782,
+      "step": 5512
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.208885788584128e-05,
+      "loss": 1.1191,
+      "step": 5513
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.20575564907807e-05,
+      "loss": 1.0877,
+      "step": 5514
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.202625691205149e-05,
+      "loss": 1.1073,
+      "step": 5515
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.199495915282212e-05,
+      "loss": 1.1195,
+      "step": 5516
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.196366321626094e-05,
+      "loss": 1.088,
+      "step": 5517
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.193236910553599e-05,
+      "loss": 1.144,
+      "step": 5518
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.190107682381527e-05,
+      "loss": 1.1299,
+      "step": 5519
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.186978637426654e-05,
+      "loss": 1.1057,
+      "step": 5520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.183849776005732e-05,
+      "loss": 1.101,
+      "step": 5521
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.1807210984355e-05,
+      "loss": 1.1344,
+      "step": 5522
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.177592605032678e-05,
+      "loss": 1.1893,
+      "step": 5523
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.174464296113963e-05,
+      "loss": 1.1378,
+      "step": 5524
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.171336171996044e-05,
+      "loss": 1.1303,
+      "step": 5525
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.168208232995578e-05,
+      "loss": 1.1466,
+      "step": 5526
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.16508047942921e-05,
+      "loss": 1.0707,
+      "step": 5527
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.161952911613569e-05,
+      "loss": 1.1286,
+      "step": 5528
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.158825529865256e-05,
+      "loss": 1.1557,
+      "step": 5529
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.155698334500867e-05,
+      "loss": 1.0605,
+      "step": 5530
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.152571325836963e-05,
+      "loss": 1.0869,
+      "step": 5531
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.149444504190094e-05,
+      "loss": 1.1824,
+      "step": 5532
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.146317869876801e-05,
+      "loss": 1.1626,
+      "step": 5533
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.143191423213589e-05,
+      "loss": 1.065,
+      "step": 5534
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.140065164516952e-05,
+      "loss": 1.0526,
+      "step": 5535
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.136939094103367e-05,
+      "loss": 1.1101,
+      "step": 5536
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.133813212289284e-05,
+      "loss": 1.1095,
+      "step": 5537
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.130687519391145e-05,
+      "loss": 1.1081,
+      "step": 5538
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.127562015725363e-05,
+      "loss": 1.1049,
+      "step": 5539
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.124436701608336e-05,
+      "loss": 1.1479,
+      "step": 5540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.121311577356445e-05,
+      "loss": 1.0854,
+      "step": 5541
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.118186643286047e-05,
+      "loss": 1.1613,
+      "step": 5542
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.115061899713486e-05,
+      "loss": 1.1121,
+      "step": 5543
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.111937346955078e-05,
+      "loss": 1.0859,
+      "step": 5544
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.108812985327128e-05,
+      "loss": 1.1386,
+      "step": 5545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.105688815145913e-05,
+      "loss": 1.1227,
+      "step": 5546
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.102564836727701e-05,
+      "loss": 1.093,
+      "step": 5547
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.099441050388737e-05,
+      "loss": 1.1174,
+      "step": 5548
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.096317456445242e-05,
+      "loss": 1.0739,
+      "step": 5549
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.093194055213421e-05,
+      "loss": 1.1299,
+      "step": 5550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.090070847009458e-05,
+      "loss": 1.1143,
+      "step": 5551
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.086947832149518e-05,
+      "loss": 1.1613,
+      "step": 5552
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.08382501094975e-05,
+      "loss": 1.0921,
+      "step": 5553
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.080702383726277e-05,
+      "loss": 1.0918,
+      "step": 5554
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.077579950795211e-05,
+      "loss": 1.1301,
+      "step": 5555
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.074457712472632e-05,
+      "loss": 1.1347,
+      "step": 5556
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.071335669074608e-05,
+      "loss": 1.1198,
+      "step": 5557
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.068213820917192e-05,
+      "loss": 1.1116,
+      "step": 5558
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.065092168316406e-05,
+      "loss": 1.0671,
+      "step": 5559
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.061970711588262e-05,
+      "loss": 1.1032,
+      "step": 5560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.058849451048747e-05,
+      "loss": 1.1138,
+      "step": 5561
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.05572838701383e-05,
+      "loss": 1.0852,
+      "step": 5562
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.052607519799461e-05,
+      "loss": 1.0736,
+      "step": 5563
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.049486849721564e-05,
+      "loss": 1.1445,
+      "step": 5564
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.046366377096048e-05,
+      "loss": 1.0718,
+      "step": 5565
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.043246102238804e-05,
+      "loss": 1.1784,
+      "step": 5566
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.040126025465702e-05,
+      "loss": 1.1064,
+      "step": 5567
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.037006147092585e-05,
+      "loss": 1.0572,
+      "step": 5568
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.033886467435286e-05,
+      "loss": 1.1164,
+      "step": 5569
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.030766986809611e-05,
+      "loss": 1.1399,
+      "step": 5570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.027647705531348e-05,
+      "loss": 1.097,
+      "step": 5571
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.02452862391627e-05,
+      "loss": 1.0813,
+      "step": 5572
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.021409742280112e-05,
+      "loss": 1.1468,
+      "step": 5573
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.018291060938617e-05,
+      "loss": 1.1056,
+      "step": 5574
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.015172580207485e-05,
+      "loss": 1.1291,
+      "step": 5575
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.012054300402401e-05,
+      "loss": 1.1077,
+      "step": 5576
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.008936221839036e-05,
+      "loss": 1.1072,
+      "step": 5577
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.005818344833031e-05,
+      "loss": 1.1609,
+      "step": 5578
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.002700669700017e-05,
+      "loss": 1.1032,
+      "step": 5579
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.999583196755597e-05,
+      "loss": 1.1115,
+      "step": 5580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.996465926315357e-05,
+      "loss": 1.12,
+      "step": 5581
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.993348858694858e-05,
+      "loss": 1.0722,
+      "step": 5582
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.990231994209647e-05,
+      "loss": 1.0865,
+      "step": 5583
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.987115333175245e-05,
+      "loss": 1.1646,
+      "step": 5584
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.98399887590716e-05,
+      "loss": 1.0575,
+      "step": 5585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.980882622720867e-05,
+      "loss": 1.0637,
+      "step": 5586
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.977766573931829e-05,
+      "loss": 1.1205,
+      "step": 5587
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.974650729855492e-05,
+      "loss": 1.0996,
+      "step": 5588
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.971535090807269e-05,
+      "loss": 1.1231,
+      "step": 5589
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.968419657102566e-05,
+      "loss": 1.0378,
+      "step": 5590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.965304429056756e-05,
+      "loss": 1.0906,
+      "step": 5591
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.962189406985201e-05,
+      "loss": 1.088,
+      "step": 5592
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.959074591203232e-05,
+      "loss": 1.1272,
+      "step": 5593
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.955959982026171e-05,
+      "loss": 1.1271,
+      "step": 5594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.952845579769307e-05,
+      "loss": 1.0673,
+      "step": 5595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.94973138474792e-05,
+      "loss": 1.1006,
+      "step": 5596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.946617397277262e-05,
+      "loss": 1.1344,
+      "step": 5597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.943503617672561e-05,
+      "loss": 1.1246,
+      "step": 5598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.94039004624903e-05,
+      "loss": 1.1195,
+      "step": 5599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.93727668332186e-05,
+      "loss": 1.1117,
+      "step": 5600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.934163529206217e-05,
+      "loss": 1.1478,
+      "step": 5601
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.931050584217252e-05,
+      "loss": 1.1381,
+      "step": 5602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.927937848670091e-05,
+      "loss": 1.1187,
+      "step": 5603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.924825322879839e-05,
+      "loss": 1.1772,
+      "step": 5604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.92171300716158e-05,
+      "loss": 1.0379,
+      "step": 5605
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.918600901830376e-05,
+      "loss": 1.1185,
+      "step": 5606
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.91548900720127e-05,
+      "loss": 1.1351,
+      "step": 5607
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.912377323589279e-05,
+      "loss": 1.1662,
+      "step": 5608
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.909265851309406e-05,
+      "loss": 1.1116,
+      "step": 5609
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.906154590676628e-05,
+      "loss": 1.1148,
+      "step": 5610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.903043542005897e-05,
+      "loss": 1.0711,
+      "step": 5611
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.899932705612149e-05,
+      "loss": 1.0647,
+      "step": 5612
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.8968220818103e-05,
+      "loss": 1.1063,
+      "step": 5613
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.893711670915237e-05,
+      "loss": 1.1988,
+      "step": 5614
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.890601473241835e-05,
+      "loss": 1.0871,
+      "step": 5615
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.887491489104938e-05,
+      "loss": 1.1246,
+      "step": 5616
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.884381718819379e-05,
+      "loss": 1.159,
+      "step": 5617
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.881272162699955e-05,
+      "loss": 1.033,
+      "step": 5618
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.878162821061455e-05,
+      "loss": 1.1467,
+      "step": 5619
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.87505369421864e-05,
+      "loss": 1.0525,
+      "step": 5620
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.871944782486247e-05,
+      "loss": 1.0025,
+      "step": 5621
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.868836086178998e-05,
+      "loss": 1.0962,
+      "step": 5622
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.865727605611588e-05,
+      "loss": 1.0822,
+      "step": 5623
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.86261934109869e-05,
+      "loss": 1.1192,
+      "step": 5624
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.859511292954957e-05,
+      "loss": 1.1145,
+      "step": 5625
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.856403461495022e-05,
+      "loss": 1.1306,
+      "step": 5626
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.853295847033495e-05,
+      "loss": 1.1091,
+      "step": 5627
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.850188449884954e-05,
+      "loss": 1.1039,
+      "step": 5628
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.847081270363974e-05,
+      "loss": 1.0707,
+      "step": 5629
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.843974308785098e-05,
+      "loss": 1.1197,
+      "step": 5630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.84086756546284e-05,
+      "loss": 1.2128,
+      "step": 5631
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.837761040711702e-05,
+      "loss": 1.163,
+      "step": 5632
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.834654734846161e-05,
+      "loss": 1.1186,
+      "step": 5633
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.831548648180671e-05,
+      "loss": 1.0847,
+      "step": 5634
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.828442781029664e-05,
+      "loss": 1.1676,
+      "step": 5635
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.82533713370755e-05,
+      "loss": 1.1209,
+      "step": 5636
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.822231706528715e-05,
+      "loss": 1.1204,
+      "step": 5637
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.819126499807529e-05,
+      "loss": 1.0538,
+      "step": 5638
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.816021513858332e-05,
+      "loss": 1.0724,
+      "step": 5639
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.812916748995445e-05,
+      "loss": 1.0606,
+      "step": 5640
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.809812205533167e-05,
+      "loss": 1.1277,
+      "step": 5641
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.80670788378577e-05,
+      "loss": 1.0933,
+      "step": 5642
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.803603784067514e-05,
+      "loss": 1.1496,
+      "step": 5643
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.800499906692628e-05,
+      "loss": 1.0929,
+      "step": 5644
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.797396251975321e-05,
+      "loss": 1.0971,
+      "step": 5645
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.794292820229777e-05,
+      "loss": 1.0796,
+      "step": 5646
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.791189611770162e-05,
+      "loss": 1.0853,
+      "step": 5647
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.788086626910615e-05,
+      "loss": 1.0779,
+      "step": 5648
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.784983865965253e-05,
+      "loss": 1.1583,
+      "step": 5649
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.781881329248173e-05,
+      "loss": 1.0815,
+      "step": 5650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.77877901707345e-05,
+      "loss": 1.1328,
+      "step": 5651
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.775676929755134e-05,
+      "loss": 1.1517,
+      "step": 5652
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.772575067607247e-05,
+      "loss": 1.1447,
+      "step": 5653
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.769473430943798e-05,
+      "loss": 1.159,
+      "step": 5654
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.766372020078768e-05,
+      "loss": 1.0867,
+      "step": 5655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.763270835326114e-05,
+      "loss": 1.1089,
+      "step": 5656
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.760169876999776e-05,
+      "loss": 1.1234,
+      "step": 5657
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.757069145413664e-05,
+      "loss": 1.1399,
+      "step": 5658
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.753968640881673e-05,
+      "loss": 1.1605,
+      "step": 5659
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.750868363717663e-05,
+      "loss": 1.0953,
+      "step": 5660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.747768314235481e-05,
+      "loss": 1.0857,
+      "step": 5661
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.744668492748948e-05,
+      "loss": 1.1319,
+      "step": 5662
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.741568899571865e-05,
+      "loss": 1.0915,
+      "step": 5663
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.738469535018005e-05,
+      "loss": 1.123,
+      "step": 5664
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.735370399401116e-05,
+      "loss": 1.1529,
+      "step": 5665
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.732271493034931e-05,
+      "loss": 1.1058,
+      "step": 5666
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.729172816233154e-05,
+      "loss": 1.129,
+      "step": 5667
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.726074369309467e-05,
+      "loss": 1.154,
+      "step": 5668
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.722976152577526e-05,
+      "loss": 1.0779,
+      "step": 5669
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.719878166350973e-05,
+      "loss": 1.1745,
+      "step": 5670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.716780410943418e-05,
+      "loss": 1.1035,
+      "step": 5671
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.713682886668449e-05,
+      "loss": 1.055,
+      "step": 5672
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.710585593839629e-05,
+      "loss": 1.0976,
+      "step": 5673
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.707488532770503e-05,
+      "loss": 1.113,
+      "step": 5674
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.704391703774588e-05,
+      "loss": 1.0872,
+      "step": 5675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.70129510716538e-05,
+      "loss": 1.0762,
+      "step": 5676
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.698198743256353e-05,
+      "loss": 1.1142,
+      "step": 5677
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.695102612360949e-05,
+      "loss": 1.1318,
+      "step": 5678
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.692006714792596e-05,
+      "loss": 1.0726,
+      "step": 5679
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.688911050864697e-05,
+      "loss": 1.1141,
+      "step": 5680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.685815620890626e-05,
+      "loss": 1.1698,
+      "step": 5681
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.682720425183736e-05,
+      "loss": 1.0967,
+      "step": 5682
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.679625464057353e-05,
+      "loss": 1.0695,
+      "step": 5683
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.676530737824796e-05,
+      "loss": 1.0815,
+      "step": 5684
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.673436246799337e-05,
+      "loss": 1.1412,
+      "step": 5685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.670341991294234e-05,
+      "loss": 1.1051,
+      "step": 5686
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.667247971622725e-05,
+      "loss": 1.1433,
+      "step": 5687
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.66415418809802e-05,
+      "loss": 1.1076,
+      "step": 5688
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.661060641033308e-05,
+      "loss": 1.1042,
+      "step": 5689
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.657967330741747e-05,
+      "loss": 1.087,
+      "step": 5690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.654874257536478e-05,
+      "loss": 1.1429,
+      "step": 5691
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.651781421730617e-05,
+      "loss": 1.0975,
+      "step": 5692
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.648688823637253e-05,
+      "loss": 1.0978,
+      "step": 5693
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.645596463569455e-05,
+      "loss": 1.0508,
+      "step": 5694
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.642504341840262e-05,
+      "loss": 1.0913,
+      "step": 5695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.639412458762697e-05,
+      "loss": 1.1424,
+      "step": 5696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.636320814649749e-05,
+      "loss": 1.0928,
+      "step": 5697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.633229409814394e-05,
+      "loss": 1.1067,
+      "step": 5698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.630138244569575e-05,
+      "loss": 1.1157,
+      "step": 5699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.627047319228214e-05,
+      "loss": 1.1049,
+      "step": 5700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.623956634103211e-05,
+      "loss": 1.1026,
+      "step": 5701
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.620866189507435e-05,
+      "loss": 1.1342,
+      "step": 5702
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.617775985753736e-05,
+      "loss": 1.0986,
+      "step": 5703
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.61468602315494e-05,
+      "loss": 1.1354,
+      "step": 5704
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.611596302023845e-05,
+      "loss": 1.0858,
+      "step": 5705
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.60850682267323e-05,
+      "loss": 1.1117,
+      "step": 5706
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.605417585415841e-05,
+      "loss": 1.0792,
+      "step": 5707
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.602328590564408e-05,
+      "loss": 1.1149,
+      "step": 5708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.599239838431631e-05,
+      "loss": 1.137,
+      "step": 5709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.596151329330189e-05,
+      "loss": 1.1095,
+      "step": 5710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.593063063572733e-05,
+      "loss": 1.1678,
+      "step": 5711
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.589975041471893e-05,
+      "loss": 1.1319,
+      "step": 5712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.586887263340274e-05,
+      "loss": 1.0977,
+      "step": 5713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.583799729490454e-05,
+      "loss": 1.1233,
+      "step": 5714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.580712440234983e-05,
+      "loss": 1.0883,
+      "step": 5715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.577625395886396e-05,
+      "loss": 1.0847,
+      "step": 5716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.574538596757194e-05,
+      "loss": 1.1505,
+      "step": 5717
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.571452043159857e-05,
+      "loss": 1.1222,
+      "step": 5718
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.568365735406844e-05,
+      "loss": 1.163,
+      "step": 5719
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.565279673810577e-05,
+      "loss": 1.146,
+      "step": 5720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.562193858683468e-05,
+      "loss": 1.1557,
+      "step": 5721
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.559108290337895e-05,
+      "loss": 1.1052,
+      "step": 5722
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.556022969086214e-05,
+      "loss": 1.1133,
+      "step": 5723
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.552937895240749e-05,
+      "loss": 1.1508,
+      "step": 5724
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.549853069113813e-05,
+      "loss": 1.1021,
+      "step": 5725
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.546768491017686e-05,
+      "loss": 1.0936,
+      "step": 5726
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.543684161264618e-05,
+      "loss": 1.1771,
+      "step": 5727
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.540600080166841e-05,
+      "loss": 1.1996,
+      "step": 5728
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.537516248036559e-05,
+      "loss": 1.0667,
+      "step": 5729
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.534432665185953e-05,
+      "loss": 1.0856,
+      "step": 5730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.531349331927176e-05,
+      "loss": 1.1088,
+      "step": 5731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.528266248572358e-05,
+      "loss": 1.066,
+      "step": 5732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.525183415433599e-05,
+      "loss": 1.0943,
+      "step": 5733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.52210083282298e-05,
+      "loss": 1.0752,
+      "step": 5734
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.519018501052555e-05,
+      "loss": 1.1108,
+      "step": 5735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.515936420434351e-05,
+      "loss": 1.0773,
+      "step": 5736
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.512854591280367e-05,
+      "loss": 1.1877,
+      "step": 5737
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.50977301390258e-05,
+      "loss": 1.1296,
+      "step": 5738
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.506691688612946e-05,
+      "loss": 1.0924,
+      "step": 5739
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.503610615723387e-05,
+      "loss": 1.172,
+      "step": 5740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.500529795545804e-05,
+      "loss": 1.1465,
+      "step": 5741
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.49744922839207e-05,
+      "loss": 1.1915,
+      "step": 5742
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.494368914574037e-05,
+      "loss": 1.1274,
+      "step": 5743
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.491288854403528e-05,
+      "loss": 1.1056,
+      "step": 5744
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.488209048192336e-05,
+      "loss": 1.08,
+      "step": 5745
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.485129496252236e-05,
+      "loss": 1.1137,
+      "step": 5746
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.482050198894974e-05,
+      "loss": 1.1123,
+      "step": 5747
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.47897115643227e-05,
+      "loss": 1.1281,
+      "step": 5748
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.475892369175819e-05,
+      "loss": 1.0845,
+      "step": 5749
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.47281383743729e-05,
+      "loss": 1.1258,
+      "step": 5750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.469735561528321e-05,
+      "loss": 1.1249,
+      "step": 5751
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.466657541760536e-05,
+      "loss": 1.1769,
+      "step": 5752
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.463579778445522e-05,
+      "loss": 1.0733,
+      "step": 5753
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.460502271894845e-05,
+      "loss": 1.1571,
+      "step": 5754
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.457425022420046e-05,
+      "loss": 1.1359,
+      "step": 5755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.454348030332636e-05,
+      "loss": 1.0476,
+      "step": 5756
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.451271295944102e-05,
+      "loss": 1.1542,
+      "step": 5757
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.448194819565904e-05,
+      "loss": 1.1585,
+      "step": 5758
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.445118601509479e-05,
+      "loss": 1.099,
+      "step": 5759
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.442042642086233e-05,
+      "loss": 1.1534,
+      "step": 5760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.438966941607552e-05,
+      "loss": 1.1207,
+      "step": 5761
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.435891500384789e-05,
+      "loss": 1.0421,
+      "step": 5762
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.432816318729274e-05,
+      "loss": 1.1645,
+      "step": 5763
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.429741396952313e-05,
+      "loss": 1.0981,
+      "step": 5764
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.42666673536518e-05,
+      "loss": 1.1043,
+      "step": 5765
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.423592334279126e-05,
+      "loss": 1.1691,
+      "step": 5766
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.420518194005381e-05,
+      "loss": 1.1332,
+      "step": 5767
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.417444314855142e-05,
+      "loss": 1.1359,
+      "step": 5768
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.414370697139577e-05,
+      "loss": 1.1851,
+      "step": 5769
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.411297341169833e-05,
+      "loss": 1.173,
+      "step": 5770
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.408224247257029e-05,
+      "loss": 1.1501,
+      "step": 5771
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.405151415712257e-05,
+      "loss": 1.0427,
+      "step": 5772
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.402078846846585e-05,
+      "loss": 1.1267,
+      "step": 5773
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.39900654097105e-05,
+      "loss": 1.137,
+      "step": 5774
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.395934498396663e-05,
+      "loss": 1.1681,
+      "step": 5775
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.392862719434415e-05,
+      "loss": 1.125,
+      "step": 5776
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.38979120439526e-05,
+      "loss": 1.1494,
+      "step": 5777
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.386719953590135e-05,
+      "loss": 1.1441,
+      "step": 5778
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.383648967329942e-05,
+      "loss": 1.1186,
+      "step": 5779
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.380578245925555e-05,
+      "loss": 1.0952,
+      "step": 5780
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.377507789687842e-05,
+      "loss": 1.0912,
+      "step": 5781
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.374437598927616e-05,
+      "loss": 1.1124,
+      "step": 5782
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.371367673955679e-05,
+      "loss": 1.1235,
+      "step": 5783
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.368298015082799e-05,
+      "loss": 1.1587,
+      "step": 5784
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.365228622619726e-05,
+      "loss": 1.1216,
+      "step": 5785
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.362159496877177e-05,
+      "loss": 1.0915,
+      "step": 5786
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.35909063816584e-05,
+      "loss": 1.1147,
+      "step": 5787
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.356022046796377e-05,
+      "loss": 1.1245,
+      "step": 5788
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.352953723079428e-05,
+      "loss": 1.1353,
+      "step": 5789
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.349885667325605e-05,
+      "loss": 1.1012,
+      "step": 5790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.346817879845482e-05,
+      "loss": 1.1103,
+      "step": 5791
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.343750360949622e-05,
+      "loss": 1.1075,
+      "step": 5792
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.340683110948545e-05,
+      "loss": 1.1289,
+      "step": 5793
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.337616130152762e-05,
+      "loss": 1.0499,
+      "step": 5794
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.334549418872739e-05,
+      "loss": 1.1273,
+      "step": 5795
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.331482977418924e-05,
+      "loss": 1.1052,
+      "step": 5796
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.328416806101739e-05,
+      "loss": 1.1635,
+      "step": 5797
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.325350905231573e-05,
+      "loss": 1.091,
+      "step": 5798
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.322285275118788e-05,
+      "loss": 1.1294,
+      "step": 5799
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.319219916073725e-05,
+      "loss": 1.1295,
+      "step": 5800
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.31615482840669e-05,
+      "loss": 1.1292,
+      "step": 5801
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.313090012427965e-05,
+      "loss": 1.0901,
+      "step": 5802
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.310025468447808e-05,
+      "loss": 1.1022,
+      "step": 5803
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.306961196776443e-05,
+      "loss": 1.103,
+      "step": 5804
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.303897197724068e-05,
+      "loss": 1.0794,
+      "step": 5805
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.300833471600856e-05,
+      "loss": 1.1405,
+      "step": 5806
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.297770018716949e-05,
+      "loss": 1.1356,
+      "step": 5807
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.294706839382471e-05,
+      "loss": 1.0212,
+      "step": 5808
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.291643933907501e-05,
+      "loss": 1.0785,
+      "step": 5809
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.288581302602106e-05,
+      "loss": 1.1645,
+      "step": 5810
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.285518945776322e-05,
+      "loss": 1.1036,
+      "step": 5811
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.282456863740145e-05,
+      "loss": 1.0456,
+      "step": 5812
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.279395056803559e-05,
+      "loss": 1.09,
+      "step": 5813
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.27633352527651e-05,
+      "loss": 1.1386,
+      "step": 5814
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.273272269468927e-05,
+      "loss": 1.1496,
+      "step": 5815
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.270211289690696e-05,
+      "loss": 1.0695,
+      "step": 5816
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.267150586251687e-05,
+      "loss": 1.0771,
+      "step": 5817
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.264090159461736e-05,
+      "loss": 1.0746,
+      "step": 5818
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.261030009630655e-05,
+      "loss": 1.0208,
+      "step": 5819
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.257970137068227e-05,
+      "loss": 1.0549,
+      "step": 5820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.254910542084199e-05,
+      "loss": 1.1108,
+      "step": 5821
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.251851224988304e-05,
+      "loss": 1.1007,
+      "step": 5822
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.248792186090242e-05,
+      "loss": 1.119,
+      "step": 5823
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.245733425699674e-05,
+      "loss": 1.0999,
+      "step": 5824
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.242674944126248e-05,
+      "loss": 1.0916,
+      "step": 5825
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.239616741679573e-05,
+      "loss": 1.1806,
+      "step": 5826
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.236558818669237e-05,
+      "loss": 1.0487,
+      "step": 5827
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.233501175404798e-05,
+      "loss": 1.0777,
+      "step": 5828
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.230443812195777e-05,
+      "loss": 1.0663,
+      "step": 5829
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.22738672935168e-05,
+      "loss": 1.1516,
+      "step": 5830
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.224329927181976e-05,
+      "loss": 1.0892,
+      "step": 5831
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.22127340599611e-05,
+      "loss": 1.0587,
+      "step": 5832
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.218217166103496e-05,
+      "loss": 1.1046,
+      "step": 5833
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.215161207813521e-05,
+      "loss": 1.1115,
+      "step": 5834
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.212105531435537e-05,
+      "loss": 1.0904,
+      "step": 5835
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.209050137278881e-05,
+      "loss": 1.1363,
+      "step": 5836
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.20599502565285e-05,
+      "loss": 1.102,
+      "step": 5837
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.202940196866717e-05,
+      "loss": 1.0992,
+      "step": 5838
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.199885651229723e-05,
+      "loss": 1.189,
+      "step": 5839
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.196831389051087e-05,
+      "loss": 1.1562,
+      "step": 5840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.193777410639991e-05,
+      "loss": 1.1421,
+      "step": 5841
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.190723716305595e-05,
+      "loss": 1.0525,
+      "step": 5842
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.187670306357023e-05,
+      "loss": 1.1012,
+      "step": 5843
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.184617181103379e-05,
+      "loss": 1.1216,
+      "step": 5844
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.181564340853736e-05,
+      "loss": 1.1121,
+      "step": 5845
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.178511785917131e-05,
+      "loss": 1.1607,
+      "step": 5846
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.175459516602576e-05,
+      "loss": 1.1028,
+      "step": 5847
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.172407533219059e-05,
+      "loss": 1.1551,
+      "step": 5848
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.169355836075536e-05,
+      "loss": 1.1375,
+      "step": 5849
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.166304425480931e-05,
+      "loss": 1.1506,
+      "step": 5850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.163253301744143e-05,
+      "loss": 1.1001,
+      "step": 5851
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.160202465174039e-05,
+      "loss": 1.0803,
+      "step": 5852
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.15715191607946e-05,
+      "loss": 1.1225,
+      "step": 5853
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.154101654769214e-05,
+      "loss": 1.1024,
+      "step": 5854
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.151051681552084e-05,
+      "loss": 1.1267,
+      "step": 5855
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.148001996736819e-05,
+      "loss": 1.157,
+      "step": 5856
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.144952600632144e-05,
+      "loss": 1.1309,
+      "step": 5857
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.141903493546753e-05,
+      "loss": 1.1363,
+      "step": 5858
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.138854675789306e-05,
+      "loss": 1.0456,
+      "step": 5859
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.135806147668441e-05,
+      "loss": 1.1324,
+      "step": 5860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.132757909492764e-05,
+      "loss": 1.1562,
+      "step": 5861
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.129709961570847e-05,
+      "loss": 1.121,
+      "step": 5862
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.126662304211244e-05,
+      "loss": 1.1064,
+      "step": 5863
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.123614937722467e-05,
+      "loss": 1.1099,
+      "step": 5864
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.120567862413008e-05,
+      "loss": 1.1367,
+      "step": 5865
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.11752107859132e-05,
+      "loss": 1.0952,
+      "step": 5866
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.114474586565836e-05,
+      "loss": 1.0839,
+      "step": 5867
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.111428386644954e-05,
+      "loss": 1.0912,
+      "step": 5868
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.108382479137045e-05,
+      "loss": 1.0853,
+      "step": 5869
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.105336864350449e-05,
+      "loss": 1.1509,
+      "step": 5870
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.102291542593473e-05,
+      "loss": 1.0722,
+      "step": 5871
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.099246514174402e-05,
+      "loss": 1.1001,
+      "step": 5872
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.096201779401488e-05,
+      "loss": 1.052,
+      "step": 5873
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.093157338582947e-05,
+      "loss": 1.1191,
+      "step": 5874
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.09011319202698e-05,
+      "loss": 1.1142,
+      "step": 5875
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.087069340041734e-05,
+      "loss": 1.1189,
+      "step": 5876
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.084025782935358e-05,
+      "loss": 1.1333,
+      "step": 5877
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.080982521015946e-05,
+      "loss": 1.1368,
+      "step": 5878
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.07793955459157e-05,
+      "loss": 1.0589,
+      "step": 5879
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.074896883970274e-05,
+      "loss": 1.0293,
+      "step": 5880
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.071854509460072e-05,
+      "loss": 1.1044,
+      "step": 5881
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.068812431368945e-05,
+      "loss": 1.0813,
+      "step": 5882
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.065770650004843e-05,
+      "loss": 1.1731,
+      "step": 5883
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.062729165675695e-05,
+      "loss": 1.1651,
+      "step": 5884
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.059687978689385e-05,
+      "loss": 1.111,
+      "step": 5885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.056647089353783e-05,
+      "loss": 1.1558,
+      "step": 5886
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.053606497976719e-05,
+      "loss": 1.1343,
+      "step": 5887
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.050566204865992e-05,
+      "loss": 1.067,
+      "step": 5888
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.047526210329376e-05,
+      "loss": 1.0488,
+      "step": 5889
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.04448651467461e-05,
+      "loss": 1.1325,
+      "step": 5890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.041447118209412e-05,
+      "loss": 1.1303,
+      "step": 5891
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.03840802124146e-05,
+      "loss": 1.102,
+      "step": 5892
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.035369224078403e-05,
+      "loss": 1.0307,
+      "step": 5893
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.032330727027862e-05,
+      "loss": 1.1856,
+      "step": 5894
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.029292530397431e-05,
+      "loss": 1.0571,
+      "step": 5895
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.026254634494662e-05,
+      "loss": 1.1662,
+      "step": 5896
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.023217039627092e-05,
+      "loss": 1.1158,
+      "step": 5897
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.020179746102213e-05,
+      "loss": 1.1075,
+      "step": 5898
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.0171427542275e-05,
+      "loss": 1.0934,
+      "step": 5899
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.014106064310387e-05,
+      "loss": 1.0679,
+      "step": 5900
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.011069676658279e-05,
+      "loss": 1.1491,
+      "step": 5901
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.008033591578555e-05,
+      "loss": 1.0927,
+      "step": 5902
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.00499780937856e-05,
+      "loss": 1.1564,
+      "step": 5903
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.001962330365612e-05,
+      "loss": 1.0492,
+      "step": 5904
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.998927154846993e-05,
+      "loss": 1.1156,
+      "step": 5905
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.995892283129959e-05,
+      "loss": 1.0996,
+      "step": 5906
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.992857715521732e-05,
+      "loss": 1.1546,
+      "step": 5907
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.989823452329503e-05,
+      "loss": 1.1561,
+      "step": 5908
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.986789493860433e-05,
+      "loss": 1.0822,
+      "step": 5909
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.983755840421653e-05,
+      "loss": 1.045,
+      "step": 5910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.980722492320267e-05,
+      "loss": 1.1359,
+      "step": 5911
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.97768944986334e-05,
+      "loss": 1.082,
+      "step": 5912
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.974656713357908e-05,
+      "loss": 1.1048,
+      "step": 5913
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.971624283110983e-05,
+      "loss": 1.1391,
+      "step": 5914
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.968592159429535e-05,
+      "loss": 1.1311,
+      "step": 5915
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.965560342620515e-05,
+      "loss": 1.1635,
+      "step": 5916
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.96252883299083e-05,
+      "loss": 1.0777,
+      "step": 5917
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.95949763084737e-05,
+      "loss": 1.1723,
+      "step": 5918
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.956466736496983e-05,
+      "loss": 1.1119,
+      "step": 5919
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.953436150246491e-05,
+      "loss": 1.0658,
+      "step": 5920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.950405872402681e-05,
+      "loss": 1.133,
+      "step": 5921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.947375903272311e-05,
+      "loss": 1.121,
+      "step": 5922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.94434624316211e-05,
+      "loss": 1.1134,
+      "step": 5923
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.941316892378775e-05,
+      "loss": 1.1356,
+      "step": 5924
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.938287851228964e-05,
+      "loss": 1.0903,
+      "step": 5925
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.935259120019314e-05,
+      "loss": 1.1125,
+      "step": 5926
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.932230699056426e-05,
+      "loss": 1.135,
+      "step": 5927
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.929202588646871e-05,
+      "loss": 1.0979,
+      "step": 5928
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.926174789097186e-05,
+      "loss": 1.0656,
+      "step": 5929
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.923147300713878e-05,
+      "loss": 1.1316,
+      "step": 5930
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.920120123803419e-05,
+      "loss": 1.0959,
+      "step": 5931
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.917093258672264e-05,
+      "loss": 1.1255,
+      "step": 5932
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.914066705626816e-05,
+      "loss": 1.133,
+      "step": 5933
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.91104046497346e-05,
+      "loss": 1.1412,
+      "step": 5934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.908014537018544e-05,
+      "loss": 1.0444,
+      "step": 5935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.904988922068384e-05,
+      "loss": 1.0427,
+      "step": 5936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.901963620429274e-05,
+      "loss": 1.1292,
+      "step": 5937
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.898938632407459e-05,
+      "loss": 1.1731,
+      "step": 5938
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.895913958309164e-05,
+      "loss": 1.1476,
+      "step": 5939
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.892889598440581e-05,
+      "loss": 1.0723,
+      "step": 5940
+    },
+    {
+      "epoch": 2.38,
+      "eval_loss": 0.8191943168640137,
+      "eval_runtime": 504.3926,
+      "eval_samples_per_second": 51.349,
+      "eval_steps_per_second": 25.674,
+      "step": 5940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.889865553107868e-05,
+      "loss": 1.0869,
+      "step": 5941
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.886841822617155e-05,
+      "loss": 1.1065,
+      "step": 5942
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.883818407274533e-05,
+      "loss": 1.1646,
+      "step": 5943
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.880795307386067e-05,
+      "loss": 1.1666,
+      "step": 5944
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.877772523257787e-05,
+      "loss": 1.168,
+      "step": 5945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.874750055195695e-05,
+      "loss": 1.1241,
+      "step": 5946
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.87172790350576e-05,
+      "loss": 1.1273,
+      "step": 5947
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.868706068493913e-05,
+      "loss": 1.0855,
+      "step": 5948
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.865684550466063e-05,
+      "loss": 1.0735,
+      "step": 5949
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.862663349728073e-05,
+      "loss": 1.1519,
+      "step": 5950
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.859642466585789e-05,
+      "loss": 1.0964,
+      "step": 5951
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.856621901345015e-05,
+      "loss": 1.1055,
+      "step": 5952
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.853601654311526e-05,
+      "loss": 1.1413,
+      "step": 5953
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.850581725791069e-05,
+      "loss": 1.0818,
+      "step": 5954
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.847562116089347e-05,
+      "loss": 1.0868,
+      "step": 5955
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.844542825512041e-05,
+      "loss": 1.148,
+      "step": 5956
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.8415238543648e-05,
+      "loss": 1.0844,
+      "step": 5957
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.83850520295323e-05,
+      "loss": 1.012,
+      "step": 5958
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.83548687158292e-05,
+      "loss": 1.0475,
+      "step": 5959
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.832468860559415e-05,
+      "loss": 1.1155,
+      "step": 5960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.829451170188233e-05,
+      "loss": 1.0841,
+      "step": 5961
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.826433800774859e-05,
+      "loss": 1.0777,
+      "step": 5962
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.823416752624739e-05,
+      "loss": 1.1376,
+      "step": 5963
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.820400026043294e-05,
+      "loss": 1.1099,
+      "step": 5964
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.817383621335911e-05,
+      "loss": 1.1869,
+      "step": 5965
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.814367538807945e-05,
+      "loss": 1.134,
+      "step": 5966
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.811351778764717e-05,
+      "loss": 1.1083,
+      "step": 5967
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.808336341511512e-05,
+      "loss": 1.1152,
+      "step": 5968
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.805321227353587e-05,
+      "loss": 1.1048,
+      "step": 5969
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.802306436596167e-05,
+      "loss": 1.0361,
+      "step": 5970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.799291969544443e-05,
+      "loss": 1.1593,
+      "step": 5971
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.796277826503563e-05,
+      "loss": 1.182,
+      "step": 5972
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.793264007778663e-05,
+      "loss": 1.1335,
+      "step": 5973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.790250513674836e-05,
+      "loss": 1.0823,
+      "step": 5974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.787237344497132e-05,
+      "loss": 1.086,
+      "step": 5975
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.78422450055058e-05,
+      "loss": 1.1116,
+      "step": 5976
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.781211982140176e-05,
+      "loss": 1.0805,
+      "step": 5977
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.778199789570878e-05,
+      "loss": 1.0788,
+      "step": 5978
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.775187923147617e-05,
+      "loss": 1.0721,
+      "step": 5979
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.77217638317528e-05,
+      "loss": 1.1389,
+      "step": 5980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.769165169958733e-05,
+      "loss": 1.131,
+      "step": 5981
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.766154283802804e-05,
+      "loss": 1.0831,
+      "step": 5982
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.763143725012286e-05,
+      "loss": 1.0733,
+      "step": 5983
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.760133493891945e-05,
+      "loss": 1.0877,
+      "step": 5984
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.757123590746505e-05,
+      "loss": 1.0626,
+      "step": 5985
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.754114015880658e-05,
+      "loss": 1.1347,
+      "step": 5986
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.751104769599079e-05,
+      "loss": 1.0637,
+      "step": 5987
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.748095852206387e-05,
+      "loss": 1.131,
+      "step": 5988
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.74508726400718e-05,
+      "loss": 1.1203,
+      "step": 5989
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.74207900530602e-05,
+      "loss": 1.1036,
+      "step": 5990
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.739071076407442e-05,
+      "loss": 1.1532,
+      "step": 5991
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.736063477615931e-05,
+      "loss": 1.1733,
+      "step": 5992
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.733056209235955e-05,
+      "loss": 1.165,
+      "step": 5993
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.730049271571943e-05,
+      "loss": 1.107,
+      "step": 5994
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.727042664928289e-05,
+      "loss": 1.0936,
+      "step": 5995
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.724036389609357e-05,
+      "loss": 1.1432,
+      "step": 5996
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.721030445919471e-05,
+      "loss": 1.1216,
+      "step": 5997
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.71802483416293e-05,
+      "loss": 1.1225,
+      "step": 5998
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.715019554643993e-05,
+      "loss": 1.1177,
+      "step": 5999
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.712014607666885e-05,
+      "loss": 1.0769,
+      "step": 6000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.709009993535804e-05,
+      "loss": 1.0806,
+      "step": 6001
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.70600571255491e-05,
+      "loss": 1.1091,
+      "step": 6002
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.703001765028327e-05,
+      "loss": 1.2237,
+      "step": 6003
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.69999815126015e-05,
+      "loss": 1.1715,
+      "step": 6004
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.696994871554435e-05,
+      "loss": 1.1372,
+      "step": 6005
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.693991926215207e-05,
+      "loss": 1.1703,
+      "step": 6006
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.690989315546459e-05,
+      "loss": 1.1031,
+      "step": 6007
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.687987039852148e-05,
+      "loss": 1.1243,
+      "step": 6008
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.684985099436201e-05,
+      "loss": 1.075,
+      "step": 6009
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.681983494602498e-05,
+      "loss": 1.1733,
+      "step": 6010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.678982225654902e-05,
+      "loss": 1.2014,
+      "step": 6011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.675981292897231e-05,
+      "loss": 1.1277,
+      "step": 6012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.672980696633272e-05,
+      "loss": 1.1325,
+      "step": 6013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.669980437166783e-05,
+      "loss": 1.1237,
+      "step": 6014
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.66698051480148e-05,
+      "loss": 1.1007,
+      "step": 6015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.66398092984105e-05,
+      "loss": 1.0906,
+      "step": 6016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.660981682589139e-05,
+      "loss": 1.1065,
+      "step": 6017
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.65798277334937e-05,
+      "loss": 1.1038,
+      "step": 6018
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.65498420242532e-05,
+      "loss": 1.1475,
+      "step": 6019
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.65198597012054e-05,
+      "loss": 1.1063,
+      "step": 6020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.648988076738548e-05,
+      "loss": 1.0797,
+      "step": 6021
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.645990522582817e-05,
+      "loss": 1.0551,
+      "step": 6022
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.642993307956795e-05,
+      "loss": 1.119,
+      "step": 6023
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.639996433163893e-05,
+      "loss": 1.043,
+      "step": 6024
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.636999898507488e-05,
+      "loss": 0.9932,
+      "step": 6025
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.634003704290923e-05,
+      "loss": 1.0826,
+      "step": 6026
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.6310078508175e-05,
+      "loss": 1.1151,
+      "step": 6027
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.6280123383905e-05,
+      "loss": 1.1738,
+      "step": 6028
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.625017167313161e-05,
+      "loss": 1.0985,
+      "step": 6029
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.622022337888683e-05,
+      "loss": 1.1068,
+      "step": 6030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.619027850420237e-05,
+      "loss": 1.1299,
+      "step": 6031
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.616033705210957e-05,
+      "loss": 1.1604,
+      "step": 6032
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.613039902563947e-05,
+      "loss": 1.123,
+      "step": 6033
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.61004644278227e-05,
+      "loss": 1.102,
+      "step": 6034
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.607053326168956e-05,
+      "loss": 1.1156,
+      "step": 6035
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.604060553027002e-05,
+      "loss": 1.1209,
+      "step": 6036
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.601068123659369e-05,
+      "loss": 1.0958,
+      "step": 6037
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.598076038368988e-05,
+      "loss": 1.0483,
+      "step": 6038
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.595084297458742e-05,
+      "loss": 1.0231,
+      "step": 6039
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.592092901231494e-05,
+      "loss": 1.0809,
+      "step": 6040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.589101849990065e-05,
+      "loss": 1.1434,
+      "step": 6041
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.586111144037242e-05,
+      "loss": 1.0902,
+      "step": 6042
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.583120783675777e-05,
+      "loss": 1.1345,
+      "step": 6043
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.580130769208388e-05,
+      "loss": 1.1331,
+      "step": 6044
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.577141100937756e-05,
+      "loss": 1.1633,
+      "step": 6045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.57415177916653e-05,
+      "loss": 1.1075,
+      "step": 6046
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.57116280419732e-05,
+      "loss": 1.1163,
+      "step": 6047
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.568174176332701e-05,
+      "loss": 1.1402,
+      "step": 6048
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.565185895875219e-05,
+      "loss": 1.1291,
+      "step": 6049
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.562197963127378e-05,
+      "loss": 1.1119,
+      "step": 6050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.559210378391651e-05,
+      "loss": 1.1241,
+      "step": 6051
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.556223141970474e-05,
+      "loss": 1.0377,
+      "step": 6052
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.553236254166246e-05,
+      "loss": 1.1223,
+      "step": 6053
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.550249715281333e-05,
+      "loss": 1.0228,
+      "step": 6054
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.547263525618064e-05,
+      "loss": 1.0681,
+      "step": 6055
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.54427768547874e-05,
+      "loss": 1.1168,
+      "step": 6056
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.541292195165614e-05,
+      "loss": 1.1838,
+      "step": 6057
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.538307054980915e-05,
+      "loss": 1.099,
+      "step": 6058
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.535322265226828e-05,
+      "loss": 1.0912,
+      "step": 6059
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.532337826205509e-05,
+      "loss": 1.101,
+      "step": 6060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.529353738219073e-05,
+      "loss": 1.1582,
+      "step": 6061
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.526370001569605e-05,
+      "loss": 1.0998,
+      "step": 6062
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.523386616559151e-05,
+      "loss": 1.1725,
+      "step": 6063
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.52040358348972e-05,
+      "loss": 1.0884,
+      "step": 6064
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.517420902663289e-05,
+      "loss": 1.1238,
+      "step": 6065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.514438574381796e-05,
+      "loss": 1.1166,
+      "step": 6066
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.511456598947148e-05,
+      "loss": 1.0953,
+      "step": 6067
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.508474976661208e-05,
+      "loss": 1.149,
+      "step": 6068
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.505493707825816e-05,
+      "loss": 1.059,
+      "step": 6069
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.502512792742762e-05,
+      "loss": 1.0898,
+      "step": 6070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.499532231713816e-05,
+      "loss": 1.2477,
+      "step": 6071
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.496552025040694e-05,
+      "loss": 1.1077,
+      "step": 6072
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.493572173025088e-05,
+      "loss": 1.1299,
+      "step": 6073
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.490592675968655e-05,
+      "loss": 1.1374,
+      "step": 6074
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.487613534173006e-05,
+      "loss": 1.0646,
+      "step": 6075
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.484634747939729e-05,
+      "loss": 1.0788,
+      "step": 6076
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.481656317570363e-05,
+      "loss": 1.1509,
+      "step": 6077
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.478678243366423e-05,
+      "loss": 1.0607,
+      "step": 6078
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.47570052562938e-05,
+      "loss": 1.1468,
+      "step": 6079
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.47272316466067e-05,
+      "loss": 1.0883,
+      "step": 6080
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.469746160761699e-05,
+      "loss": 1.1793,
+      "step": 6081
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.466769514233823e-05,
+      "loss": 1.1117,
+      "step": 6082
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.463793225378382e-05,
+      "loss": 1.171,
+      "step": 6083
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.460817294496662e-05,
+      "loss": 1.0937,
+      "step": 6084
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.457841721889922e-05,
+      "loss": 1.0771,
+      "step": 6085
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.454866507859378e-05,
+      "loss": 1.1276,
+      "step": 6086
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.451891652706219e-05,
+      "loss": 1.1079,
+      "step": 6087
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.448917156731592e-05,
+      "loss": 1.171,
+      "step": 6088
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.445943020236604e-05,
+      "loss": 1.0972,
+      "step": 6089
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.442969243522334e-05,
+      "loss": 1.1071,
+      "step": 6090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.439995826889817e-05,
+      "loss": 1.1315,
+      "step": 6091
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.437022770640058e-05,
+      "loss": 1.0488,
+      "step": 6092
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.434050075074023e-05,
+      "loss": 1.1313,
+      "step": 6093
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.431077740492636e-05,
+      "loss": 1.0998,
+      "step": 6094
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.428105767196793e-05,
+      "loss": 1.1029,
+      "step": 6095
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.425134155487347e-05,
+      "loss": 1.1642,
+      "step": 6096
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.422162905665122e-05,
+      "loss": 1.1392,
+      "step": 6097
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.419192018030897e-05,
+      "loss": 1.1184,
+      "step": 6098
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.41622149288542e-05,
+      "loss": 1.125,
+      "step": 6099
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.4132513305294e-05,
+      "loss": 1.1193,
+      "step": 6100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.410281531263509e-05,
+      "loss": 1.1488,
+      "step": 6101
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.40731209538838e-05,
+      "loss": 1.0494,
+      "step": 6102
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.404343023204615e-05,
+      "loss": 1.1452,
+      "step": 6103
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.401374315012776e-05,
+      "loss": 1.0884,
+      "step": 6104
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.398405971113388e-05,
+      "loss": 1.0966,
+      "step": 6105
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.395437991806938e-05,
+      "loss": 1.0746,
+      "step": 6106
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.39247037739388e-05,
+      "loss": 1.1061,
+      "step": 6107
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.389503128174625e-05,
+      "loss": 1.1085,
+      "step": 6108
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.386536244449553e-05,
+      "loss": 1.1117,
+      "step": 6109
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.383569726519003e-05,
+      "loss": 1.0948,
+      "step": 6110
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.380603574683281e-05,
+      "loss": 1.1887,
+      "step": 6111
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.377637789242655e-05,
+      "loss": 1.1021,
+      "step": 6112
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.37467237049735e-05,
+      "loss": 1.0981,
+      "step": 6113
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.37170731874756e-05,
+      "loss": 1.1403,
+      "step": 6114
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.368742634293441e-05,
+      "loss": 1.0696,
+      "step": 6115
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.365778317435109e-05,
+      "loss": 1.1693,
+      "step": 6116
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.362814368472645e-05,
+      "loss": 1.0954,
+      "step": 6117
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.359850787706096e-05,
+      "loss": 1.0892,
+      "step": 6118
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.356887575435464e-05,
+      "loss": 1.1252,
+      "step": 6119
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.353924731960719e-05,
+      "loss": 1.0958,
+      "step": 6120
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.350962257581794e-05,
+      "loss": 1.1368,
+      "step": 6121
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.348000152598581e-05,
+      "loss": 1.1102,
+      "step": 6122
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.345038417310938e-05,
+      "loss": 1.0862,
+      "step": 6123
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.342077052018684e-05,
+      "loss": 1.0862,
+      "step": 6124
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.339116057021606e-05,
+      "loss": 1.1114,
+      "step": 6125
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.33615543261944e-05,
+      "loss": 1.0877,
+      "step": 6126
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.333195179111898e-05,
+      "loss": 1.0977,
+      "step": 6127
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.330235296798647e-05,
+      "loss": 1.0712,
+      "step": 6128
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.32727578597932e-05,
+      "loss": 1.14,
+      "step": 6129
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.324316646953517e-05,
+      "loss": 1.1452,
+      "step": 6130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.321357880020783e-05,
+      "loss": 1.0835,
+      "step": 6131
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.318399485480644e-05,
+      "loss": 1.1171,
+      "step": 6132
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.31544146363258e-05,
+      "loss": 1.1488,
+      "step": 6133
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.312483814776034e-05,
+      "loss": 1.1156,
+      "step": 6134
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.309526539210416e-05,
+      "loss": 1.1041,
+      "step": 6135
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.306569637235086e-05,
+      "loss": 1.1364,
+      "step": 6136
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.303613109149375e-05,
+      "loss": 1.1157,
+      "step": 6137
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.300656955252587e-05,
+      "loss": 1.0538,
+      "step": 6138
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.297701175843963e-05,
+      "loss": 1.111,
+      "step": 6139
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.294745771222726e-05,
+      "loss": 1.1891,
+      "step": 6140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.291790741688053e-05,
+      "loss": 1.0823,
+      "step": 6141
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.288836087539083e-05,
+      "loss": 1.125,
+      "step": 6142
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.285881809074922e-05,
+      "loss": 1.0885,
+      "step": 6143
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.282927906594633e-05,
+      "loss": 1.0442,
+      "step": 6144
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.279974380397243e-05,
+      "loss": 1.0506,
+      "step": 6145
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.277021230781736e-05,
+      "loss": 1.0744,
+      "step": 6146
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.274068458047069e-05,
+      "loss": 1.1232,
+      "step": 6147
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.27111606249215e-05,
+      "loss": 1.1159,
+      "step": 6148
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.268164044415854e-05,
+      "loss": 1.0869,
+      "step": 6149
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.265212404117015e-05,
+      "loss": 1.1203,
+      "step": 6150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.262261141894432e-05,
+      "loss": 1.1205,
+      "step": 6151
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.259310258046867e-05,
+      "loss": 1.0615,
+      "step": 6152
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.256359752873036e-05,
+      "loss": 1.1264,
+      "step": 6153
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.253409626671623e-05,
+      "loss": 1.1664,
+      "step": 6154
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.250459879741277e-05,
+      "loss": 1.0982,
+      "step": 6155
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.247510512380597e-05,
+      "loss": 1.0969,
+      "step": 6156
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.244561524888153e-05,
+      "loss": 1.0891,
+      "step": 6157
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.241612917562475e-05,
+      "loss": 1.0912,
+      "step": 6158
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.238664690702051e-05,
+      "loss": 1.1424,
+      "step": 6159
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.23571684460534e-05,
+      "loss": 1.0748,
+      "step": 6160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.232769379570744e-05,
+      "loss": 1.0904,
+      "step": 6161
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.229822295896645e-05,
+      "loss": 1.0674,
+      "step": 6162
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.226875593881378e-05,
+      "loss": 1.1777,
+      "step": 6163
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.22392927382324e-05,
+      "loss": 1.0621,
+      "step": 6164
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.22098333602049e-05,
+      "loss": 1.1434,
+      "step": 6165
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.218037780771349e-05,
+      "loss": 1.1213,
+      "step": 6166
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.215092608373997e-05,
+      "loss": 1.0829,
+      "step": 6167
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.212147819126581e-05,
+      "loss": 1.0997,
+      "step": 6168
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.2092034133272e-05,
+      "loss": 1.1117,
+      "step": 6169
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.206259391273921e-05,
+      "loss": 1.1585,
+      "step": 6170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.203315753264769e-05,
+      "loss": 1.119,
+      "step": 6171
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.200372499597735e-05,
+      "loss": 1.13,
+      "step": 6172
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.197429630570763e-05,
+      "loss": 1.0916,
+      "step": 6173
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.194487146481767e-05,
+      "loss": 1.0339,
+      "step": 6174
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.191545047628612e-05,
+      "loss": 1.089,
+      "step": 6175
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.188603334309133e-05,
+      "loss": 1.1175,
+      "step": 6176
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.185662006821125e-05,
+      "loss": 1.0579,
+      "step": 6177
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.182721065462333e-05,
+      "loss": 1.14,
+      "step": 6178
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.17978051053048e-05,
+      "loss": 1.089,
+      "step": 6179
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.176840342323242e-05,
+      "loss": 1.167,
+      "step": 6180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.173900561138249e-05,
+      "loss": 1.101,
+      "step": 6181
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.170961167273102e-05,
+      "loss": 1.105,
+      "step": 6182
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.168022161025355e-05,
+      "loss": 1.1293,
+      "step": 6183
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.165083542692529e-05,
+      "loss": 1.1236,
+      "step": 6184
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.162145312572108e-05,
+      "loss": 1.1044,
+      "step": 6185
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.159207470961522e-05,
+      "loss": 1.0631,
+      "step": 6186
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.156270018158177e-05,
+      "loss": 1.1049,
+      "step": 6187
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.153332954459437e-05,
+      "loss": 1.1371,
+      "step": 6188
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.150396280162618e-05,
+      "loss": 1.1144,
+      "step": 6189
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.147459995565008e-05,
+      "loss": 1.1235,
+      "step": 6190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.144524100963844e-05,
+      "loss": 1.152,
+      "step": 6191
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.141588596656333e-05,
+      "loss": 1.1293,
+      "step": 6192
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.13865348293964e-05,
+      "loss": 1.1015,
+      "step": 6193
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.135718760110888e-05,
+      "loss": 1.0859,
+      "step": 6194
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.132784428467163e-05,
+      "loss": 1.1082,
+      "step": 6195
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.129850488305509e-05,
+      "loss": 1.0811,
+      "step": 6196
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.126916939922933e-05,
+      "loss": 1.1634,
+      "step": 6197
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.1239837836164e-05,
+      "loss": 1.0997,
+      "step": 6198
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.121051019682837e-05,
+      "loss": 1.0752,
+      "step": 6199
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.118118648419128e-05,
+      "loss": 1.1397,
+      "step": 6200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.115186670122123e-05,
+      "loss": 1.1422,
+      "step": 6201
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.112255085088632e-05,
+      "loss": 1.0626,
+      "step": 6202
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.109323893615414e-05,
+      "loss": 1.1642,
+      "step": 6203
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.1063930959992e-05,
+      "loss": 1.1022,
+      "step": 6204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.103462692536681e-05,
+      "loss": 1.0804,
+      "step": 6205
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.100532683524498e-05,
+      "loss": 1.1538,
+      "step": 6206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.097603069259265e-05,
+      "loss": 1.1056,
+      "step": 6207
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0946738500375485e-05,
+      "loss": 1.0912,
+      "step": 6208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.091745026155875e-05,
+      "loss": 1.0992,
+      "step": 6209
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0888165979107334e-05,
+      "loss": 1.1581,
+      "step": 6210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0858885655985696e-05,
+      "loss": 1.1587,
+      "step": 6211
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0829609295157916e-05,
+      "loss": 1.1139,
+      "step": 6212
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.080033689958768e-05,
+      "loss": 1.1314,
+      "step": 6213
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.077106847223826e-05,
+      "loss": 1.1304,
+      "step": 6214
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0741804016072524e-05,
+      "loss": 1.1386,
+      "step": 6215
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0712543534052934e-05,
+      "loss": 1.1193,
+      "step": 6216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0683287029141565e-05,
+      "loss": 1.191,
+      "step": 6217
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.065403450430007e-05,
+      "loss": 1.1728,
+      "step": 6218
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.062478596248975e-05,
+      "loss": 1.1872,
+      "step": 6219
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0595541406671385e-05,
+      "loss": 1.1065,
+      "step": 6220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.056630083980552e-05,
+      "loss": 1.1069,
+      "step": 6221
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0537064264852195e-05,
+      "loss": 1.1591,
+      "step": 6222
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0507831684771e-05,
+      "loss": 1.0463,
+      "step": 6223
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0478603102521204e-05,
+      "loss": 1.146,
+      "step": 6224
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.044937852106167e-05,
+      "loss": 1.1244,
+      "step": 6225
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.042015794335081e-05,
+      "loss": 1.101,
+      "step": 6226
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.039094137234667e-05,
+      "loss": 1.1203,
+      "step": 6227
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.036172881100683e-05,
+      "loss": 1.0638,
+      "step": 6228
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0332520262288536e-05,
+      "loss": 1.1353,
+      "step": 6229
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0303315729148604e-05,
+      "loss": 1.0735,
+      "step": 6230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0274115214543424e-05,
+      "loss": 1.0857,
+      "step": 6231
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.024491872142902e-05,
+      "loss": 1.1085,
+      "step": 6232
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0215726252760904e-05,
+      "loss": 1.1008,
+      "step": 6233
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.018653781149436e-05,
+      "loss": 1.1473,
+      "step": 6234
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0157353400584126e-05,
+      "loss": 1.1257,
+      "step": 6235
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.012817302298455e-05,
+      "loss": 1.0909,
+      "step": 6236
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.00989966816496e-05,
+      "loss": 1.0982,
+      "step": 6237
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0069824379532835e-05,
+      "loss": 1.0995,
+      "step": 6238
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.00406561195874e-05,
+      "loss": 1.1376,
+      "step": 6239
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.0011491904766e-05,
+      "loss": 1.1076,
+      "step": 6240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.998233173802098e-05,
+      "loss": 1.1018,
+      "step": 6241
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9953175622304236e-05,
+      "loss": 1.1467,
+      "step": 6242
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9924023560567286e-05,
+      "loss": 1.0883,
+      "step": 6243
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9894875555761234e-05,
+      "loss": 1.1217,
+      "step": 6244
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.986573161083673e-05,
+      "loss": 1.1144,
+      "step": 6245
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.983659172874405e-05,
+      "loss": 1.1077,
+      "step": 6246
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.980745591243304e-05,
+      "loss": 1.1592,
+      "step": 6247
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.977832416485321e-05,
+      "loss": 1.1437,
+      "step": 6248
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.974919648895354e-05,
+      "loss": 1.0993,
+      "step": 6249
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.972007288768266e-05,
+      "loss": 1.1524,
+      "step": 6250
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9690953363988814e-05,
+      "loss": 1.1161,
+      "step": 6251
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.966183792081978e-05,
+      "loss": 1.1045,
+      "step": 6252
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.963272656112292e-05,
+      "loss": 1.1241,
+      "step": 6253
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.960361928784523e-05,
+      "loss": 1.1005,
+      "step": 6254
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.957451610393327e-05,
+      "loss": 1.0658,
+      "step": 6255
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.954541701233316e-05,
+      "loss": 1.1861,
+      "step": 6256
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.951632201599068e-05,
+      "loss": 1.1306,
+      "step": 6257
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.948723111785108e-05,
+      "loss": 1.2004,
+      "step": 6258
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.945814432085932e-05,
+      "loss": 1.0685,
+      "step": 6259
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.942906162795984e-05,
+      "loss": 1.1348,
+      "step": 6260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9399983042096705e-05,
+      "loss": 1.0887,
+      "step": 6261
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.937090856621362e-05,
+      "loss": 1.073,
+      "step": 6262
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9341838203253794e-05,
+      "loss": 1.05,
+      "step": 6263
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.931277195616007e-05,
+      "loss": 1.0818,
+      "step": 6264
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9283709827874825e-05,
+      "loss": 1.1179,
+      "step": 6265
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9254651821340046e-05,
+      "loss": 1.1225,
+      "step": 6266
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.922559793949731e-05,
+      "loss": 1.1617,
+      "step": 6267
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.919654818528777e-05,
+      "loss": 1.1005,
+      "step": 6268
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.91675025616522e-05,
+      "loss": 1.0915,
+      "step": 6269
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.913846107153085e-05,
+      "loss": 1.1189,
+      "step": 6270
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.9109423717863665e-05,
+      "loss": 1.1192,
+      "step": 6271
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.9080390503590114e-05,
+      "loss": 1.0217,
+      "step": 6272
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.905136143164923e-05,
+      "loss": 1.0509,
+      "step": 6273
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.902233650497972e-05,
+      "loss": 1.1578,
+      "step": 6274
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8993315726519706e-05,
+      "loss": 1.1101,
+      "step": 6275
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8964299099207085e-05,
+      "loss": 1.0977,
+      "step": 6276
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.893528662597922e-05,
+      "loss": 1.1752,
+      "step": 6277
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8906278309773056e-05,
+      "loss": 1.1111,
+      "step": 6278
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.887727415352513e-05,
+      "loss": 1.1066,
+      "step": 6279
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.884827416017156e-05,
+      "loss": 1.0858,
+      "step": 6280
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.881927833264804e-05,
+      "loss": 1.0537,
+      "step": 6281
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8790286673889906e-05,
+      "loss": 1.0724,
+      "step": 6282
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.876129918683192e-05,
+      "loss": 1.1754,
+      "step": 6283
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.873231587440856e-05,
+      "loss": 1.088,
+      "step": 6284
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.870333673955384e-05,
+      "loss": 1.1553,
+      "step": 6285
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.867436178520136e-05,
+      "loss": 1.1999,
+      "step": 6286
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8645391014284236e-05,
+      "loss": 1.0991,
+      "step": 6287
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.861642442973521e-05,
+      "loss": 1.0785,
+      "step": 6288
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.858746203448666e-05,
+      "loss": 1.1398,
+      "step": 6289
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.855850383147044e-05,
+      "loss": 1.1061,
+      "step": 6290
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8529549823618e-05,
+      "loss": 1.1411,
+      "step": 6291
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.850060001386042e-05,
+      "loss": 1.107,
+      "step": 6292
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8471654405128295e-05,
+      "loss": 1.1274,
+      "step": 6293
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.844271300035185e-05,
+      "loss": 1.1015,
+      "step": 6294
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.841377580246079e-05,
+      "loss": 1.0982,
+      "step": 6295
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.838484281438449e-05,
+      "loss": 1.0909,
+      "step": 6296
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.83559140390519e-05,
+      "loss": 1.1582,
+      "step": 6297
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.832698947939141e-05,
+      "loss": 1.0942,
+      "step": 6298
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.829806913833121e-05,
+      "loss": 1.1278,
+      "step": 6299
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.8269153018798825e-05,
+      "loss": 1.0893,
+      "step": 6300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.824024112372155e-05,
+      "loss": 1.1641,
+      "step": 6301
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.821133345602607e-05,
+      "loss": 1.0408,
+      "step": 6302
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.8182430018638814e-05,
+      "loss": 1.0667,
+      "step": 6303
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.815353081448571e-05,
+      "loss": 1.0735,
+      "step": 6304
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.812463584649223e-05,
+      "loss": 1.1071,
+      "step": 6305
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.8095745117583376e-05,
+      "loss": 1.1218,
+      "step": 6306
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.806685863068391e-05,
+      "loss": 1.143,
+      "step": 6307
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.803797638871793e-05,
+      "loss": 1.1069,
+      "step": 6308
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.800909839460929e-05,
+      "loss": 1.1497,
+      "step": 6309
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.79802246512813e-05,
+      "loss": 1.0869,
+      "step": 6310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.795135516165686e-05,
+      "loss": 1.0997,
+      "step": 6311
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.792248992865852e-05,
+      "loss": 1.2039,
+      "step": 6312
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7893628955208244e-05,
+      "loss": 1.0884,
+      "step": 6313
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.786477224422776e-05,
+      "loss": 1.133,
+      "step": 6314
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7835919798638206e-05,
+      "loss": 1.0987,
+      "step": 6315
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.780707162136032e-05,
+      "loss": 1.0761,
+      "step": 6316
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.777822771531446e-05,
+      "loss": 1.1848,
+      "step": 6317
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.774938808342056e-05,
+      "loss": 1.1071,
+      "step": 6318
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.772055272859801e-05,
+      "loss": 1.162,
+      "step": 6319
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.769172165376592e-05,
+      "loss": 1.0682,
+      "step": 6320
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.766289486184284e-05,
+      "loss": 1.0287,
+      "step": 6321
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.763407235574693e-05,
+      "loss": 1.0881,
+      "step": 6322
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.760525413839595e-05,
+      "loss": 1.1024,
+      "step": 6323
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.757644021270716e-05,
+      "loss": 1.1365,
+      "step": 6324
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.754763058159747e-05,
+      "loss": 1.1068,
+      "step": 6325
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7518825247983285e-05,
+      "loss": 1.0925,
+      "step": 6326
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.749002421478056e-05,
+      "loss": 1.0865,
+      "step": 6327
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7461227484904914e-05,
+      "loss": 1.0827,
+      "step": 6328
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.743243506127141e-05,
+      "loss": 1.1433,
+      "step": 6329
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7403646946794766e-05,
+      "loss": 1.0999,
+      "step": 6330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.737486314438926e-05,
+      "loss": 1.1167,
+      "step": 6331
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.734608365696863e-05,
+      "loss": 1.1538,
+      "step": 6332
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7317308487446356e-05,
+      "loss": 1.073,
+      "step": 6333
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7288537638735294e-05,
+      "loss": 1.0592,
+      "step": 6334
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.725977111374795e-05,
+      "loss": 1.0969,
+      "step": 6335
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7231008915396444e-05,
+      "loss": 1.0852,
+      "step": 6336
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.720225104659231e-05,
+      "loss": 1.1079,
+      "step": 6337
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7173497510246856e-05,
+      "loss": 1.0835,
+      "step": 6338
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.714474830927074e-05,
+      "loss": 1.0949,
+      "step": 6339
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7116003446574284e-05,
+      "loss": 1.166,
+      "step": 6340
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.708726292506738e-05,
+      "loss": 1.1779,
+      "step": 6341
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.7058526747659435e-05,
+      "loss": 1.0755,
+      "step": 6342
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.702979491725946e-05,
+      "loss": 1.1539,
+      "step": 6343
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.700106743677604e-05,
+      "loss": 1.0926,
+      "step": 6344
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.69723443091172e-05,
+      "loss": 1.0958,
+      "step": 6345
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.694362553719072e-05,
+      "loss": 1.1036,
+      "step": 6346
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.691491112390376e-05,
+      "loss": 1.0945,
+      "step": 6347
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.688620107216308e-05,
+      "loss": 1.2125,
+      "step": 6348
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.685749538487512e-05,
+      "loss": 1.0783,
+      "step": 6349
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.682879406494568e-05,
+      "loss": 1.0876,
+      "step": 6350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6800097115280314e-05,
+      "loss": 1.0943,
+      "step": 6351
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6771404538784e-05,
+      "loss": 1.1422,
+      "step": 6352
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.67427163383613e-05,
+      "loss": 1.0947,
+      "step": 6353
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6714032516916405e-05,
+      "loss": 1.1076,
+      "step": 6354
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.668535307735292e-05,
+      "loss": 1.1006,
+      "step": 6355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.665667802257418e-05,
+      "loss": 1.1186,
+      "step": 6356
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.662800735548293e-05,
+      "loss": 1.0997,
+      "step": 6357
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.659934107898155e-05,
+      "loss": 1.086,
+      "step": 6358
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.657067919597199e-05,
+      "loss": 1.1496,
+      "step": 6359
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6542021709355696e-05,
+      "loss": 1.0896,
+      "step": 6360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.651336862203365e-05,
+      "loss": 1.0872,
+      "step": 6361
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6484719936906494e-05,
+      "loss": 1.0762,
+      "step": 6362
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6456075656874364e-05,
+      "loss": 1.1279,
+      "step": 6363
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6427435784836867e-05,
+      "loss": 1.1281,
+      "step": 6364
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.639880032369335e-05,
+      "loss": 1.1164,
+      "step": 6365
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.637016927634251e-05,
+      "loss": 1.0973,
+      "step": 6366
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.634154264568281e-05,
+      "loss": 1.0998,
+      "step": 6367
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.631292043461207e-05,
+      "loss": 1.1139,
+      "step": 6368
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.628430264602773e-05,
+      "loss": 1.0928,
+      "step": 6369
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6255689282826884e-05,
+      "loss": 1.1349,
+      "step": 6370
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.622708034790599e-05,
+      "loss": 1.0906,
+      "step": 6371
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.619847584416121e-05,
+      "loss": 1.0858,
+      "step": 6372
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6169875774488245e-05,
+      "loss": 1.0691,
+      "step": 6373
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.614128014178224e-05,
+      "loss": 1.1496,
+      "step": 6374
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.611268894893801e-05,
+      "loss": 1.1327,
+      "step": 6375
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6084102198849854e-05,
+      "loss": 1.086,
+      "step": 6376
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.605551989441158e-05,
+      "loss": 1.0816,
+      "step": 6377
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.602694203851671e-05,
+      "loss": 1.1287,
+      "step": 6378
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.599836863405811e-05,
+      "loss": 1.1406,
+      "step": 6379
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.596979968392837e-05,
+      "loss": 1.1865,
+      "step": 6380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5941235191019524e-05,
+      "loss": 1.1141,
+      "step": 6381
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5912675158223135e-05,
+      "loss": 1.1012,
+      "step": 6382
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.588411958843045e-05,
+      "loss": 1.0496,
+      "step": 6383
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.585556848453209e-05,
+      "loss": 1.0895,
+      "step": 6384
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.582702184941837e-05,
+      "loss": 1.1296,
+      "step": 6385
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.579847968597911e-05,
+      "loss": 1.0785,
+      "step": 6386
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5769941997103606e-05,
+      "loss": 1.1522,
+      "step": 6387
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5741408785680816e-05,
+      "loss": 1.0907,
+      "step": 6388
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.571288005459916e-05,
+      "loss": 1.1369,
+      "step": 6389
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.568435580674657e-05,
+      "loss": 1.0604,
+      "step": 6390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5655836045010675e-05,
+      "loss": 1.1002,
+      "step": 6391
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.562732077227849e-05,
+      "loss": 1.1564,
+      "step": 6392
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5598809991436715e-05,
+      "loss": 1.0863,
+      "step": 6393
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.5570303705371485e-05,
+      "loss": 1.1321,
+      "step": 6394
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.5541801916968473e-05,
+      "loss": 1.075,
+      "step": 6395
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.551330462911304e-05,
+      "loss": 1.0943,
+      "step": 6396
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.548481184468989e-05,
+      "loss": 1.1536,
+      "step": 6397
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.545632356658344e-05,
+      "loss": 1.0598,
+      "step": 6398
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.54278397976776e-05,
+      "loss": 1.0378,
+      "step": 6399
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.539936054085577e-05,
+      "loss": 1.0874,
+      "step": 6400
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.537088579900098e-05,
+      "loss": 1.077,
+      "step": 6401
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.5342415574995734e-05,
+      "loss": 1.0543,
+      "step": 6402
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.531394987172204e-05,
+      "loss": 1.1329,
+      "step": 6403
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.528548869206161e-05,
+      "loss": 1.0586,
+      "step": 6404
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.5257032038895506e-05,
+      "loss": 1.0663,
+      "step": 6405
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.522857991510452e-05,
+      "loss": 1.053,
+      "step": 6406
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.520013232356882e-05,
+      "loss": 1.0884,
+      "step": 6407
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.517168926716818e-05,
+      "loss": 1.1217,
+      "step": 6408
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.514325074878197e-05,
+      "loss": 1.1171,
+      "step": 6409
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.511481677128901e-05,
+      "loss": 1.1089,
+      "step": 6410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.508638733756768e-05,
+      "loss": 1.1044,
+      "step": 6411
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.5057962450495947e-05,
+      "loss": 1.1534,
+      "step": 6412
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.502954211295133e-05,
+      "loss": 1.1336,
+      "step": 6413
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.500112632781076e-05,
+      "loss": 1.0649,
+      "step": 6414
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.49727150979509e-05,
+      "loss": 1.1618,
+      "step": 6415
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4944308426247745e-05,
+      "loss": 1.0688,
+      "step": 6416
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.491590631557703e-05,
+      "loss": 1.0947,
+      "step": 6417
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4887508768813854e-05,
+      "loss": 1.0849,
+      "step": 6418
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.485911578883294e-05,
+      "loss": 1.1598,
+      "step": 6419
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.483072737850856e-05,
+      "loss": 1.1614,
+      "step": 6420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.480234354071446e-05,
+      "loss": 1.1374,
+      "step": 6421
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4773964278324045e-05,
+      "loss": 1.1152,
+      "step": 6422
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.474558959421011e-05,
+      "loss": 1.1067,
+      "step": 6423
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.471721949124504e-05,
+      "loss": 1.1674,
+      "step": 6424
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.468885397230082e-05,
+      "loss": 1.1176,
+      "step": 6425
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4660493040248875e-05,
+      "loss": 1.0589,
+      "step": 6426
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.463213669796021e-05,
+      "loss": 1.1094,
+      "step": 6427
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.460378494830545e-05,
+      "loss": 1.1702,
+      "step": 6428
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.457543779415455e-05,
+      "loss": 1.0414,
+      "step": 6429
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.454709523837722e-05,
+      "loss": 1.1024,
+      "step": 6430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.451875728384257e-05,
+      "loss": 1.0954,
+      "step": 6431
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4490423933419234e-05,
+      "loss": 1.1512,
+      "step": 6432
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.446209518997549e-05,
+      "loss": 1.1555,
+      "step": 6433
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4433771056379036e-05,
+      "loss": 1.0595,
+      "step": 6434
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4405451535497196e-05,
+      "loss": 1.1879,
+      "step": 6435
+    },
+    {
+      "epoch": 2.58,
+      "eval_loss": 0.8183497190475464,
+      "eval_runtime": 504.4277,
+      "eval_samples_per_second": 51.345,
+      "eval_steps_per_second": 25.673,
+      "step": 6435
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4377136630196765e-05,
+      "loss": 1.114,
+      "step": 6436
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.434882634334406e-05,
+      "loss": 1.1637,
+      "step": 6437
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.432052067780502e-05,
+      "loss": 1.1221,
+      "step": 6438
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.429221963644497e-05,
+      "loss": 1.0562,
+      "step": 6439
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.426392322212891e-05,
+      "loss": 1.1209,
+      "step": 6440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.423563143772134e-05,
+      "loss": 1.1429,
+      "step": 6441
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4207344286086204e-05,
+      "loss": 1.0832,
+      "step": 6442
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.417906177008712e-05,
+      "loss": 1.0784,
+      "step": 6443
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.4150783892587074e-05,
+      "loss": 1.082,
+      "step": 6444
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.412251065644867e-05,
+      "loss": 1.1363,
+      "step": 6445
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.409424206453408e-05,
+      "loss": 1.1301,
+      "step": 6446
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.4065978119704905e-05,
+      "loss": 1.0676,
+      "step": 6447
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.4037718824822424e-05,
+      "loss": 1.1037,
+      "step": 6448
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.4009464182747284e-05,
+      "loss": 1.0547,
+      "step": 6449
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3981214196339705e-05,
+      "loss": 1.1015,
+      "step": 6450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.395296886845953e-05,
+      "loss": 1.1206,
+      "step": 6451
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.392472820196603e-05,
+      "loss": 1.0893,
+      "step": 6452
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3896492199717975e-05,
+      "loss": 1.0992,
+      "step": 6453
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.386826086457385e-05,
+      "loss": 1.1553,
+      "step": 6454
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3840034199391474e-05,
+      "loss": 1.1109,
+      "step": 6455
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.381181220702822e-05,
+      "loss": 1.0913,
+      "step": 6456
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.378359489034113e-05,
+      "loss": 1.1149,
+      "step": 6457
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.375538225218655e-05,
+      "loss": 1.0894,
+      "step": 6458
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.372717429542058e-05,
+      "loss": 1.1782,
+      "step": 6459
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.369897102289869e-05,
+      "loss": 1.1075,
+      "step": 6460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.36707724374759e-05,
+      "loss": 1.1305,
+      "step": 6461
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.364257854200685e-05,
+      "loss": 1.0727,
+      "step": 6462
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.361438933934555e-05,
+      "loss": 1.1928,
+      "step": 6463
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.358620483234571e-05,
+      "loss": 1.0592,
+      "step": 6464
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.355802502386042e-05,
+      "loss": 1.1386,
+      "step": 6465
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.352984991674235e-05,
+      "loss": 1.1241,
+      "step": 6466
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3501679513843687e-05,
+      "loss": 1.155,
+      "step": 6467
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3473513818016205e-05,
+      "loss": 1.1406,
+      "step": 6468
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.344535283211108e-05,
+      "loss": 1.0357,
+      "step": 6469
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.341719655897914e-05,
+      "loss": 1.0965,
+      "step": 6470
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.338904500147061e-05,
+      "loss": 1.0453,
+      "step": 6471
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3360898162435346e-05,
+      "loss": 1.1525,
+      "step": 6472
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3332756044722674e-05,
+      "loss": 1.1701,
+      "step": 6473
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.330461865118141e-05,
+      "loss": 1.1371,
+      "step": 6474
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3276485984659975e-05,
+      "loss": 1.1264,
+      "step": 6475
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.324835804800623e-05,
+      "loss": 1.0931,
+      "step": 6476
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.322023484406764e-05,
+      "loss": 1.1511,
+      "step": 6477
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3192116375691115e-05,
+      "loss": 1.1249,
+      "step": 6478
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3164002645723085e-05,
+      "loss": 1.1009,
+      "step": 6479
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3135893657009604e-05,
+      "loss": 1.0877,
+      "step": 6480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.310778941239609e-05,
+      "loss": 1.1794,
+      "step": 6481
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.307968991472761e-05,
+      "loss": 1.1158,
+      "step": 6482
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.3051595166848745e-05,
+      "loss": 1.1043,
+      "step": 6483
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.302350517160346e-05,
+      "loss": 1.0779,
+      "step": 6484
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2995419931835436e-05,
+      "loss": 1.1373,
+      "step": 6485
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2967339450387696e-05,
+      "loss": 1.1145,
+      "step": 6486
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.293926373010285e-05,
+      "loss": 1.1058,
+      "step": 6487
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2911192773823104e-05,
+      "loss": 1.1107,
+      "step": 6488
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2883126584390005e-05,
+      "loss": 1.0572,
+      "step": 6489
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.285506516464483e-05,
+      "loss": 1.1391,
+      "step": 6490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.282700851742821e-05,
+      "loss": 1.1042,
+      "step": 6491
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.27989566455803e-05,
+      "loss": 1.0906,
+      "step": 6492
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.277090955194092e-05,
+      "loss": 1.0908,
+      "step": 6493
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2742867239349206e-05,
+      "loss": 1.0815,
+      "step": 6494
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2714829710643965e-05,
+      "loss": 1.0837,
+      "step": 6495
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.268679696866348e-05,
+      "loss": 1.0767,
+      "step": 6496
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.265876901624552e-05,
+      "loss": 1.1107,
+      "step": 6497
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2630745856227335e-05,
+      "loss": 1.1449,
+      "step": 6498
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.26027274914458e-05,
+      "loss": 1.0997,
+      "step": 6499
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.257471392473719e-05,
+      "loss": 1.0862,
+      "step": 6500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.25467051589374e-05,
+      "loss": 1.1283,
+      "step": 6501
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.251870119688176e-05,
+      "loss": 1.0376,
+      "step": 6502
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.249070204140509e-05,
+      "loss": 1.1038,
+      "step": 6503
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.246270769534187e-05,
+      "loss": 1.0381,
+      "step": 6504
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.24347181615259e-05,
+      "loss": 1.1091,
+      "step": 6505
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.240673344279068e-05,
+      "loss": 1.0876,
+      "step": 6506
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.237875354196906e-05,
+      "loss": 1.0773,
+      "step": 6507
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.235077846189342e-05,
+      "loss": 1.1037,
+      "step": 6508
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2322808205395876e-05,
+      "loss": 1.0604,
+      "step": 6509
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2294842775307784e-05,
+      "loss": 1.1542,
+      "step": 6510
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.226688217446007e-05,
+      "loss": 1.1127,
+      "step": 6511
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.223892640568331e-05,
+      "loss": 1.1413,
+      "step": 6512
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.22109754718074e-05,
+      "loss": 1.137,
+      "step": 6513
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.218302937566193e-05,
+      "loss": 1.1035,
+      "step": 6514
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.215508812007587e-05,
+      "loss": 1.1326,
+      "step": 6515
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2127151707877696e-05,
+      "loss": 1.0876,
+      "step": 6516
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.209922014189552e-05,
+      "loss": 1.0849,
+      "step": 6517
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.20712934249568e-05,
+      "loss": 1.0859,
+      "step": 6518
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.204337155988867e-05,
+      "loss": 1.1021,
+      "step": 6519
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.201545454951764e-05,
+      "loss": 1.1392,
+      "step": 6520
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1987542396669756e-05,
+      "loss": 1.061,
+      "step": 6521
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.195963510417061e-05,
+      "loss": 1.1206,
+      "step": 6522
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.193173267484535e-05,
+      "loss": 1.1291,
+      "step": 6523
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.190383511151846e-05,
+      "loss": 1.0702,
+      "step": 6524
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1875942417014124e-05,
+      "loss": 1.1386,
+      "step": 6525
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.184805459415588e-05,
+      "loss": 1.1204,
+      "step": 6526
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.182017164576691e-05,
+      "loss": 1.1378,
+      "step": 6527
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1792293574669804e-05,
+      "loss": 1.1268,
+      "step": 6528
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.176442038368663e-05,
+      "loss": 1.1305,
+      "step": 6529
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.173655207563911e-05,
+      "loss": 1.0675,
+      "step": 6530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.17086886533483e-05,
+      "loss": 1.1355,
+      "step": 6531
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.168083011963492e-05,
+      "loss": 1.0997,
+      "step": 6532
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.165297647731907e-05,
+      "loss": 1.0213,
+      "step": 6533
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.162512772922037e-05,
+      "loss": 1.1072,
+      "step": 6534
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1597283878158055e-05,
+      "loss": 1.0809,
+      "step": 6535
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1569444926950704e-05,
+      "loss": 1.1529,
+      "step": 6536
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1541610878416536e-05,
+      "loss": 1.0968,
+      "step": 6537
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.151378173537323e-05,
+      "loss": 1.0782,
+      "step": 6538
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1485957500637906e-05,
+      "loss": 1.1121,
+      "step": 6539
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1458138177027296e-05,
+      "loss": 1.0932,
+      "step": 6540
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1430323767357556e-05,
+      "loss": 1.0982,
+      "step": 6541
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.1402514274444326e-05,
+      "loss": 1.1191,
+      "step": 6542
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.137470970110285e-05,
+      "loss": 1.1214,
+      "step": 6543
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.134691005014779e-05,
+      "loss": 1.0543,
+      "step": 6544
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.1319115324393286e-05,
+      "loss": 1.0329,
+      "step": 6545
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.12913255266531e-05,
+      "loss": 1.1156,
+      "step": 6546
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.126354065974035e-05,
+      "loss": 1.0698,
+      "step": 6547
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.123576072646781e-05,
+      "loss": 1.1481,
+      "step": 6548
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.1207985729647625e-05,
+      "loss": 1.112,
+      "step": 6549
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.118021567209139e-05,
+      "loss": 1.0653,
+      "step": 6550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.1152450556610474e-05,
+      "loss": 1.0671,
+      "step": 6551
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.112469038601548e-05,
+      "loss": 1.0975,
+      "step": 6552
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.1096935163116575e-05,
+      "loss": 1.1103,
+      "step": 6553
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.10691848907235e-05,
+      "loss": 1.1297,
+      "step": 6554
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.1041439571645374e-05,
+      "loss": 1.0439,
+      "step": 6555
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.101369920869097e-05,
+      "loss": 1.1479,
+      "step": 6556
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0985963804668427e-05,
+      "loss": 1.1071,
+      "step": 6557
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.095823336238539e-05,
+      "loss": 1.0641,
+      "step": 6558
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.093050788464911e-05,
+      "loss": 1.1638,
+      "step": 6559
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.090278737426619e-05,
+      "loss": 1.155,
+      "step": 6560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0875071834042875e-05,
+      "loss": 1.1898,
+      "step": 6561
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.084736126678482e-05,
+      "loss": 1.1249,
+      "step": 6562
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.081965567529712e-05,
+      "loss": 1.205,
+      "step": 6563
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0791955062384554e-05,
+      "loss": 1.1609,
+      "step": 6564
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0764259430851236e-05,
+      "loss": 1.0932,
+      "step": 6565
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.073656878350077e-05,
+      "loss": 1.1327,
+      "step": 6566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.070888312313641e-05,
+      "loss": 1.0266,
+      "step": 6567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0681202452560695e-05,
+      "loss": 1.1701,
+      "step": 6568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0653526774575855e-05,
+      "loss": 1.0878,
+      "step": 6569
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0625856091983494e-05,
+      "loss": 1.0808,
+      "step": 6570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.059819040758469e-05,
+      "loss": 1.1235,
+      "step": 6571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.057052972418017e-05,
+      "loss": 1.1033,
+      "step": 6572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.054287404456994e-05,
+      "loss": 1.1297,
+      "step": 6573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.051522337155372e-05,
+      "loss": 1.1023,
+      "step": 6574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0487577707930565e-05,
+      "loss": 1.077,
+      "step": 6575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.045993705649903e-05,
+      "loss": 1.1513,
+      "step": 6576
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0432301420057256e-05,
+      "loss": 1.1639,
+      "step": 6577
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0404670801402855e-05,
+      "loss": 1.1065,
+      "step": 6578
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.037704520333282e-05,
+      "loss": 1.0885,
+      "step": 6579
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0349424628643806e-05,
+      "loss": 1.1254,
+      "step": 6580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.032180908013178e-05,
+      "loss": 1.1502,
+      "step": 6581
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.029419856059239e-05,
+      "loss": 1.0937,
+      "step": 6582
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.026659307282062e-05,
+      "loss": 1.1079,
+      "step": 6583
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0238992619610967e-05,
+      "loss": 1.0808,
+      "step": 6584
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.021139720375754e-05,
+      "loss": 1.1113,
+      "step": 6585
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0183806828053794e-05,
+      "loss": 1.076,
+      "step": 6586
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.015622149529271e-05,
+      "loss": 1.1154,
+      "step": 6587
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.012864120826683e-05,
+      "loss": 1.0919,
+      "step": 6588
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.010106596976808e-05,
+      "loss": 1.1156,
+      "step": 6589
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.0073495782588e-05,
+      "loss": 1.0731,
+      "step": 6590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.004593064951746e-05,
+      "loss": 1.0791,
+      "step": 6591
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.001837057334696e-05,
+      "loss": 1.1074,
+      "step": 6592
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9990815556866467e-05,
+      "loss": 1.0909,
+      "step": 6593
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.996326560286536e-05,
+      "loss": 1.016,
+      "step": 6594
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9935720714132514e-05,
+      "loss": 1.126,
+      "step": 6595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.99081808934564e-05,
+      "loss": 1.0539,
+      "step": 6596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.988064614362482e-05,
+      "loss": 1.1651,
+      "step": 6597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9853116467425234e-05,
+      "loss": 1.1294,
+      "step": 6598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.982559186764445e-05,
+      "loss": 1.0813,
+      "step": 6599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.979807234706877e-05,
+      "loss": 1.0679,
+      "step": 6600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.977055790848411e-05,
+      "loss": 1.0987,
+      "step": 6601
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9743048554675696e-05,
+      "loss": 1.1216,
+      "step": 6602
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.971554428842841e-05,
+      "loss": 1.082,
+      "step": 6603
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9688045112526496e-05,
+      "loss": 1.0892,
+      "step": 6604
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.966055102975364e-05,
+      "loss": 1.1309,
+      "step": 6605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.963306204289326e-05,
+      "loss": 1.1562,
+      "step": 6606
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9605578154728015e-05,
+      "loss": 1.0838,
+      "step": 6607
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.957809936804009e-05,
+      "loss": 1.1624,
+      "step": 6608
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9550625685611244e-05,
+      "loss": 1.1423,
+      "step": 6609
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.952315711022262e-05,
+      "loss": 1.0995,
+      "step": 6610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.949569364465494e-05,
+      "loss": 1.1005,
+      "step": 6611
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.946823529168835e-05,
+      "loss": 1.0341,
+      "step": 6612
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.944078205410241e-05,
+      "loss": 1.0621,
+      "step": 6613
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.941333393467634e-05,
+      "loss": 1.1469,
+      "step": 6614
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9385890936188664e-05,
+      "loss": 1.0662,
+      "step": 6615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.935845306141753e-05,
+      "loss": 1.0532,
+      "step": 6616
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9331020313140465e-05,
+      "loss": 1.1442,
+      "step": 6617
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9303592694134445e-05,
+      "loss": 1.0959,
+      "step": 6618
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.927617020717614e-05,
+      "loss": 1.0815,
+      "step": 6619
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.924875285504148e-05,
+      "loss": 1.0935,
+      "step": 6620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.922134064050592e-05,
+      "loss": 1.0964,
+      "step": 6621
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9193933566344484e-05,
+      "loss": 1.1044,
+      "step": 6622
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.916653163533157e-05,
+      "loss": 1.1426,
+      "step": 6623
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9139134850241154e-05,
+      "loss": 1.1296,
+      "step": 6624
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9111743213846604e-05,
+      "loss": 1.1197,
+      "step": 6625
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9084356728920776e-05,
+      "loss": 1.155,
+      "step": 6626
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.9056975398236104e-05,
+      "loss": 1.1226,
+      "step": 6627
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.902959922456435e-05,
+      "loss": 1.1114,
+      "step": 6628
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.90022282106769e-05,
+      "loss": 1.0974,
+      "step": 6629
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.897486235934452e-05,
+      "loss": 1.1623,
+      "step": 6630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8947501673337445e-05,
+      "loss": 1.1008,
+      "step": 6631
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.892014615542545e-05,
+      "loss": 1.0948,
+      "step": 6632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.889279580837781e-05,
+      "loss": 1.1106,
+      "step": 6633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.886545063496316e-05,
+      "loss": 1.1989,
+      "step": 6634
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.883811063794973e-05,
+      "loss": 1.1081,
+      "step": 6635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.881077582010516e-05,
+      "loss": 1.0739,
+      "step": 6636
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8783446184196534e-05,
+      "loss": 1.0569,
+      "step": 6637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8756121732990535e-05,
+      "loss": 1.0951,
+      "step": 6638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.872880246925317e-05,
+      "loss": 1.0178,
+      "step": 6639
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8701488395750064e-05,
+      "loss": 1.0826,
+      "step": 6640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.867417951524622e-05,
+      "loss": 1.0227,
+      "step": 6641
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.86468758305061e-05,
+      "loss": 1.1462,
+      "step": 6642
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8619577344293755e-05,
+      "loss": 1.1051,
+      "step": 6643
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.859228405937256e-05,
+      "loss": 1.1258,
+      "step": 6644
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.856499597850553e-05,
+      "loss": 1.0962,
+      "step": 6645
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.853771310445498e-05,
+      "loss": 1.1287,
+      "step": 6646
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.851043543998284e-05,
+      "loss": 1.1228,
+      "step": 6647
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.848316298785046e-05,
+      "loss": 1.1418,
+      "step": 6648
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8455895750818635e-05,
+      "loss": 1.0862,
+      "step": 6649
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.842863373164761e-05,
+      "loss": 1.116,
+      "step": 6650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.840137693309724e-05,
+      "loss": 1.0806,
+      "step": 6651
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.837412535792666e-05,
+      "loss": 1.0774,
+      "step": 6652
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.834687900889467e-05,
+      "loss": 1.1547,
+      "step": 6653
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8319637888759395e-05,
+      "loss": 1.0927,
+      "step": 6654
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8292402000278434e-05,
+      "loss": 1.0749,
+      "step": 6655
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8265171346209004e-05,
+      "loss": 1.0843,
+      "step": 6656
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.823794592930759e-05,
+      "loss": 1.0859,
+      "step": 6657
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8210725752330345e-05,
+      "loss": 1.0876,
+      "step": 6658
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.818351081803275e-05,
+      "loss": 1.1247,
+      "step": 6659
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.8156301129169713e-05,
+      "loss": 1.1069,
+      "step": 6660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.812909668849586e-05,
+      "loss": 1.1706,
+      "step": 6661
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.810189749876504e-05,
+      "loss": 1.1271,
+      "step": 6662
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.807470356273062e-05,
+      "loss": 1.1041,
+      "step": 6663
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.804751488314555e-05,
+      "loss": 1.1063,
+      "step": 6664
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.802033146276208e-05,
+      "loss": 1.2019,
+      "step": 6665
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7993153304332105e-05,
+      "loss": 1.0925,
+      "step": 6666
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.796598041060685e-05,
+      "loss": 1.1211,
+      "step": 6667
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7938812784337014e-05,
+      "loss": 1.1428,
+      "step": 6668
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.791165042827289e-05,
+      "loss": 1.1128,
+      "step": 6669
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.788449334516406e-05,
+      "loss": 1.1169,
+      "step": 6670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.785734153775975e-05,
+      "loss": 1.1198,
+      "step": 6671
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.783019500880852e-05,
+      "loss": 1.0993,
+      "step": 6672
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.780305376105836e-05,
+      "loss": 1.1384,
+      "step": 6673
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7775917797256966e-05,
+      "loss": 1.08,
+      "step": 6674
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.774878712015126e-05,
+      "loss": 1.121,
+      "step": 6675
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7721661732487666e-05,
+      "loss": 1.1381,
+      "step": 6676
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.76945416370122e-05,
+      "loss": 1.1471,
+      "step": 6677
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.766742683647021e-05,
+      "loss": 1.1435,
+      "step": 6678
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.764031733360651e-05,
+      "loss": 1.1174,
+      "step": 6679
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7613213131165514e-05,
+      "loss": 1.0943,
+      "step": 6680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.758611423189092e-05,
+      "loss": 1.0975,
+      "step": 6681
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7559020638526054e-05,
+      "loss": 1.0777,
+      "step": 6682
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.753193235381358e-05,
+      "loss": 1.1204,
+      "step": 6683
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.750484938049565e-05,
+      "loss": 1.1155,
+      "step": 6684
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.747777172131398e-05,
+      "loss": 1.0733,
+      "step": 6685
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7450699379009575e-05,
+      "loss": 1.0955,
+      "step": 6686
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.742363235632304e-05,
+      "loss": 1.0934,
+      "step": 6687
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.739657065599444e-05,
+      "loss": 1.134,
+      "step": 6688
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7369514280763185e-05,
+      "loss": 1.1087,
+      "step": 6689
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.734246323336827e-05,
+      "loss": 1.105,
+      "step": 6690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.7315417516548085e-05,
+      "loss": 1.0959,
+      "step": 6691
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.728837713304045e-05,
+      "loss": 1.073,
+      "step": 6692
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.726134208558277e-05,
+      "loss": 1.0991,
+      "step": 6693
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.723431237691175e-05,
+      "loss": 1.1789,
+      "step": 6694
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.7207288009763706e-05,
+      "loss": 1.104,
+      "step": 6695
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.718026898687431e-05,
+      "loss": 1.1395,
+      "step": 6696
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.715325531097867e-05,
+      "loss": 1.0747,
+      "step": 6697
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.712624698481152e-05,
+      "loss": 1.0915,
+      "step": 6698
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.709924401110682e-05,
+      "loss": 1.1521,
+      "step": 6699
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.707224639259822e-05,
+      "loss": 1.108,
+      "step": 6700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.7045254132018625e-05,
+      "loss": 1.0807,
+      "step": 6701
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.701826723210051e-05,
+      "loss": 1.0628,
+      "step": 6702
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.699128569557585e-05,
+      "loss": 1.0999,
+      "step": 6703
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6964309525175963e-05,
+      "loss": 1.1307,
+      "step": 6704
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6937338723631644e-05,
+      "loss": 1.0639,
+      "step": 6705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.691037329367324e-05,
+      "loss": 1.1382,
+      "step": 6706
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.68834132380304e-05,
+      "loss": 1.1484,
+      "step": 6707
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.685645855943242e-05,
+      "loss": 1.0894,
+      "step": 6708
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6829509260607895e-05,
+      "loss": 1.089,
+      "step": 6709
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.68025653442849e-05,
+      "loss": 1.1151,
+      "step": 6710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6775626813191065e-05,
+      "loss": 1.0917,
+      "step": 6711
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.674869367005334e-05,
+      "loss": 1.1456,
+      "step": 6712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.672176591759824e-05,
+      "loss": 1.074,
+      "step": 6713
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.669484355855167e-05,
+      "loss": 1.1026,
+      "step": 6714
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6667926595638934e-05,
+      "loss": 1.1176,
+      "step": 6715
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.664101503158501e-05,
+      "loss": 1.1405,
+      "step": 6716
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6614108869114114e-05,
+      "loss": 1.0938,
+      "step": 6717
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6587208110949934e-05,
+      "loss": 1.0838,
+      "step": 6718
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.656031275981574e-05,
+      "loss": 1.1603,
+      "step": 6719
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.653342281843414e-05,
+      "loss": 1.0858,
+      "step": 6720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.650653828952719e-05,
+      "loss": 1.1381,
+      "step": 6721
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.647965917581652e-05,
+      "loss": 1.1494,
+      "step": 6722
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.645278548002303e-05,
+      "loss": 1.1164,
+      "step": 6723
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.642591720486728e-05,
+      "loss": 1.0751,
+      "step": 6724
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6399054353069104e-05,
+      "loss": 1.1557,
+      "step": 6725
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.637219692734783e-05,
+      "loss": 1.1312,
+      "step": 6726
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.634534493042235e-05,
+      "loss": 1.0804,
+      "step": 6727
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.631849836501083e-05,
+      "loss": 1.0518,
+      "step": 6728
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.629165723383101e-05,
+      "loss": 1.0972,
+      "step": 6729
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.626482153960009e-05,
+      "loss": 1.1106,
+      "step": 6730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6237991285034585e-05,
+      "loss": 1.0988,
+      "step": 6731
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.621116647285064e-05,
+      "loss": 1.0276,
+      "step": 6732
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.618434710576371e-05,
+      "loss": 1.087,
+      "step": 6733
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.615753318648872e-05,
+      "loss": 1.0825,
+      "step": 6734
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6130724717740125e-05,
+      "loss": 1.0889,
+      "step": 6735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.610392170223171e-05,
+      "loss": 1.1217,
+      "step": 6736
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.607712414267685e-05,
+      "loss": 1.1327,
+      "step": 6737
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6050332041788244e-05,
+      "loss": 1.1333,
+      "step": 6738
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.6023545402278045e-05,
+      "loss": 1.1551,
+      "step": 6739
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.599676422685797e-05,
+      "loss": 1.1423,
+      "step": 6740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5969988518239025e-05,
+      "loss": 1.0965,
+      "step": 6741
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5943218279131786e-05,
+      "loss": 1.0922,
+      "step": 6742
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.591645351224626e-05,
+      "loss": 1.1619,
+      "step": 6743
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.588969422029179e-05,
+      "loss": 1.1133,
+      "step": 6744
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.586294040597734e-05,
+      "loss": 1.147,
+      "step": 6745
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.583619207201119e-05,
+      "loss": 1.1292,
+      "step": 6746
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5809449221101034e-05,
+      "loss": 1.0783,
+      "step": 6747
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.578271185595418e-05,
+      "loss": 1.1541,
+      "step": 6748
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5755979979277175e-05,
+      "loss": 1.0665,
+      "step": 6749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.572925359377622e-05,
+      "loss": 1.0889,
+      "step": 6750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.57025327021568e-05,
+      "loss": 1.1116,
+      "step": 6751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.567581730712386e-05,
+      "loss": 1.0642,
+      "step": 6752
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5649107411381895e-05,
+      "loss": 1.0626,
+      "step": 6753
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.562240301763472e-05,
+      "loss": 1.0758,
+      "step": 6754
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5595704128585706e-05,
+      "loss": 1.1452,
+      "step": 6755
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.556901074693754e-05,
+      "loss": 1.0584,
+      "step": 6756
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.554232287539244e-05,
+      "loss": 1.1702,
+      "step": 6757
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.55156405166521e-05,
+      "loss": 1.1484,
+      "step": 6758
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.548896367341757e-05,
+      "loss": 1.0881,
+      "step": 6759
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5462292348389325e-05,
+      "loss": 1.1414,
+      "step": 6760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.54356265442674e-05,
+      "loss": 1.0877,
+      "step": 6761
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.540896626375113e-05,
+      "loss": 1.1084,
+      "step": 6762
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.538231150953944e-05,
+      "loss": 1.1515,
+      "step": 6763
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.535566228433058e-05,
+      "loss": 1.1234,
+      "step": 6764
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.532901859082225e-05,
+      "loss": 1.0936,
+      "step": 6765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.530238043171167e-05,
+      "loss": 1.002,
+      "step": 6766
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.527574780969541e-05,
+      "loss": 1.1125,
+      "step": 6767
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5249120727469505e-05,
+      "loss": 1.0894,
+      "step": 6768
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.52224991877295e-05,
+      "loss": 1.1197,
+      "step": 6769
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.519588319317024e-05,
+      "loss": 1.1935,
+      "step": 6770
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5169272746486136e-05,
+      "loss": 1.1189,
+      "step": 6771
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.514266785037102e-05,
+      "loss": 1.1035,
+      "step": 6772
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.511606850751806e-05,
+      "loss": 1.1047,
+      "step": 6773
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.508947472062002e-05,
+      "loss": 1.1033,
+      "step": 6774
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.506288649236895e-05,
+      "loss": 1.1343,
+      "step": 6775
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.503630382545639e-05,
+      "loss": 1.1061,
+      "step": 6776
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.50097267225734e-05,
+      "loss": 1.1537,
+      "step": 6777
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.498315518641033e-05,
+      "loss": 1.0941,
+      "step": 6778
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.495658921965711e-05,
+      "loss": 1.1,
+      "step": 6779
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.493002882500301e-05,
+      "loss": 1.1612,
+      "step": 6780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.490347400513672e-05,
+      "loss": 1.1486,
+      "step": 6781
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.487692476274651e-05,
+      "loss": 1.1122,
+      "step": 6782
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.485038110051988e-05,
+      "loss": 1.102,
+      "step": 6783
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.482384302114393e-05,
+      "loss": 1.1085,
+      "step": 6784
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.479731052730516e-05,
+      "loss": 1.1443,
+      "step": 6785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.477078362168941e-05,
+      "loss": 1.1071,
+      "step": 6786
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.47442623069821e-05,
+      "loss": 1.1366,
+      "step": 6787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.471774658586797e-05,
+      "loss": 1.1125,
+      "step": 6788
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4691236461031206e-05,
+      "loss": 1.102,
+      "step": 6789
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4664731935155515e-05,
+      "loss": 1.1317,
+      "step": 6790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.463823301092391e-05,
+      "loss": 1.1623,
+      "step": 6791
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.461173969101897e-05,
+      "loss": 1.0747,
+      "step": 6792
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.458525197812262e-05,
+      "loss": 1.0653,
+      "step": 6793
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.455876987491619e-05,
+      "loss": 1.0648,
+      "step": 6794
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4532293384080545e-05,
+      "loss": 1.0745,
+      "step": 6795
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.450582250829589e-05,
+      "loss": 1.1163,
+      "step": 6796
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.447935725024191e-05,
+      "loss": 1.1609,
+      "step": 6797
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.445289761259774e-05,
+      "loss": 1.0901,
+      "step": 6798
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4426443598041856e-05,
+      "loss": 1.112,
+      "step": 6799
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.43999952092523e-05,
+      "loss": 1.0675,
+      "step": 6800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.437355244890643e-05,
+      "loss": 1.1605,
+      "step": 6801
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.434711531968102e-05,
+      "loss": 1.1364,
+      "step": 6802
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.432068382425243e-05,
+      "loss": 1.1195,
+      "step": 6803
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4294257965296246e-05,
+      "loss": 1.1294,
+      "step": 6804
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4267837745487674e-05,
+      "loss": 1.1051,
+      "step": 6805
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.424142316750121e-05,
+      "loss": 1.082,
+      "step": 6806
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.421501423401079e-05,
+      "loss": 1.0532,
+      "step": 6807
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.418861094768991e-05,
+      "loss": 1.106,
+      "step": 6808
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.416221331121131e-05,
+      "loss": 1.1176,
+      "step": 6809
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4135821327247326e-05,
+      "loss": 1.1067,
+      "step": 6810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4109434998469565e-05,
+      "loss": 1.095,
+      "step": 6811
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.408305432754923e-05,
+      "loss": 1.0912,
+      "step": 6812
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.405667931715679e-05,
+      "loss": 1.0701,
+      "step": 6813
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.403030996996227e-05,
+      "loss": 1.0664,
+      "step": 6814
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.400394628863499e-05,
+      "loss": 1.0877,
+      "step": 6815
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.397758827584386e-05,
+      "loss": 1.1079,
+      "step": 6816
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3951235934257096e-05,
+      "loss": 1.1299,
+      "step": 6817
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3924889266542315e-05,
+      "loss": 1.0714,
+      "step": 6818
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.38985482753667e-05,
+      "loss": 1.0464,
+      "step": 6819
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.387221296339671e-05,
+      "loss": 1.1003,
+      "step": 6820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.384588333329836e-05,
+      "loss": 1.1017,
+      "step": 6821
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.381955938773697e-05,
+      "loss": 1.1105,
+      "step": 6822
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3793241129377335e-05,
+      "loss": 1.1778,
+      "step": 6823
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3766928560883736e-05,
+      "loss": 1.1336,
+      "step": 6824
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.374062168491975e-05,
+      "loss": 1.0961,
+      "step": 6825
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.371432050414848e-05,
+      "loss": 1.0576,
+      "step": 6826
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.368802502123246e-05,
+      "loss": 1.1017,
+      "step": 6827
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3661735238833535e-05,
+      "loss": 1.0952,
+      "step": 6828
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.363545115961312e-05,
+      "loss": 1.0904,
+      "step": 6829
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.360917278623193e-05,
+      "loss": 1.1494,
+      "step": 6830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.358290012135015e-05,
+      "loss": 1.1862,
+      "step": 6831
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.355663316762741e-05,
+      "loss": 1.1612,
+      "step": 6832
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3530371927722715e-05,
+      "loss": 1.1106,
+      "step": 6833
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.350411640429455e-05,
+      "loss": 1.1151,
+      "step": 6834
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3477866600000775e-05,
+      "loss": 1.0692,
+      "step": 6835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.345162251749864e-05,
+      "loss": 1.1348,
+      "step": 6836
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3425384159444935e-05,
+      "loss": 1.1203,
+      "step": 6837
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.339915152849572e-05,
+      "loss": 1.0856,
+      "step": 6838
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.337292462730659e-05,
+      "loss": 1.0861,
+      "step": 6839
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.334670345853254e-05,
+      "loss": 1.1639,
+      "step": 6840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.332048802482792e-05,
+      "loss": 1.0917,
+      "step": 6841
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.329427832884659e-05,
+      "loss": 1.1288,
+      "step": 6842
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.326807437324177e-05,
+      "loss": 1.0959,
+      "step": 6843
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.3241876160666075e-05,
+      "loss": 1.1639,
+      "step": 6844
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.3215683693771626e-05,
+      "loss": 1.1451,
+      "step": 6845
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.318949697520986e-05,
+      "loss": 1.1348,
+      "step": 6846
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.316331600763175e-05,
+      "loss": 1.137,
+      "step": 6847
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.313714079368758e-05,
+      "loss": 1.0832,
+      "step": 6848
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.311097133602707e-05,
+      "loss": 1.0926,
+      "step": 6849
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.308480763729944e-05,
+      "loss": 1.1262,
+      "step": 6850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.305864970015321e-05,
+      "loss": 1.1297,
+      "step": 6851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.303249752723638e-05,
+      "loss": 1.1086,
+      "step": 6852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.300635112119643e-05,
+      "loss": 1.106,
+      "step": 6853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.298021048468014e-05,
+      "loss": 1.0951,
+      "step": 6854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2954075620333714e-05,
+      "loss": 1.0377,
+      "step": 6855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.292794653080288e-05,
+      "loss": 1.1318,
+      "step": 6856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.290182321873264e-05,
+      "loss": 1.1274,
+      "step": 6857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.287570568676755e-05,
+      "loss": 1.1555,
+      "step": 6858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2849593937551494e-05,
+      "loss": 1.0411,
+      "step": 6859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2823487973727736e-05,
+      "loss": 1.1286,
+      "step": 6860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.27973877979391e-05,
+      "loss": 1.1278,
+      "step": 6861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.277129341282764e-05,
+      "loss": 1.1188,
+      "step": 6862
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2745204821035015e-05,
+      "loss": 1.1194,
+      "step": 6863
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.271912202520213e-05,
+      "loss": 1.0765,
+      "step": 6864
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2693045027969356e-05,
+      "loss": 1.1465,
+      "step": 6865
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.266697383197652e-05,
+      "loss": 1.1036,
+      "step": 6866
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.264090843986289e-05,
+      "loss": 1.1632,
+      "step": 6867
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2614848854267e-05,
+      "loss": 1.0829,
+      "step": 6868
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2588795077826974e-05,
+      "loss": 1.0473,
+      "step": 6869
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.256274711318017e-05,
+      "loss": 1.0849,
+      "step": 6870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2536704962963534e-05,
+      "loss": 1.095,
+      "step": 6871
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.251066862981331e-05,
+      "loss": 1.1605,
+      "step": 6872
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2484638116365136e-05,
+      "loss": 1.0647,
+      "step": 6873
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.245861342525418e-05,
+      "loss": 1.0797,
+      "step": 6874
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.243259455911487e-05,
+      "loss": 1.1003,
+      "step": 6875
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.24065815205812e-05,
+      "loss": 1.1577,
+      "step": 6876
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.238057431228647e-05,
+      "loss": 1.1193,
+      "step": 6877
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2354572936863366e-05,
+      "loss": 1.1187,
+      "step": 6878
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2328577396944105e-05,
+      "loss": 1.0694,
+      "step": 6879
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2302587695160165e-05,
+      "loss": 1.1197,
+      "step": 6880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.227660383414256e-05,
+      "loss": 1.0564,
+      "step": 6881
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2250625816521696e-05,
+      "loss": 1.0937,
+      "step": 6882
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.222465364492727e-05,
+      "loss": 1.101,
+      "step": 6883
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.219868732198855e-05,
+      "loss": 1.1111,
+      "step": 6884
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.217272685033409e-05,
+      "loss": 1.1402,
+      "step": 6885
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.214677223259187e-05,
+      "loss": 1.1263,
+      "step": 6886
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.212082347138936e-05,
+      "loss": 1.1111,
+      "step": 6887
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.209488056935332e-05,
+      "loss": 1.089,
+      "step": 6888
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.206894352911005e-05,
+      "loss": 1.1186,
+      "step": 6889
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.2043012353285126e-05,
+      "loss": 1.1014,
+      "step": 6890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.2017087044503564e-05,
+      "loss": 1.065,
+      "step": 6891
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.199116760538989e-05,
+      "loss": 1.1597,
+      "step": 6892
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.196525403856786e-05,
+      "loss": 1.0746,
+      "step": 6893
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.193934634666079e-05,
+      "loss": 1.0677,
+      "step": 6894
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.191344453229136e-05,
+      "loss": 1.1431,
+      "step": 6895
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.188754859808156e-05,
+      "loss": 1.1302,
+      "step": 6896
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.186165854665296e-05,
+      "loss": 1.1638,
+      "step": 6897
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.183577438062638e-05,
+      "loss": 1.1263,
+      "step": 6898
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1809896102622057e-05,
+      "loss": 1.084,
+      "step": 6899
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.178402371525977e-05,
+      "loss": 1.1509,
+      "step": 6900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1758157221158546e-05,
+      "loss": 1.1293,
+      "step": 6901
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.173229662293685e-05,
+      "loss": 1.0986,
+      "step": 6902
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.170644192321266e-05,
+      "loss": 1.0931,
+      "step": 6903
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.168059312460317e-05,
+      "loss": 1.1376,
+      "step": 6904
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.165475022972519e-05,
+      "loss": 1.1322,
+      "step": 6905
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1628913241194766e-05,
+      "loss": 1.1389,
+      "step": 6906
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.160308216162733e-05,
+      "loss": 1.1489,
+      "step": 6907
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1577256993637924e-05,
+      "loss": 1.1338,
+      "step": 6908
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1551437739840804e-05,
+      "loss": 1.1534,
+      "step": 6909
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.152562440284963e-05,
+      "loss": 1.049,
+      "step": 6910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.149981698527757e-05,
+      "loss": 1.0689,
+      "step": 6911
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1474015489737093e-05,
+      "loss": 1.0899,
+      "step": 6912
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.144821991884016e-05,
+      "loss": 1.1455,
+      "step": 6913
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.142243027519806e-05,
+      "loss": 1.1194,
+      "step": 6914
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.139664656142145e-05,
+      "loss": 1.1794,
+      "step": 6915
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.137086878012053e-05,
+      "loss": 1.0075,
+      "step": 6916
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.1345096933904735e-05,
+      "loss": 1.1021,
+      "step": 6917
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.131933102538304e-05,
+      "loss": 1.1568,
+      "step": 6918
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.1293571057163714e-05,
+      "loss": 1.1607,
+      "step": 6919
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.126781703185445e-05,
+      "loss": 1.0783,
+      "step": 6920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.1242068952062365e-05,
+      "loss": 1.1366,
+      "step": 6921
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.121632682039402e-05,
+      "loss": 1.1111,
+      "step": 6922
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.1190590639455215e-05,
+      "loss": 1.1392,
+      "step": 6923
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.1164860411851346e-05,
+      "loss": 1.1159,
+      "step": 6924
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.113913614018704e-05,
+      "loss": 1.1363,
+      "step": 6925
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.111341782706645e-05,
+      "loss": 1.0753,
+      "step": 6926
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.108770547509303e-05,
+      "loss": 1.0891,
+      "step": 6927
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.106199908686964e-05,
+      "loss": 1.1031,
+      "step": 6928
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.103629866499862e-05,
+      "loss": 1.1562,
+      "step": 6929
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.1010604212081594e-05,
+      "loss": 1.1282,
+      "step": 6930
+    },
+    {
+      "epoch": 2.78,
+      "eval_loss": 0.8156223297119141,
+      "eval_runtime": 503.4665,
+      "eval_samples_per_second": 51.443,
+      "eval_steps_per_second": 25.722,
+      "step": 6930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0984915730719707e-05,
+      "loss": 1.0827,
+      "step": 6931
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.095923322351338e-05,
+      "loss": 1.1226,
+      "step": 6932
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.093355669306246e-05,
+      "loss": 1.1006,
+      "step": 6933
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0907886141966256e-05,
+      "loss": 1.1334,
+      "step": 6934
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0882221572823375e-05,
+      "loss": 1.1459,
+      "step": 6935
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0856562988231904e-05,
+      "loss": 1.1278,
+      "step": 6936
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.08309103907893e-05,
+      "loss": 1.1069,
+      "step": 6937
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0805263783092354e-05,
+      "loss": 1.0624,
+      "step": 6938
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0779623167737346e-05,
+      "loss": 1.1181,
+      "step": 6939
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.075398854731988e-05,
+      "loss": 1.1208,
+      "step": 6940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0728359924434936e-05,
+      "loss": 1.1248,
+      "step": 6941
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0702737301676985e-05,
+      "loss": 1.0377,
+      "step": 6942
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.067712068163978e-05,
+      "loss": 1.0804,
+      "step": 6943
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0651510066916575e-05,
+      "loss": 1.0956,
+      "step": 6944
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0625905460099935e-05,
+      "loss": 1.1169,
+      "step": 6945
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.060030686378178e-05,
+      "loss": 1.1212,
+      "step": 6946
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.057471428055357e-05,
+      "loss": 1.1193,
+      "step": 6947
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0549127713006e-05,
+      "loss": 1.1247,
+      "step": 6948
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.052354716372925e-05,
+      "loss": 1.1194,
+      "step": 6949
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0497972635312896e-05,
+      "loss": 1.1166,
+      "step": 6950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0472404130345855e-05,
+      "loss": 1.1352,
+      "step": 6951
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0446841651416386e-05,
+      "loss": 1.0887,
+      "step": 6952
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.042128520111231e-05,
+      "loss": 1.0768,
+      "step": 6953
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0395734782020644e-05,
+      "loss": 1.1402,
+      "step": 6954
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0370190396727947e-05,
+      "loss": 1.1328,
+      "step": 6955
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.034465204782009e-05,
+      "loss": 1.1352,
+      "step": 6956
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.031911973788227e-05,
+      "loss": 1.0838,
+      "step": 6957
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.029359346949926e-05,
+      "loss": 1.1077,
+      "step": 6958
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.026807324525502e-05,
+      "loss": 1.1156,
+      "step": 6959
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.024255906773305e-05,
+      "loss": 1.0994,
+      "step": 6960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0217050939516167e-05,
+      "loss": 1.0444,
+      "step": 6961
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.019154886318649e-05,
+      "loss": 1.0753,
+      "step": 6962
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.016605284132577e-05,
+      "loss": 1.1029,
+      "step": 6963
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.014056287651492e-05,
+      "loss": 1.0787,
+      "step": 6964
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.011507897133427e-05,
+      "loss": 1.1286,
+      "step": 6965
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.008960112836369e-05,
+      "loss": 1.1445,
+      "step": 6966
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.00641293501822e-05,
+      "loss": 1.1623,
+      "step": 6967
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.0038663639368447e-05,
+      "loss": 1.1213,
+      "step": 6968
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.001320399850031e-05,
+      "loss": 1.1881,
+      "step": 6969
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.998775043015504e-05,
+      "loss": 1.0957,
+      "step": 6970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.99623029369094e-05,
+      "loss": 1.119,
+      "step": 6971
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.993686152133941e-05,
+      "loss": 1.1034,
+      "step": 6972
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.991142618602058e-05,
+      "loss": 1.0936,
+      "step": 6973
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.988599693352775e-05,
+      "loss": 1.1325,
+      "step": 6974
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.986057376643508e-05,
+      "loss": 1.0844,
+      "step": 6975
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9835156687316246e-05,
+      "loss": 1.0817,
+      "step": 6976
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.980974569874425e-05,
+      "loss": 1.053,
+      "step": 6977
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.978434080329143e-05,
+      "loss": 1.1148,
+      "step": 6978
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.97589420035296e-05,
+      "loss": 1.15,
+      "step": 6979
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.973354930202984e-05,
+      "loss": 1.1416,
+      "step": 6980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.970816270136274e-05,
+      "loss": 1.1552,
+      "step": 6981
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.96827822040982e-05,
+      "loss": 1.0959,
+      "step": 6982
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.965740781280546e-05,
+      "loss": 1.0993,
+      "step": 6983
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.963203953005326e-05,
+      "loss": 1.0486,
+      "step": 6984
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.96066773584096e-05,
+      "loss": 1.0581,
+      "step": 6985
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.958132130044199e-05,
+      "loss": 1.1896,
+      "step": 6986
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.955597135871719e-05,
+      "loss": 1.1209,
+      "step": 6987
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.953062753580139e-05,
+      "loss": 1.1115,
+      "step": 6988
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.950528983426023e-05,
+      "loss": 1.0451,
+      "step": 6989
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.94799582566586e-05,
+      "loss": 1.1185,
+      "step": 6990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.945463280556089e-05,
+      "loss": 1.0803,
+      "step": 6991
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9429313483530826e-05,
+      "loss": 1.0354,
+      "step": 6992
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9404000293131504e-05,
+      "loss": 1.0647,
+      "step": 6993
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9378693236925345e-05,
+      "loss": 1.068,
+      "step": 6994
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.935339231747429e-05,
+      "loss": 1.1068,
+      "step": 6995
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.932809753733951e-05,
+      "loss": 1.0957,
+      "step": 6996
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.930280889908168e-05,
+      "loss": 1.1651,
+      "step": 6997
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9277526405260754e-05,
+      "loss": 1.18,
+      "step": 6998
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.925225005843609e-05,
+      "loss": 1.1077,
+      "step": 6999
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.922697986116649e-05,
+      "loss": 1.1451,
+      "step": 7000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.920171581601002e-05,
+      "loss": 1.1175,
+      "step": 7001
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9176457925524254e-05,
+      "loss": 1.1456,
+      "step": 7002
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9151206192266e-05,
+      "loss": 1.0952,
+      "step": 7003
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9125960618791546e-05,
+      "loss": 1.1223,
+      "step": 7004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9100721207656565e-05,
+      "loss": 1.0707,
+      "step": 7005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.9075487961416046e-05,
+      "loss": 1.1498,
+      "step": 7006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.905026088262432e-05,
+      "loss": 1.097,
+      "step": 7007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.902503997383523e-05,
+      "loss": 1.1229,
+      "step": 7008
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8999825237601836e-05,
+      "loss": 1.107,
+      "step": 7009
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.897461667647674e-05,
+      "loss": 1.1076,
+      "step": 7010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.894941429301178e-05,
+      "loss": 1.1193,
+      "step": 7011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.892421808975818e-05,
+      "loss": 1.1084,
+      "step": 7012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.889902806926665e-05,
+      "loss": 1.1048,
+      "step": 7013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.887384423408713e-05,
+      "loss": 1.112,
+      "step": 7014
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.884866658676909e-05,
+      "loss": 1.206,
+      "step": 7015
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.882349512986123e-05,
+      "loss": 1.105,
+      "step": 7016
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.879832986591162e-05,
+      "loss": 1.0632,
+      "step": 7017
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8773170797467914e-05,
+      "loss": 1.0722,
+      "step": 7018
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8748017927076906e-05,
+      "loss": 1.0911,
+      "step": 7019
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8722871257284823e-05,
+      "loss": 1.0652,
+      "step": 7020
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.869773079063734e-05,
+      "loss": 1.1234,
+      "step": 7021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8672596529679404e-05,
+      "loss": 1.1221,
+      "step": 7022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.864746847695544e-05,
+      "loss": 1.0832,
+      "step": 7023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8622346635009146e-05,
+      "loss": 1.1392,
+      "step": 7024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8597231006383595e-05,
+      "loss": 1.115,
+      "step": 7025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8572121593621344e-05,
+      "loss": 1.1003,
+      "step": 7026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.854701839926419e-05,
+      "loss": 1.1098,
+      "step": 7027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.852192142585339e-05,
+      "loss": 1.113,
+      "step": 7028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8496830675929516e-05,
+      "loss": 1.0886,
+      "step": 7029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8471746152032506e-05,
+      "loss": 1.0717,
+      "step": 7030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8446667856701715e-05,
+      "loss": 1.0935,
+      "step": 7031
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.842159579247587e-05,
+      "loss": 1.1246,
+      "step": 7032
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.839652996189299e-05,
+      "loss": 1.1413,
+      "step": 7033
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.837147036749057e-05,
+      "loss": 1.119,
+      "step": 7034
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.834641701180539e-05,
+      "loss": 1.0928,
+      "step": 7035
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.832136989737358e-05,
+      "loss": 1.076,
+      "step": 7036
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.829632902673077e-05,
+      "loss": 1.1276,
+      "step": 7037
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.827129440241179e-05,
+      "loss": 1.1624,
+      "step": 7038
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8246266026951e-05,
+      "loss": 1.0925,
+      "step": 7039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.822124390288201e-05,
+      "loss": 1.0804,
+      "step": 7040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.819622803273779e-05,
+      "loss": 1.1729,
+      "step": 7041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.81712184190508e-05,
+      "loss": 1.0751,
+      "step": 7042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.814621506435272e-05,
+      "loss": 1.1084,
+      "step": 7043
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.812121797117473e-05,
+      "loss": 1.0519,
+      "step": 7044
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.809622714204724e-05,
+      "loss": 1.0515,
+      "step": 7045
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.807124257950012e-05,
+      "loss": 1.1141,
+      "step": 7046
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.804626428606264e-05,
+      "loss": 1.1398,
+      "step": 7047
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.802129226426333e-05,
+      "loss": 1.1221,
+      "step": 7048
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.79963265166301e-05,
+      "loss": 1.0742,
+      "step": 7049
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.797136704569032e-05,
+      "loss": 1.056,
+      "step": 7050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.794641385397061e-05,
+      "loss": 1.1214,
+      "step": 7051
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7921466943997064e-05,
+      "loss": 1.1201,
+      "step": 7052
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.789652631829504e-05,
+      "loss": 1.1291,
+      "step": 7053
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.787159197938928e-05,
+      "loss": 1.0684,
+      "step": 7054
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.784666392980398e-05,
+      "loss": 1.0799,
+      "step": 7055
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.782174217206256e-05,
+      "loss": 1.1518,
+      "step": 7056
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.779682670868794e-05,
+      "loss": 1.0177,
+      "step": 7057
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.777191754220226e-05,
+      "loss": 1.1106,
+      "step": 7058
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.774701467512716e-05,
+      "loss": 1.0846,
+      "step": 7059
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7722118109983595e-05,
+      "loss": 1.0985,
+      "step": 7060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.769722784929185e-05,
+      "loss": 1.1418,
+      "step": 7061
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.767234389557154e-05,
+      "loss": 1.0877,
+      "step": 7062
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.764746625134177e-05,
+      "loss": 1.1498,
+      "step": 7063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.762259491912086e-05,
+      "loss": 1.1541,
+      "step": 7064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.759772990142661e-05,
+      "loss": 1.0115,
+      "step": 7065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.757287120077614e-05,
+      "loss": 1.0223,
+      "step": 7066
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7548018819685834e-05,
+      "loss": 1.0858,
+      "step": 7067
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.752317276067163e-05,
+      "loss": 1.0805,
+      "step": 7068
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7498333026248646e-05,
+      "loss": 1.0239,
+      "step": 7069
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7473499618931484e-05,
+      "loss": 1.0238,
+      "step": 7070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.744867254123403e-05,
+      "loss": 1.1118,
+      "step": 7071
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7423851795669485e-05,
+      "loss": 1.1054,
+      "step": 7072
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7399037384750615e-05,
+      "loss": 1.1536,
+      "step": 7073
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.737422931098934e-05,
+      "loss": 1.0874,
+      "step": 7074
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.734942757689697e-05,
+      "loss": 1.1025,
+      "step": 7075
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.732463218498426e-05,
+      "loss": 1.1082,
+      "step": 7076
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7299843137761245e-05,
+      "loss": 1.1477,
+      "step": 7077
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7275060437737387e-05,
+      "loss": 1.1143,
+      "step": 7078
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7250284087421424e-05,
+      "loss": 1.0579,
+      "step": 7079
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.722551408932147e-05,
+      "loss": 1.1264,
+      "step": 7080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7200750445945074e-05,
+      "loss": 1.0972,
+      "step": 7081
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.717599315979906e-05,
+      "loss": 1.1138,
+      "step": 7082
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7151242233389594e-05,
+      "loss": 1.0677,
+      "step": 7083
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.71264976692223e-05,
+      "loss": 1.1085,
+      "step": 7084
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.710175946980203e-05,
+      "loss": 1.0549,
+      "step": 7085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.70770276376331e-05,
+      "loss": 1.166,
+      "step": 7086
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.7052302175219156e-05,
+      "loss": 1.0927,
+      "step": 7087
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.702758308506312e-05,
+      "loss": 1.0394,
+      "step": 7088
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.700287036966741e-05,
+      "loss": 1.1175,
+      "step": 7089
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.697816403153367e-05,
+      "loss": 1.0755,
+      "step": 7090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.695346407316291e-05,
+      "loss": 1.1176,
+      "step": 7091
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.692877049705561e-05,
+      "loss": 1.1071,
+      "step": 7092
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6904083305711455e-05,
+      "loss": 1.177,
+      "step": 7093
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6879402501629625e-05,
+      "loss": 1.0739,
+      "step": 7094
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.685472808730853e-05,
+      "loss": 1.1707,
+      "step": 7095
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6830060065245974e-05,
+      "loss": 1.118,
+      "step": 7096
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.680539843793919e-05,
+      "loss": 1.0934,
+      "step": 7097
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6780743207884616e-05,
+      "loss": 1.1249,
+      "step": 7098
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.67560943775782e-05,
+      "loss": 1.1431,
+      "step": 7099
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.67314519495151e-05,
+      "loss": 1.0706,
+      "step": 7100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.670681592618995e-05,
+      "loss": 1.1121,
+      "step": 7101
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6682186310096676e-05,
+      "loss": 1.137,
+      "step": 7102
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.665756310372854e-05,
+      "loss": 1.0921,
+      "step": 7103
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6632946309578154e-05,
+      "loss": 1.1262,
+      "step": 7104
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.660833593013755e-05,
+      "loss": 1.1346,
+      "step": 7105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6583731967897994e-05,
+      "loss": 1.0621,
+      "step": 7106
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.655913442535025e-05,
+      "loss": 1.1328,
+      "step": 7107
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.65345433049843e-05,
+      "loss": 1.083,
+      "step": 7108
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.650995860928952e-05,
+      "loss": 1.1078,
+      "step": 7109
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.648538034075468e-05,
+      "loss": 1.1074,
+      "step": 7110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6460808501867804e-05,
+      "loss": 1.1107,
+      "step": 7111
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6436243095116406e-05,
+      "loss": 1.1197,
+      "step": 7112
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.641168412298719e-05,
+      "loss": 1.1149,
+      "step": 7113
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6387131587966315e-05,
+      "loss": 1.1194,
+      "step": 7114
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.63625854925393e-05,
+      "loss": 1.0712,
+      "step": 7115
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.633804583919094e-05,
+      "loss": 1.0918,
+      "step": 7116
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.631351263040536e-05,
+      "loss": 1.0773,
+      "step": 7117
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6288985868666145e-05,
+      "loss": 1.0909,
+      "step": 7118
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6264465556456115e-05,
+      "loss": 1.0373,
+      "step": 7119
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.623995169625754e-05,
+      "loss": 1.1469,
+      "step": 7120
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.621544429055196e-05,
+      "loss": 1.1659,
+      "step": 7121
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.619094334182025e-05,
+      "loss": 1.0597,
+      "step": 7122
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.616644885254271e-05,
+      "loss": 1.172,
+      "step": 7123
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.614196082519891e-05,
+      "loss": 1.1095,
+      "step": 7124
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.611747926226783e-05,
+      "loss": 1.0987,
+      "step": 7125
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.609300416622775e-05,
+      "loss": 1.0727,
+      "step": 7126
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.606853553955626e-05,
+      "loss": 1.1499,
+      "step": 7127
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.604407338473039e-05,
+      "loss": 1.0326,
+      "step": 7128
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.60196177042265e-05,
+      "loss": 1.1802,
+      "step": 7129
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.59951685005202e-05,
+      "loss": 1.0722,
+      "step": 7130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5970725776086565e-05,
+      "loss": 1.0879,
+      "step": 7131
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5946289533399925e-05,
+      "loss": 1.0857,
+      "step": 7132
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.592185977493395e-05,
+      "loss": 1.0925,
+      "step": 7133
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.589743650316176e-05,
+      "loss": 1.1289,
+      "step": 7134
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.587301972055568e-05,
+      "loss": 1.1502,
+      "step": 7135
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.584860942958751e-05,
+      "loss": 1.0773,
+      "step": 7136
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5824205632728294e-05,
+      "loss": 1.1612,
+      "step": 7137
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5799808332448415e-05,
+      "loss": 1.0178,
+      "step": 7138
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.577541753121771e-05,
+      "loss": 1.1278,
+      "step": 7139
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.575103323150522e-05,
+      "loss": 1.1134,
+      "step": 7140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.572665543577941e-05,
+      "loss": 1.0977,
+      "step": 7141
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.570228414650811e-05,
+      "loss": 1.102,
+      "step": 7142
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5677919366158396e-05,
+      "loss": 1.0824,
+      "step": 7143
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5653561097196794e-05,
+      "loss": 1.1362,
+      "step": 7144
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5629209342089074e-05,
+      "loss": 1.1829,
+      "step": 7145
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5604864103300374e-05,
+      "loss": 1.0859,
+      "step": 7146
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.558052538329524e-05,
+      "loss": 1.058,
+      "step": 7147
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5556193184537435e-05,
+      "loss": 1.1095,
+      "step": 7148
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.553186750949021e-05,
+      "loss": 1.1109,
+      "step": 7149
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.550754836061604e-05,
+      "loss": 1.0845,
+      "step": 7150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.548323574037674e-05,
+      "loss": 1.0656,
+      "step": 7151
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5458929651233576e-05,
+      "loss": 1.0381,
+      "step": 7152
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.543463009564699e-05,
+      "loss": 1.1042,
+      "step": 7153
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.541033707607694e-05,
+      "loss": 1.1065,
+      "step": 7154
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.538605059498256e-05,
+      "loss": 1.0737,
+      "step": 7155
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.536177065482241e-05,
+      "loss": 1.1214,
+      "step": 7156
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.533749725805444e-05,
+      "loss": 1.0961,
+      "step": 7157
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.53132304071358e-05,
+      "loss": 1.0888,
+      "step": 7158
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5288970104523044e-05,
+      "loss": 1.0724,
+      "step": 7159
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5264716352672114e-05,
+      "loss": 1.1246,
+      "step": 7160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.524046915403818e-05,
+      "loss": 1.1152,
+      "step": 7161
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5216228511075875e-05,
+      "loss": 1.0881,
+      "step": 7162
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.519199442623908e-05,
+      "loss": 1.0882,
+      "step": 7163
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5167766901981004e-05,
+      "loss": 1.1084,
+      "step": 7164
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5143545940754266e-05,
+      "loss": 1.1617,
+      "step": 7165
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5119331545010725e-05,
+      "loss": 1.0881,
+      "step": 7166
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.509512371720171e-05,
+      "loss": 1.1829,
+      "step": 7167
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5070922459777714e-05,
+      "loss": 1.0669,
+      "step": 7168
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.504672777518873e-05,
+      "loss": 1.1456,
+      "step": 7169
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.502253966588394e-05,
+      "loss": 1.1018,
+      "step": 7170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4998358134312e-05,
+      "loss": 1.1354,
+      "step": 7171
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4974183182920764e-05,
+      "loss": 1.1015,
+      "step": 7172
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.495001481415756e-05,
+      "loss": 1.0851,
+      "step": 7173
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4925853030468936e-05,
+      "loss": 1.1114,
+      "step": 7174
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.490169783430077e-05,
+      "loss": 1.0911,
+      "step": 7175
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.48775492280984e-05,
+      "loss": 1.1095,
+      "step": 7176
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4853407214306335e-05,
+      "loss": 1.1522,
+      "step": 7177
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.482927179536857e-05,
+      "loss": 1.1297,
+      "step": 7178
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.480514297372832e-05,
+      "loss": 1.0706,
+      "step": 7179
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.478102075182813e-05,
+      "loss": 1.0877,
+      "step": 7180
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4756905132109994e-05,
+      "loss": 1.1054,
+      "step": 7181
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.47327961170151e-05,
+      "loss": 1.1361,
+      "step": 7182
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.470869370898404e-05,
+      "loss": 1.0588,
+      "step": 7183
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4684597910456764e-05,
+      "loss": 1.1737,
+      "step": 7184
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.466050872387245e-05,
+      "loss": 1.1389,
+      "step": 7185
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.463642615166976e-05,
+      "loss": 1.1388,
+      "step": 7186
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.461235019628651e-05,
+      "loss": 1.1024,
+      "step": 7187
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4588280860159946e-05,
+      "loss": 1.0561,
+      "step": 7188
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.456421814572669e-05,
+      "loss": 1.1148,
+      "step": 7189
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.454016205542254e-05,
+      "loss": 1.1192,
+      "step": 7190
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.451611259168282e-05,
+      "loss": 1.067,
+      "step": 7191
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.449206975694204e-05,
+      "loss": 1.0669,
+      "step": 7192
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4468033553634015e-05,
+      "loss": 1.0687,
+      "step": 7193
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.444400398419206e-05,
+      "loss": 1.1115,
+      "step": 7194
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.441998105104863e-05,
+      "loss": 1.0785,
+      "step": 7195
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.439596475663562e-05,
+      "loss": 1.0872,
+      "step": 7196
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.437195510338426e-05,
+      "loss": 1.0615,
+      "step": 7197
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.434795209372501e-05,
+      "loss": 1.1397,
+      "step": 7198
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.432395573008777e-05,
+      "loss": 1.1255,
+      "step": 7199
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4299966014901695e-05,
+      "loss": 1.1338,
+      "step": 7200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4275982950595254e-05,
+      "loss": 1.0155,
+      "step": 7201
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.425200653959635e-05,
+      "loss": 1.1192,
+      "step": 7202
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.422803678433205e-05,
+      "loss": 1.1684,
+      "step": 7203
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.420407368722892e-05,
+      "loss": 1.1197,
+      "step": 7204
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.418011725071273e-05,
+      "loss": 1.1603,
+      "step": 7205
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.415616747720858e-05,
+      "loss": 1.0683,
+      "step": 7206
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4132224369140994e-05,
+      "loss": 1.0881,
+      "step": 7207
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.41082879289337e-05,
+      "loss": 1.1113,
+      "step": 7208
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.408435815900986e-05,
+      "loss": 1.0971,
+      "step": 7209
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.406043506179186e-05,
+      "loss": 1.0968,
+      "step": 7210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.4036518639701475e-05,
+      "loss": 1.0926,
+      "step": 7211
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.401260889515982e-05,
+      "loss": 1.1629,
+      "step": 7212
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.398870583058728e-05,
+      "loss": 1.0965,
+      "step": 7213
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.396480944840356e-05,
+      "loss": 1.092,
+      "step": 7214
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.394091975102776e-05,
+      "loss": 1.1407,
+      "step": 7215
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.391703674087824e-05,
+      "loss": 1.136,
+      "step": 7216
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3893160420372646e-05,
+      "loss": 1.0633,
+      "step": 7217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.386929079192811e-05,
+      "loss": 1.1043,
+      "step": 7218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3845427857960865e-05,
+      "loss": 1.1325,
+      "step": 7219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3821571620886686e-05,
+      "loss": 1.1287,
+      "step": 7220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.37977220831205e-05,
+      "loss": 1.1285,
+      "step": 7221
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.377387924707659e-05,
+      "loss": 1.0917,
+      "step": 7222
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.375004311516864e-05,
+      "loss": 1.1157,
+      "step": 7223
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3726213689809624e-05,
+      "loss": 1.0339,
+      "step": 7224
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.370239097341177e-05,
+      "loss": 1.0906,
+      "step": 7225
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.367857496838672e-05,
+      "loss": 1.1172,
+      "step": 7226
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.365476567714533e-05,
+      "loss": 1.1423,
+      "step": 7227
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.363096310209791e-05,
+      "loss": 1.0845,
+      "step": 7228
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.360716724565399e-05,
+      "loss": 1.0976,
+      "step": 7229
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3583378110222406e-05,
+      "loss": 1.1466,
+      "step": 7230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.355959569821143e-05,
+      "loss": 1.1162,
+      "step": 7231
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.35358200120285e-05,
+      "loss": 1.1576,
+      "step": 7232
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3512051054080526e-05,
+      "loss": 1.0636,
+      "step": 7233
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.348828882677364e-05,
+      "loss": 1.0622,
+      "step": 7234
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.346453333251327e-05,
+      "loss": 1.1078,
+      "step": 7235
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3440784573704275e-05,
+      "loss": 1.0879,
+      "step": 7236
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.34170425527507e-05,
+      "loss": 1.0734,
+      "step": 7237
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.339330727205601e-05,
+      "loss": 1.0972,
+      "step": 7238
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.336957873402299e-05,
+      "loss": 1.1074,
+      "step": 7239
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.334585694105362e-05,
+      "loss": 1.1182,
+      "step": 7240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.332214189554935e-05,
+      "loss": 1.0763,
+      "step": 7241
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.329843359991087e-05,
+      "loss": 1.1325,
+      "step": 7242
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3274732056538124e-05,
+      "loss": 1.1299,
+      "step": 7243
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.325103726783054e-05,
+      "loss": 1.1151,
+      "step": 7244
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.322734923618669e-05,
+      "loss": 1.1023,
+      "step": 7245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.320366796400459e-05,
+      "loss": 1.1142,
+      "step": 7246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3179993453681504e-05,
+      "loss": 1.1533,
+      "step": 7247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3156325707613975e-05,
+      "loss": 1.1289,
+      "step": 7248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3132664728197996e-05,
+      "loss": 1.0612,
+      "step": 7249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.310901051782871e-05,
+      "loss": 1.107,
+      "step": 7250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.308536307890069e-05,
+      "loss": 1.0974,
+      "step": 7251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.306172241380784e-05,
+      "loss": 1.0528,
+      "step": 7252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3038088524943247e-05,
+      "loss": 1.0912,
+      "step": 7253
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.301446141469946e-05,
+      "loss": 1.1523,
+      "step": 7254
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.299084108546825e-05,
+      "loss": 1.0955,
+      "step": 7255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.296722753964069e-05,
+      "loss": 1.1262,
+      "step": 7256
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2943620779607263e-05,
+      "loss": 1.1652,
+      "step": 7257
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.292002080775765e-05,
+      "loss": 1.1571,
+      "step": 7258
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.289642762648094e-05,
+      "loss": 1.1059,
+      "step": 7259
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.28728412381655e-05,
+      "loss": 1.1176,
+      "step": 7260
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.284926164519894e-05,
+      "loss": 1.1835,
+      "step": 7261
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2825688849968317e-05,
+      "loss": 1.1141,
+      "step": 7262
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.280212285485991e-05,
+      "loss": 1.1556,
+      "step": 7263
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.277856366225927e-05,
+      "loss": 1.102,
+      "step": 7264
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.275501127455138e-05,
+      "loss": 1.065,
+      "step": 7265
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.273146569412049e-05,
+      "loss": 1.1125,
+      "step": 7266
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.270792692335008e-05,
+      "loss": 1.0546,
+      "step": 7267
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.268439496462305e-05,
+      "loss": 1.1092,
+      "step": 7268
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.266086982032152e-05,
+      "loss": 1.0975,
+      "step": 7269
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.263735149282702e-05,
+      "loss": 1.0773,
+      "step": 7270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.261383998452031e-05,
+      "loss": 1.0969,
+      "step": 7271
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.259033529778144e-05,
+      "loss": 1.1436,
+      "step": 7272
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.256683743498988e-05,
+      "loss": 1.1872,
+      "step": 7273
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.254334639852429e-05,
+      "loss": 1.1102,
+      "step": 7274
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.251986219076274e-05,
+      "loss": 1.1554,
+      "step": 7275
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2496384814082524e-05,
+      "loss": 1.1278,
+      "step": 7276
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.247291427086027e-05,
+      "loss": 1.0935,
+      "step": 7277
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.244945056347193e-05,
+      "loss": 1.1229,
+      "step": 7278
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2425993694292814e-05,
+      "loss": 1.1019,
+      "step": 7279
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2402543665697394e-05,
+      "loss": 1.1109,
+      "step": 7280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.237910048005963e-05,
+      "loss": 1.1922,
+      "step": 7281
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.235566413975262e-05,
+      "loss": 1.1153,
+      "step": 7282
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2332234647148915e-05,
+      "loss": 1.0863,
+      "step": 7283
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2308812004620284e-05,
+      "loss": 1.1163,
+      "step": 7284
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2285396214537775e-05,
+      "loss": 1.0747,
+      "step": 7285
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.226198727927187e-05,
+      "loss": 1.0764,
+      "step": 7286
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2238585201192204e-05,
+      "loss": 1.1292,
+      "step": 7287
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.221518998266786e-05,
+      "loss": 1.0755,
+      "step": 7288
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.219180162606713e-05,
+      "loss": 1.0853,
+      "step": 7289
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2168420133757605e-05,
+      "loss": 1.0783,
+      "step": 7290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.214504550810629e-05,
+      "loss": 1.1065,
+      "step": 7291
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.212167775147934e-05,
+      "loss": 1.1016,
+      "step": 7292
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.209831686624235e-05,
+      "loss": 1.1356,
+      "step": 7293
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.207496285476017e-05,
+      "loss": 1.1014,
+      "step": 7294
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2051615719396925e-05,
+      "loss": 1.1345,
+      "step": 7295
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.202827546251609e-05,
+      "loss": 1.1566,
+      "step": 7296
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2004942086480415e-05,
+      "loss": 1.1094,
+      "step": 7297
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1981615593651925e-05,
+      "loss": 1.1136,
+      "step": 7298
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.195829598639204e-05,
+      "loss": 1.1128,
+      "step": 7299
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.193498326706138e-05,
+      "loss": 1.077,
+      "step": 7300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.191167743801996e-05,
+      "loss": 1.0891,
+      "step": 7301
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.188837850162704e-05,
+      "loss": 1.0897,
+      "step": 7302
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1865086460241135e-05,
+      "loss": 1.0931,
+      "step": 7303
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.184180131622021e-05,
+      "loss": 1.1092,
+      "step": 7304
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.18185230719214e-05,
+      "loss": 1.1761,
+      "step": 7305
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1795251729701126e-05,
+      "loss": 1.0868,
+      "step": 7306
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1771987291915293e-05,
+      "loss": 1.0763,
+      "step": 7307
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1748729760918925e-05,
+      "loss": 1.1468,
+      "step": 7308
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.172547913906636e-05,
+      "loss": 1.1478,
+      "step": 7309
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.170223542871136e-05,
+      "loss": 1.0915,
+      "step": 7310
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1678998632206846e-05,
+      "loss": 1.0949,
+      "step": 7311
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1655768751905145e-05,
+      "loss": 1.1366,
+      "step": 7312
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.163254579015783e-05,
+      "loss": 1.0945,
+      "step": 7313
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.160932974931575e-05,
+      "loss": 1.0931,
+      "step": 7314
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1586120631729146e-05,
+      "loss": 1.1273,
+      "step": 7315
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1562918439747426e-05,
+      "loss": 1.0984,
+      "step": 7316
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.153972317571946e-05,
+      "loss": 1.0916,
+      "step": 7317
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.151653484199327e-05,
+      "loss": 1.0837,
+      "step": 7318
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1493353440916215e-05,
+      "loss": 1.0826,
+      "step": 7319
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.147017897483501e-05,
+      "loss": 1.1325,
+      "step": 7320
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.144701144609565e-05,
+      "loss": 1.0239,
+      "step": 7321
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1423850857043344e-05,
+      "loss": 1.1373,
+      "step": 7322
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.140069721002273e-05,
+      "loss": 1.1407,
+      "step": 7323
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.137755050737761e-05,
+      "loss": 1.1251,
+      "step": 7324
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1354410751451215e-05,
+      "loss": 1.1406,
+      "step": 7325
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1331277944585955e-05,
+      "loss": 1.1064,
+      "step": 7326
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1308152089123575e-05,
+      "loss": 1.1144,
+      "step": 7327
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1285033187405186e-05,
+      "loss": 1.0559,
+      "step": 7328
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.126192124177108e-05,
+      "loss": 1.0782,
+      "step": 7329
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1238816254560946e-05,
+      "loss": 1.0242,
+      "step": 7330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1215718228113714e-05,
+      "loss": 1.1757,
+      "step": 7331
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.119262716476758e-05,
+      "loss": 1.1267,
+      "step": 7332
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.116954306686011e-05,
+      "loss": 1.1177,
+      "step": 7333
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1146465936728154e-05,
+      "loss": 1.1179,
+      "step": 7334
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.112339577670777e-05,
+      "loss": 1.1663,
+      "step": 7335
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.110033258913445e-05,
+      "loss": 1.1546,
+      "step": 7336
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1077276376342826e-05,
+      "loss": 1.1095,
+      "step": 7337
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.105422714066698e-05,
+      "loss": 1.0706,
+      "step": 7338
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1031184884440157e-05,
+      "loss": 1.1525,
+      "step": 7339
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.100814960999493e-05,
+      "loss": 1.1853,
+      "step": 7340
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.098512131966325e-05,
+      "loss": 1.1239,
+      "step": 7341
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0962100015776205e-05,
+      "loss": 1.1315,
+      "step": 7342
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.093908570066436e-05,
+      "loss": 1.1204,
+      "step": 7343
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.091607837665742e-05,
+      "loss": 1.1607,
+      "step": 7344
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0893078046084424e-05,
+      "loss": 1.101,
+      "step": 7345
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0870084711273783e-05,
+      "loss": 1.0394,
+      "step": 7346
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.084709837455305e-05,
+      "loss": 1.1794,
+      "step": 7347
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.082411903824921e-05,
+      "loss": 1.122,
+      "step": 7348
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.080114670468851e-05,
+      "loss": 1.1188,
+      "step": 7349
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.077818137619643e-05,
+      "loss": 1.1478,
+      "step": 7350
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.075522305509773e-05,
+      "loss": 1.0597,
+      "step": 7351
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.073227174371657e-05,
+      "loss": 1.1194,
+      "step": 7352
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.070932744437628e-05,
+      "loss": 1.0776,
+      "step": 7353
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0686390159399604e-05,
+      "loss": 1.0681,
+      "step": 7354
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0663459891108446e-05,
+      "loss": 1.1565,
+      "step": 7355
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0640536641824055e-05,
+      "loss": 1.0765,
+      "step": 7356
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0617620413867045e-05,
+      "loss": 1.1001,
+      "step": 7357
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.059471120955715e-05,
+      "loss": 1.1086,
+      "step": 7358
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.057180903121358e-05,
+      "loss": 1.154,
+      "step": 7359
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0548913881154704e-05,
+      "loss": 1.0688,
+      "step": 7360
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.052602576169816e-05,
+      "loss": 1.151,
+      "step": 7361
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0503144675161065e-05,
+      "loss": 1.1277,
+      "step": 7362
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0480270623859642e-05,
+      "loss": 1.1443,
+      "step": 7363
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.045740361010939e-05,
+      "loss": 1.1506,
+      "step": 7364
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0434543636225253e-05,
+      "loss": 1.096,
+      "step": 7365
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.041169070452129e-05,
+      "loss": 1.0932,
+      "step": 7366
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.038884481731099e-05,
+      "loss": 1.0695,
+      "step": 7367
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0366005976907043e-05,
+      "loss": 1.0917,
+      "step": 7368
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0343174185621402e-05,
+      "loss": 1.0828,
+      "step": 7369
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.032034944576543e-05,
+      "loss": 1.1138,
+      "step": 7370
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.029753175964962e-05,
+      "loss": 1.0901,
+      "step": 7371
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0274721129583905e-05,
+      "loss": 1.0039,
+      "step": 7372
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0251917557877384e-05,
+      "loss": 1.1006,
+      "step": 7373
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.022912104683847e-05,
+      "loss": 1.0917,
+      "step": 7374
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0206331598774896e-05,
+      "loss": 1.0779,
+      "step": 7375
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.018354921599369e-05,
+      "loss": 1.1422,
+      "step": 7376
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0160773900801075e-05,
+      "loss": 1.0342,
+      "step": 7377
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.013800565550269e-05,
+      "loss": 1.09,
+      "step": 7378
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0115244482403315e-05,
+      "loss": 1.1313,
+      "step": 7379
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.009249038380715e-05,
+      "loss": 1.0907,
+      "step": 7380
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0069743362017588e-05,
+      "loss": 1.1066,
+      "step": 7381
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0047003419337295e-05,
+      "loss": 1.0688,
+      "step": 7382
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.0024270558068322e-05,
+      "loss": 1.0971,
+      "step": 7383
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.000154478051188e-05,
+      "loss": 1.1595,
+      "step": 7384
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9978826088968593e-05,
+      "loss": 1.0904,
+      "step": 7385
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.995611448573824e-05,
+      "loss": 1.1425,
+      "step": 7386
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.993340997311993e-05,
+      "loss": 1.1073,
+      "step": 7387
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.991071255341209e-05,
+      "loss": 1.1297,
+      "step": 7388
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.988802222891244e-05,
+      "loss": 1.1191,
+      "step": 7389
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.986533900191786e-05,
+      "loss": 1.1237,
+      "step": 7390
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.984266287472467e-05,
+      "loss": 1.1546,
+      "step": 7391
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9819993849628348e-05,
+      "loss": 1.0817,
+      "step": 7392
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9797331928923745e-05,
+      "loss": 1.107,
+      "step": 7393
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9774677114904914e-05,
+      "loss": 1.1408,
+      "step": 7394
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.975202940986522e-05,
+      "loss": 1.1192,
+      "step": 7395
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.972938881609736e-05,
+      "loss": 1.1395,
+      "step": 7396
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.970675533589322e-05,
+      "loss": 1.1537,
+      "step": 7397
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9684128971544e-05,
+      "loss": 1.0578,
+      "step": 7398
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.966150972534024e-05,
+      "loss": 1.0926,
+      "step": 7399
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9638897599571645e-05,
+      "loss": 1.0984,
+      "step": 7400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.961629259652734e-05,
+      "loss": 1.1772,
+      "step": 7401
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9593694718495567e-05,
+      "loss": 1.1151,
+      "step": 7402
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.957110396776398e-05,
+      "loss": 1.0693,
+      "step": 7403
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.954852034661949e-05,
+      "loss": 1.0805,
+      "step": 7404
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9525943857348225e-05,
+      "loss": 1.0574,
+      "step": 7405
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9503374502235604e-05,
+      "loss": 1.1213,
+      "step": 7406
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9480812283566383e-05,
+      "loss": 1.1207,
+      "step": 7407
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.945825720362453e-05,
+      "loss": 1.1137,
+      "step": 7408
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9435709264693345e-05,
+      "loss": 1.1383,
+      "step": 7409
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9413168469055375e-05,
+      "loss": 1.1276,
+      "step": 7410
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9390634818992412e-05,
+      "loss": 1.1107,
+      "step": 7411
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9368108316785602e-05,
+      "loss": 1.1597,
+      "step": 7412
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9345588964715286e-05,
+      "loss": 1.066,
+      "step": 7413
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9323076765061176e-05,
+      "loss": 1.1028,
+      "step": 7414
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9300571720102156e-05,
+      "loss": 1.0607,
+      "step": 7415
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9278073832116405e-05,
+      "loss": 1.1309,
+      "step": 7416
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.92555831033815e-05,
+      "loss": 1.0977,
+      "step": 7417
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9233099536174146e-05,
+      "loss": 1.0539,
+      "step": 7418
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9210623132770365e-05,
+      "loss": 1.1353,
+      "step": 7419
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9188153895445503e-05,
+      "loss": 1.085,
+      "step": 7420
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9165691826474085e-05,
+      "loss": 1.1008,
+      "step": 7421
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9143236928130048e-05,
+      "loss": 1.0956,
+      "step": 7422
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.912078920268647e-05,
+      "loss": 1.0979,
+      "step": 7423
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.909834865241574e-05,
+      "loss": 1.046,
+      "step": 7424
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9075915279589606e-05,
+      "loss": 1.0974,
+      "step": 7425
+    },
+    {
+      "epoch": 2.98,
+      "eval_loss": 0.8140046000480652,
+      "eval_runtime": 500.3099,
+      "eval_samples_per_second": 51.768,
+      "eval_steps_per_second": 25.884,
+      "step": 7425
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9053489086478935e-05,
+      "loss": 1.0834,
+      "step": 7426
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9031070075354028e-05,
+      "loss": 1.1541,
+      "step": 7427
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9008658248484366e-05,
+      "loss": 1.1332,
+      "step": 7428
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8986253608138657e-05,
+      "loss": 1.1417,
+      "step": 7429
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8963856156585013e-05,
+      "loss": 1.0411,
+      "step": 7430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8941465896090746e-05,
+      "loss": 1.0875,
+      "step": 7431
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8919082828922417e-05,
+      "loss": 1.1373,
+      "step": 7432
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.889670695734592e-05,
+      "loss": 1.1343,
+      "step": 7433
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8874338283626336e-05,
+      "loss": 1.112,
+      "step": 7434
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8851976810028136e-05,
+      "loss": 1.0857,
+      "step": 7435
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.882962253881495e-05,
+      "loss": 1.1429,
+      "step": 7436
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8807275472249707e-05,
+      "loss": 1.0799,
+      "step": 7437
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8784935612594667e-05,
+      "loss": 1.0916,
+      "step": 7438
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8762602962111294e-05,
+      "loss": 1.1468,
+      "step": 7439
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.874027752306032e-05,
+      "loss": 1.099,
+      "step": 7440
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8717959297701825e-05,
+      "loss": 1.0984,
+      "step": 7441
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.869564828829505e-05,
+      "loss": 1.0794,
+      "step": 7442
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8673344497098608e-05,
+      "loss": 1.0707,
+      "step": 7443
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8651047926370287e-05,
+      "loss": 1.1233,
+      "step": 7444
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.862875857836722e-05,
+      "loss": 1.1536,
+      "step": 7445
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.860647645534581e-05,
+      "loss": 1.0632,
+      "step": 7446
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8584201559561663e-05,
+      "loss": 1.1208,
+      "step": 7447
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8561933893269656e-05,
+      "loss": 1.1126,
+      "step": 7448
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8539673458724035e-05,
+      "loss": 1.0934,
+      "step": 7449
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8517420258178184e-05,
+      "loss": 1.1123,
+      "step": 7450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8495174293884874e-05,
+      "loss": 1.0943,
+      "step": 7451
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.847293556809606e-05,
+      "loss": 1.0785,
+      "step": 7452
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.845070408306296e-05,
+      "loss": 1.0741,
+      "step": 7453
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.8428479841036147e-05,
+      "loss": 1.106,
+      "step": 7454
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.840626284426534e-05,
+      "loss": 1.1078,
+      "step": 7455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.838405309499965e-05,
+      "loss": 1.0973,
+      "step": 7456
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8361850595487327e-05,
+      "loss": 1.1612,
+      "step": 7457
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8339655347976e-05,
+      "loss": 1.0826,
+      "step": 7458
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8317467354712524e-05,
+      "loss": 1.1004,
+      "step": 7459
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8295286617943005e-05,
+      "loss": 1.1586,
+      "step": 7460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8273113139912764e-05,
+      "loss": 1.2077,
+      "step": 7461
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8250946922866526e-05,
+      "loss": 1.1469,
+      "step": 7462
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8228787969048122e-05,
+      "loss": 1.1154,
+      "step": 7463
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8206636280700803e-05,
+      "loss": 1.0337,
+      "step": 7464
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8184491860066965e-05,
+      "loss": 1.127,
+      "step": 7465
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.816235470938827e-05,
+      "loss": 1.112,
+      "step": 7466
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.814022483090577e-05,
+      "loss": 1.0905,
+      "step": 7467
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.811810222685962e-05,
+      "loss": 1.0583,
+      "step": 7468
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8095986899489367e-05,
+      "loss": 1.1413,
+      "step": 7469
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.807387885103374e-05,
+      "loss": 1.0918,
+      "step": 7470
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.805177808373072e-05,
+      "loss": 1.1264,
+      "step": 7471
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.8029684599817674e-05,
+      "loss": 1.1338,
+      "step": 7472
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.800759840153111e-05,
+      "loss": 1.0715,
+      "step": 7473
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7985519491106803e-05,
+      "loss": 1.1174,
+      "step": 7474
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.796344787077988e-05,
+      "loss": 1.1403,
+      "step": 7475
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7941383542784615e-05,
+      "loss": 1.1178,
+      "step": 7476
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.791932650935467e-05,
+      "loss": 1.1251,
+      "step": 7477
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7897276772722847e-05,
+      "loss": 1.0512,
+      "step": 7478
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7875234335121257e-05,
+      "loss": 1.142,
+      "step": 7479
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.785319919878132e-05,
+      "loss": 1.154,
+      "step": 7480
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.783117136593363e-05,
+      "loss": 1.1102,
+      "step": 7481
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.780915083880814e-05,
+      "loss": 1.1867,
+      "step": 7482
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7787137619633974e-05,
+      "loss": 1.1142,
+      "step": 7483
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7765131710639526e-05,
+      "loss": 1.0862,
+      "step": 7484
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.774313311405251e-05,
+      "loss": 1.0566,
+      "step": 7485
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7721141832099895e-05,
+      "loss": 1.086,
+      "step": 7486
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.769915786700782e-05,
+      "loss": 1.1183,
+      "step": 7487
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.767718122100179e-05,
+      "loss": 1.0403,
+      "step": 7488
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.765521189630652e-05,
+      "loss": 1.0704,
+      "step": 7489
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.763324989514593e-05,
+      "loss": 1.0507,
+      "step": 7490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7611295219743326e-05,
+      "loss": 1.143,
+      "step": 7491
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.758934787232115e-05,
+      "loss": 1.1341,
+      "step": 7492
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7567407855101203e-05,
+      "loss": 1.115,
+      "step": 7493
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.754547517030446e-05,
+      "loss": 1.0202,
+      "step": 7494
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7523549820151173e-05,
+      "loss": 1.1876,
+      "step": 7495
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7501631806860918e-05,
+      "loss": 1.0841,
+      "step": 7496
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7479721132652415e-05,
+      "loss": 1.0947,
+      "step": 7497
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.745781779974377e-05,
+      "loss": 1.1097,
+      "step": 7498
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7435921810352216e-05,
+      "loss": 1.0339,
+      "step": 7499
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.741403316669434e-05,
+      "loss": 1.076,
+      "step": 7500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7392151870985973e-05,
+      "loss": 1.0995,
+      "step": 7501
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7370277925442155e-05,
+      "loss": 1.0846,
+      "step": 7502
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.734841133227718e-05,
+      "loss": 1.0548,
+      "step": 7503
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7326552093704682e-05,
+      "loss": 1.1489,
+      "step": 7504
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.7304700211937428e-05,
+      "loss": 1.1142,
+      "step": 7505
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.728285568918757e-05,
+      "loss": 0.984,
+      "step": 7506
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7261018527666426e-05,
+      "loss": 1.0674,
+      "step": 7507
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7239188729584554e-05,
+      "loss": 1.0575,
+      "step": 7508
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7217366297151882e-05,
+      "loss": 1.1114,
+      "step": 7509
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7195551232577433e-05,
+      "loss": 1.0696,
+      "step": 7510
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.717374353806965e-05,
+      "loss": 1.0906,
+      "step": 7511
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7151943215836075e-05,
+      "loss": 1.1081,
+      "step": 7512
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.71301502680836e-05,
+      "loss": 1.1421,
+      "step": 7513
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7108364697018396e-05,
+      "loss": 1.0846,
+      "step": 7514
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.70865865048458e-05,
+      "loss": 1.0503,
+      "step": 7515
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7064815693770396e-05,
+      "loss": 1.0229,
+      "step": 7516
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.704305226599615e-05,
+      "loss": 1.1005,
+      "step": 7517
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.7021296223726112e-05,
+      "loss": 1.1272,
+      "step": 7518
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6999547569162742e-05,
+      "loss": 1.1463,
+      "step": 7519
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6977806304507647e-05,
+      "loss": 1.1239,
+      "step": 7520
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6956072431961678e-05,
+      "loss": 1.0175,
+      "step": 7521
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6934345953725048e-05,
+      "loss": 1.044,
+      "step": 7522
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6912626871997083e-05,
+      "loss": 1.1295,
+      "step": 7523
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6890915188976484e-05,
+      "loss": 1.1302,
+      "step": 7524
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.686921090686112e-05,
+      "loss": 0.9947,
+      "step": 7525
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6847514027848077e-05,
+      "loss": 1.0871,
+      "step": 7526
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6825824554133883e-05,
+      "loss": 1.0985,
+      "step": 7527
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6804142487914096e-05,
+      "loss": 1.0605,
+      "step": 7528
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6782467831383617e-05,
+      "loss": 1.0455,
+      "step": 7529
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6760800586736636e-05,
+      "loss": 1.0388,
+      "step": 7530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.6739140756166514e-05,
+      "loss": 1.1007,
+      "step": 7531
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6717488341865872e-05,
+      "loss": 1.0055,
+      "step": 7532
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.669584334602667e-05,
+      "loss": 1.0455,
+      "step": 7533
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.667420577083999e-05,
+      "loss": 1.074,
+      "step": 7534
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.665257561849629e-05,
+      "loss": 1.0892,
+      "step": 7535
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6630952891185168e-05,
+      "loss": 1.0287,
+      "step": 7536
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6609337591095495e-05,
+      "loss": 1.0841,
+      "step": 7537
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6587729720415476e-05,
+      "loss": 1.0538,
+      "step": 7538
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6566129281332418e-05,
+      "loss": 1.0659,
+      "step": 7539
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6544536276033006e-05,
+      "loss": 1.1001,
+      "step": 7540
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6522950706703132e-05,
+      "loss": 1.0824,
+      "step": 7541
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6501372575527873e-05,
+      "loss": 0.9835,
+      "step": 7542
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.647980188469167e-05,
+      "loss": 1.0829,
+      "step": 7543
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6458238636378107e-05,
+      "loss": 1.1026,
+      "step": 7544
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6436682832770032e-05,
+      "loss": 1.0828,
+      "step": 7545
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6415134476049606e-05,
+      "loss": 1.041,
+      "step": 7546
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6393593568398144e-05,
+      "loss": 1.0388,
+      "step": 7547
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.637206011199631e-05,
+      "loss": 1.0335,
+      "step": 7548
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.635053410902393e-05,
+      "loss": 1.1764,
+      "step": 7549
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6329015561660052e-05,
+      "loss": 1.0692,
+      "step": 7550
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.630750447208311e-05,
+      "loss": 1.1121,
+      "step": 7551
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6286000842470616e-05,
+      "loss": 1.0385,
+      "step": 7552
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6264504674999468e-05,
+      "loss": 1.0543,
+      "step": 7553
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6243015971845685e-05,
+      "loss": 1.0467,
+      "step": 7554
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6221534735184618e-05,
+      "loss": 1.0582,
+      "step": 7555
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.6200060967190863e-05,
+      "loss": 1.1235,
+      "step": 7556
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6178594670038204e-05,
+      "loss": 1.0411,
+      "step": 7557
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.615713584589966e-05,
+      "loss": 1.1517,
+      "step": 7558
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.61356844969476e-05,
+      "loss": 1.0445,
+      "step": 7559
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6114240625353492e-05,
+      "loss": 1.109,
+      "step": 7560
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6092804233288194e-05,
+      "loss": 1.0303,
+      "step": 7561
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6071375322921686e-05,
+      "loss": 1.0381,
+      "step": 7562
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.604995389642322e-05,
+      "loss": 1.0882,
+      "step": 7563
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.6028539955961374e-05,
+      "loss": 1.0501,
+      "step": 7564
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.600713350370383e-05,
+      "loss": 1.0296,
+      "step": 7565
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5985734541817643e-05,
+      "loss": 1.0483,
+      "step": 7566
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5964343072469e-05,
+      "loss": 1.112,
+      "step": 7567
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5942959097823405e-05,
+      "loss": 1.0787,
+      "step": 7568
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5921582620045627e-05,
+      "loss": 1.0441,
+      "step": 7569
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5900213641299563e-05,
+      "loss": 1.1578,
+      "step": 7570
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5878852163748412e-05,
+      "loss": 1.0749,
+      "step": 7571
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5857498189554673e-05,
+      "loss": 1.068,
+      "step": 7572
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5836151720879998e-05,
+      "loss": 1.0137,
+      "step": 7573
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5814812759885287e-05,
+      "loss": 1.0646,
+      "step": 7574
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.579348130873075e-05,
+      "loss": 1.0605,
+      "step": 7575
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5772157369575745e-05,
+      "loss": 1.155,
+      "step": 7576
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.575084094457897e-05,
+      "loss": 1.0176,
+      "step": 7577
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5729532035898275e-05,
+      "loss": 1.0714,
+      "step": 7578
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5708230645690757e-05,
+      "loss": 1.109,
+      "step": 7579
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5686936776112835e-05,
+      "loss": 1.1462,
+      "step": 7580
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.5665650429320042e-05,
+      "loss": 1.0858,
+      "step": 7581
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5644371607467265e-05,
+      "loss": 1.0886,
+      "step": 7582
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.562310031270859e-05,
+      "loss": 1.0925,
+      "step": 7583
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5601836547197278e-05,
+      "loss": 1.0202,
+      "step": 7584
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5580580313085945e-05,
+      "loss": 1.0629,
+      "step": 7585
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.555933161252635e-05,
+      "loss": 1.1011,
+      "step": 7586
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5538090447669484e-05,
+      "loss": 1.0176,
+      "step": 7587
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5516856820665668e-05,
+      "loss": 1.0367,
+      "step": 7588
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5495630733664366e-05,
+      "loss": 1.0266,
+      "step": 7589
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5474412188814344e-05,
+      "loss": 1.0443,
+      "step": 7590
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5453201188263577e-05,
+      "loss": 1.0691,
+      "step": 7591
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.543199773415922e-05,
+      "loss": 1.1051,
+      "step": 7592
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5410801828647802e-05,
+      "loss": 1.1469,
+      "step": 7593
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5389613473874917e-05,
+      "loss": 1.0902,
+      "step": 7594
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5368432671985542e-05,
+      "loss": 1.0843,
+      "step": 7595
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.534725942512385e-05,
+      "loss": 1.0585,
+      "step": 7596
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5326093735433164e-05,
+      "loss": 1.101,
+      "step": 7597
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5304935605056178e-05,
+      "loss": 1.1082,
+      "step": 7598
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.528378503613471e-05,
+      "loss": 1.0871,
+      "step": 7599
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5262642030809837e-05,
+      "loss": 1.0812,
+      "step": 7600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.524150659122193e-05,
+      "loss": 1.0721,
+      "step": 7601
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5220378719510507e-05,
+      "loss": 1.0506,
+      "step": 7602
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.519925841781441e-05,
+      "loss": 1.0839,
+      "step": 7603
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.5178145688271636e-05,
+      "loss": 1.1111,
+      "step": 7604
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.515704053301943e-05,
+      "loss": 1.0615,
+      "step": 7605
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5135942954194337e-05,
+      "loss": 1.0751,
+      "step": 7606
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5114852953932033e-05,
+      "loss": 1.0156,
+      "step": 7607
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5093770534367545e-05,
+      "loss": 1.0235,
+      "step": 7608
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5072695697634995e-05,
+      "loss": 1.0043,
+      "step": 7609
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.5051628445867846e-05,
+      "loss": 1.0287,
+      "step": 7610
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.503056878119878e-05,
+      "loss": 1.1157,
+      "step": 7611
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.500951670575967e-05,
+      "loss": 1.0532,
+      "step": 7612
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4988472221681604e-05,
+      "loss": 1.0307,
+      "step": 7613
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.496743533109498e-05,
+      "loss": 1.0897,
+      "step": 7614
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4946406036129355e-05,
+      "loss": 1.0803,
+      "step": 7615
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4925384338913572e-05,
+      "loss": 1.0893,
+      "step": 7616
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.490437024157567e-05,
+      "loss": 1.0416,
+      "step": 7617
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4883363746242893e-05,
+      "loss": 1.005,
+      "step": 7618
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.486236485504181e-05,
+      "loss": 1.1145,
+      "step": 7619
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4841373570098116e-05,
+      "loss": 1.0828,
+      "step": 7620
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4820389893536767e-05,
+      "loss": 1.0638,
+      "step": 7621
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4799413827481977e-05,
+      "loss": 1.0708,
+      "step": 7622
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4778445374057223e-05,
+      "loss": 1.1013,
+      "step": 7623
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.475748453538509e-05,
+      "loss": 1.0516,
+      "step": 7624
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.473653131358752e-05,
+      "loss": 1.034,
+      "step": 7625
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.471558571078557e-05,
+      "loss": 1.0882,
+      "step": 7626
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4694647729099653e-05,
+      "loss": 1.0262,
+      "step": 7627
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.467371737064931e-05,
+      "loss": 1.1095,
+      "step": 7628
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.46527946375533e-05,
+      "loss": 1.051,
+      "step": 7629
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4631879531929735e-05,
+      "loss": 1.0581,
+      "step": 7630
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4610972055895788e-05,
+      "loss": 1.1057,
+      "step": 7631
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4590072211568017e-05,
+      "loss": 1.0725,
+      "step": 7632
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.45691800010621e-05,
+      "loss": 1.0688,
+      "step": 7633
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.454829542649294e-05,
+      "loss": 1.136,
+      "step": 7634
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4527418489974774e-05,
+      "loss": 1.0211,
+      "step": 7635
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.450654919362093e-05,
+      "loss": 1.055,
+      "step": 7636
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4485687539544056e-05,
+      "loss": 1.096,
+      "step": 7637
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4464833529856036e-05,
+      "loss": 1.0744,
+      "step": 7638
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.444398716666787e-05,
+      "loss": 1.0587,
+      "step": 7639
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.442314845208993e-05,
+      "loss": 1.0587,
+      "step": 7640
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.44023173882317e-05,
+      "loss": 1.1185,
+      "step": 7641
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4381493977201895e-05,
+      "loss": 1.0559,
+      "step": 7642
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4360678221108557e-05,
+      "loss": 1.0812,
+      "step": 7643
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.433987012205883e-05,
+      "loss": 1.0911,
+      "step": 7644
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.431906968215919e-05,
+      "loss": 1.0378,
+      "step": 7645
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.429827690351526e-05,
+      "loss": 1.0614,
+      "step": 7646
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4277491788231887e-05,
+      "loss": 1.0961,
+      "step": 7647
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4256714338413223e-05,
+      "loss": 1.0217,
+      "step": 7648
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4235944556162528e-05,
+      "loss": 1.0557,
+      "step": 7649
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4215182443582385e-05,
+      "loss": 1.0561,
+      "step": 7650
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4194428002774595e-05,
+      "loss": 1.0814,
+      "step": 7651
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4173681235840075e-05,
+      "loss": 1.0601,
+      "step": 7652
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.415294214487911e-05,
+      "loss": 1.0944,
+      "step": 7653
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.413221073199111e-05,
+      "loss": 1.0531,
+      "step": 7654
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.411148699927471e-05,
+      "loss": 1.0715,
+      "step": 7655
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4090770948827857e-05,
+      "loss": 1.1211,
+      "step": 7656
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4070062582747586e-05,
+      "loss": 1.1279,
+      "step": 7657
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4049361903130286e-05,
+      "loss": 1.1324,
+      "step": 7658
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4028668912071483e-05,
+      "loss": 1.0169,
+      "step": 7659
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4007983611665928e-05,
+      "loss": 1.079,
+      "step": 7660
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3987306004007648e-05,
+      "loss": 1.036,
+      "step": 7661
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.396663609118982e-05,
+      "loss": 1.0994,
+      "step": 7662
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.394597387530494e-05,
+      "loss": 1.0702,
+      "step": 7663
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.39253193584446e-05,
+      "loss": 0.9946,
+      "step": 7664
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3904672542699723e-05,
+      "loss": 1.0819,
+      "step": 7665
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3884033430160367e-05,
+      "loss": 1.0554,
+      "step": 7666
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.386340202291589e-05,
+      "loss": 1.0601,
+      "step": 7667
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3842778323054803e-05,
+      "loss": 1.0065,
+      "step": 7668
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.382216233266489e-05,
+      "loss": 1.0758,
+      "step": 7669
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3801554053833108e-05,
+      "loss": 1.0392,
+      "step": 7670
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.378095348864563e-05,
+      "loss": 1.061,
+      "step": 7671
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.376036063918793e-05,
+      "loss": 1.0878,
+      "step": 7672
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3739775507544582e-05,
+      "loss": 1.0536,
+      "step": 7673
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.371919809579949e-05,
+      "loss": 1.0819,
+      "step": 7674
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3698628406035706e-05,
+      "loss": 1.1048,
+      "step": 7675
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.367806644033549e-05,
+      "loss": 1.0024,
+      "step": 7676
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3657512200780373e-05,
+      "loss": 1.0581,
+      "step": 7677
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.363696568945113e-05,
+      "loss": 1.0816,
+      "step": 7678
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.3616426908427614e-05,
+      "loss": 1.1409,
+      "step": 7679
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3595895859789076e-05,
+      "loss": 1.0402,
+      "step": 7680
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3575372545613804e-05,
+      "loss": 1.1153,
+      "step": 7681
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3554856967979488e-05,
+      "loss": 1.0949,
+      "step": 7682
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3534349128962884e-05,
+      "loss": 1.0555,
+      "step": 7683
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.351384903064e-05,
+      "loss": 1.0778,
+      "step": 7684
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3493356675086154e-05,
+      "loss": 1.1124,
+      "step": 7685
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3472872064375717e-05,
+      "loss": 1.1278,
+      "step": 7686
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3452395200582456e-05,
+      "loss": 1.0949,
+      "step": 7687
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3431926085779223e-05,
+      "loss": 1.0915,
+      "step": 7688
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.341146472203809e-05,
+      "loss": 1.0903,
+      "step": 7689
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.339101111143045e-05,
+      "loss": 1.0618,
+      "step": 7690
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3370565256026778e-05,
+      "loss": 1.0685,
+      "step": 7691
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.335012715789686e-05,
+      "loss": 1.0977,
+      "step": 7692
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3329696819109693e-05,
+      "loss": 1.0623,
+      "step": 7693
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3309274241733414e-05,
+      "loss": 1.1265,
+      "step": 7694
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.328885942783545e-05,
+      "loss": 1.1044,
+      "step": 7695
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.326845237948242e-05,
+      "loss": 1.1092,
+      "step": 7696
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3248053098740098e-05,
+      "loss": 1.069,
+      "step": 7697
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3227661587673578e-05,
+      "loss": 1.1228,
+      "step": 7698
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3207277848347076e-05,
+      "loss": 1.0353,
+      "step": 7699
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3186901882824097e-05,
+      "loss": 1.0592,
+      "step": 7700
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.316653369316729e-05,
+      "loss": 1.0595,
+      "step": 7701
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3146173281438533e-05,
+      "loss": 1.0825,
+      "step": 7702
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3125820649698983e-05,
+      "loss": 1.0192,
+      "step": 7703
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.3105475800008903e-05,
+      "loss": 1.0923,
+      "step": 7704
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3085138734427836e-05,
+      "loss": 1.115,
+      "step": 7705
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3064809455014557e-05,
+      "loss": 1.0346,
+      "step": 7706
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.304448796382701e-05,
+      "loss": 1.0383,
+      "step": 7707
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3024174262922304e-05,
+      "loss": 1.0439,
+      "step": 7708
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.3003868354356895e-05,
+      "loss": 1.1314,
+      "step": 7709
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.29835702401863e-05,
+      "loss": 1.0801,
+      "step": 7710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2963279922465375e-05,
+      "loss": 1.0481,
+      "step": 7711
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2942997403248102e-05,
+      "loss": 1.1019,
+      "step": 7712
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.292272268458767e-05,
+      "loss": 0.9958,
+      "step": 7713
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2902455768536578e-05,
+      "loss": 1.1431,
+      "step": 7714
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2882196657146394e-05,
+      "loss": 1.0483,
+      "step": 7715
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.286194535246803e-05,
+      "loss": 1.0991,
+      "step": 7716
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2841701856551522e-05,
+      "loss": 1.0885,
+      "step": 7717
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2821466171446104e-05,
+      "loss": 1.0638,
+      "step": 7718
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.280123829920028e-05,
+      "loss": 1.066,
+      "step": 7719
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2781018241861774e-05,
+      "loss": 1.1332,
+      "step": 7720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.276080600147743e-05,
+      "loss": 1.0925,
+      "step": 7721
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.27406015800934e-05,
+      "loss": 1.1293,
+      "step": 7722
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2720404979754937e-05,
+      "loss": 0.9927,
+      "step": 7723
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.270021620250663e-05,
+      "loss": 1.0061,
+      "step": 7724
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2680035250392184e-05,
+      "loss": 1.0485,
+      "step": 7725
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.26598621254545e-05,
+      "loss": 1.0561,
+      "step": 7726
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.263969682973578e-05,
+      "loss": 1.0774,
+      "step": 7727
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.261953936527732e-05,
+      "loss": 1.111,
+      "step": 7728
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.2599389734119734e-05,
+      "loss": 1.0911,
+      "step": 7729
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.257924793830277e-05,
+      "loss": 1.1094,
+      "step": 7730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2559113979865377e-05,
+      "loss": 1.088,
+      "step": 7731
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.253898786084575e-05,
+      "loss": 1.0897,
+      "step": 7732
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.251886958328131e-05,
+      "loss": 1.0237,
+      "step": 7733
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2498759149208594e-05,
+      "loss": 1.0782,
+      "step": 7734
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.247865656066346e-05,
+      "loss": 1.0575,
+      "step": 7735
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2458561819680856e-05,
+      "loss": 1.108,
+      "step": 7736
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.243847492829504e-05,
+      "loss": 1.047,
+      "step": 7737
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2418395888539413e-05,
+      "loss": 1.0271,
+      "step": 7738
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.239832470244656e-05,
+      "loss": 1.0812,
+      "step": 7739
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.237826137204837e-05,
+      "loss": 1.0361,
+      "step": 7740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2358205899375805e-05,
+      "loss": 1.0596,
+      "step": 7741
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2338158286459178e-05,
+      "loss": 1.0568,
+      "step": 7742
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.231811853532788e-05,
+      "loss": 1.0384,
+      "step": 7743
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.229808664801053e-05,
+      "loss": 1.0814,
+      "step": 7744
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2278062626535047e-05,
+      "loss": 1.0817,
+      "step": 7745
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2258046472928407e-05,
+      "loss": 1.0948,
+      "step": 7746
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2238038189216905e-05,
+      "loss": 1.0836,
+      "step": 7747
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2218037777426027e-05,
+      "loss": 1.1136,
+      "step": 7748
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.219804523958037e-05,
+      "loss": 0.9897,
+      "step": 7749
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2178060577703864e-05,
+      "loss": 1.0846,
+      "step": 7750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.215808379381954e-05,
+      "loss": 1.1226,
+      "step": 7751
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2138114889949646e-05,
+      "loss": 1.0236,
+      "step": 7752
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.2118153868115708e-05,
+      "loss": 1.0494,
+      "step": 7753
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.209820073033837e-05,
+      "loss": 1.0207,
+      "step": 7754
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.2078255478637487e-05,
+      "loss": 1.0196,
+      "step": 7755
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.2058318115032195e-05,
+      "loss": 1.0867,
+      "step": 7756
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.2038388641540697e-05,
+      "loss": 1.1481,
+      "step": 7757
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.2018467060180548e-05,
+      "loss": 1.0645,
+      "step": 7758
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1998553372968388e-05,
+      "loss": 1.1082,
+      "step": 7759
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1978647581920063e-05,
+      "loss": 1.1215,
+      "step": 7760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.195874968905075e-05,
+      "loss": 1.1629,
+      "step": 7761
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1938859696374682e-05,
+      "loss": 1.0918,
+      "step": 7762
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1918977605905322e-05,
+      "loss": 1.0763,
+      "step": 7763
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.18991034196554e-05,
+      "loss": 0.998,
+      "step": 7764
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1879237139636732e-05,
+      "loss": 1.0525,
+      "step": 7765
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1859378767860482e-05,
+      "loss": 1.0801,
+      "step": 7766
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.18395283063369e-05,
+      "loss": 1.0325,
+      "step": 7767
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1819685757075435e-05,
+      "loss": 1.0708,
+      "step": 7768
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1799851122084814e-05,
+      "loss": 1.1398,
+      "step": 7769
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.178002440337288e-05,
+      "loss": 1.0224,
+      "step": 7770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.176020560294675e-05,
+      "loss": 1.0958,
+      "step": 7771
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1740394722812685e-05,
+      "loss": 1.1026,
+      "step": 7772
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1720591764976128e-05,
+      "loss": 1.0572,
+      "step": 7773
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1700796731441776e-05,
+      "loss": 1.0634,
+      "step": 7774
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1681009624213533e-05,
+      "loss": 1.0329,
+      "step": 7775
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.1661230445294413e-05,
+      "loss": 1.0778,
+      "step": 7776
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.164145919668672e-05,
+      "loss": 1.0894,
+      "step": 7777
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.162169588039189e-05,
+      "loss": 1.0824,
+      "step": 7778
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1601940498410622e-05,
+      "loss": 1.0783,
+      "step": 7779
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.158219305274275e-05,
+      "loss": 1.1528,
+      "step": 7780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1562453545387297e-05,
+      "loss": 1.0474,
+      "step": 7781
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1542721978342573e-05,
+      "loss": 1.0258,
+      "step": 7782
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.152299835360596e-05,
+      "loss": 1.1269,
+      "step": 7783
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.150328267317416e-05,
+      "loss": 1.0814,
+      "step": 7784
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1483574939042983e-05,
+      "loss": 1.0638,
+      "step": 7785
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1463875153207435e-05,
+      "loss": 1.0537,
+      "step": 7786
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.144418331766178e-05,
+      "loss": 1.0677,
+      "step": 7787
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.142449943439946e-05,
+      "loss": 1.062,
+      "step": 7788
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1404823505413028e-05,
+      "loss": 1.0121,
+      "step": 7789
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.138515553269439e-05,
+      "loss": 1.0412,
+      "step": 7790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1365495518234456e-05,
+      "loss": 1.078,
+      "step": 7791
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.134584346402352e-05,
+      "loss": 1.0466,
+      "step": 7792
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1326199372050924e-05,
+      "loss": 1.0555,
+      "step": 7793
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1306563244305243e-05,
+      "loss": 1.0561,
+      "step": 7794
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1286935082774316e-05,
+      "loss": 1.0878,
+      "step": 7795
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1267314889445066e-05,
+      "loss": 1.06,
+      "step": 7796
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1247702666303725e-05,
+      "loss": 1.0494,
+      "step": 7797
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1228098415335628e-05,
+      "loss": 1.1284,
+      "step": 7798
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.12085021385253e-05,
+      "loss": 1.0924,
+      "step": 7799
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.118891383785655e-05,
+      "loss": 1.1009,
+      "step": 7800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.116933351531226e-05,
+      "loss": 1.049,
+      "step": 7801
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.1149761172874605e-05,
+      "loss": 1.0913,
+      "step": 7802
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.113019681252493e-05,
+      "loss": 1.051,
+      "step": 7803
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.1110640436243723e-05,
+      "loss": 1.0855,
+      "step": 7804
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.1091092046010684e-05,
+      "loss": 1.1167,
+      "step": 7805
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.1071551643804765e-05,
+      "loss": 1.1012,
+      "step": 7806
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.1052019231604e-05,
+      "loss": 1.1108,
+      "step": 7807
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.1032494811385738e-05,
+      "loss": 1.061,
+      "step": 7808
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.1012978385126413e-05,
+      "loss": 1.0849,
+      "step": 7809
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.099346995480168e-05,
+      "loss": 1.0631,
+      "step": 7810
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.097396952238644e-05,
+      "loss": 1.0945,
+      "step": 7811
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0954477089854696e-05,
+      "loss": 1.1387,
+      "step": 7812
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0934992659179742e-05,
+      "loss": 1.0571,
+      "step": 7813
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0915516232333965e-05,
+      "loss": 1.0661,
+      "step": 7814
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0896047811288943e-05,
+      "loss": 1.0408,
+      "step": 7815
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0876587398015578e-05,
+      "loss": 1.0718,
+      "step": 7816
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.085713499448384e-05,
+      "loss": 0.9968,
+      "step": 7817
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0837690602662852e-05,
+      "loss": 1.0818,
+      "step": 7818
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0818254224521073e-05,
+      "loss": 1.0822,
+      "step": 7819
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0798825862025996e-05,
+      "loss": 1.077,
+      "step": 7820
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0779405517144423e-05,
+      "loss": 1.0877,
+      "step": 7821
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0759993191842285e-05,
+      "loss": 1.1184,
+      "step": 7822
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0740588888084677e-05,
+      "loss": 0.9965,
+      "step": 7823
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.072119260783597e-05,
+      "loss": 1.0502,
+      "step": 7824
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0701804353059605e-05,
+      "loss": 1.0577,
+      "step": 7825
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0682424125718347e-05,
+      "loss": 1.0607,
+      "step": 7826
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.0663051927774037e-05,
+      "loss": 1.0468,
+      "step": 7827
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.06436877611877e-05,
+      "loss": 1.0292,
+      "step": 7828
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.062433162791966e-05,
+      "loss": 1.1003,
+      "step": 7829
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0604983529929335e-05,
+      "loss": 1.0609,
+      "step": 7830
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.058564346917533e-05,
+      "loss": 1.1289,
+      "step": 7831
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0566311447615506e-05,
+      "loss": 1.067,
+      "step": 7832
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0546987467206802e-05,
+      "loss": 0.9968,
+      "step": 7833
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0527671529905456e-05,
+      "loss": 1.1192,
+      "step": 7834
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0508363637666827e-05,
+      "loss": 1.1041,
+      "step": 7835
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.048906379244543e-05,
+      "loss": 1.1296,
+      "step": 7836
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0469771996195076e-05,
+      "loss": 1.0352,
+      "step": 7837
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.045048825086863e-05,
+      "loss": 1.0513,
+      "step": 7838
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0431212558418246e-05,
+      "loss": 1.0917,
+      "step": 7839
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0411944920795223e-05,
+      "loss": 0.9737,
+      "step": 7840
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.039268533994999e-05,
+      "loss": 1.1132,
+      "step": 7841
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0373433817832256e-05,
+      "loss": 1.0656,
+      "step": 7842
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.03541903563909e-05,
+      "loss": 1.0973,
+      "step": 7843
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0334954957573883e-05,
+      "loss": 1.0213,
+      "step": 7844
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0315727623328507e-05,
+      "loss": 1.0389,
+      "step": 7845
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0296508355601118e-05,
+      "loss": 1.113,
+      "step": 7846
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0277297156337293e-05,
+      "loss": 1.0813,
+      "step": 7847
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0258094027481854e-05,
+      "loss": 1.096,
+      "step": 7848
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.023889897097868e-05,
+      "loss": 1.0316,
+      "step": 7849
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0219711988770994e-05,
+      "loss": 1.0577,
+      "step": 7850
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.0200533082801053e-05,
+      "loss": 1.0375,
+      "step": 7851
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.018136225501035e-05,
+      "loss": 1.0388,
+      "step": 7852
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.016219950733961e-05,
+      "loss": 1.0881,
+      "step": 7853
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.014304484172864e-05,
+      "loss": 1.0827,
+      "step": 7854
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0123898260116557e-05,
+      "loss": 1.0722,
+      "step": 7855
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0104759764441528e-05,
+      "loss": 1.0833,
+      "step": 7856
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.008562935664098e-05,
+      "loss": 1.0096,
+      "step": 7857
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0066507038651527e-05,
+      "loss": 1.112,
+      "step": 7858
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0047392812408926e-05,
+      "loss": 1.0278,
+      "step": 7859
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0028286679848106e-05,
+      "loss": 1.0715,
+      "step": 7860
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.0009188642903244e-05,
+      "loss": 1.0742,
+      "step": 7861
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9990098703507587e-05,
+      "loss": 1.1262,
+      "step": 7862
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9971016863593705e-05,
+      "loss": 1.0607,
+      "step": 7863
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9951943125093243e-05,
+      "loss": 1.0796,
+      "step": 7864
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9932877489937007e-05,
+      "loss": 1.052,
+      "step": 7865
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9913819960055102e-05,
+      "loss": 1.1311,
+      "step": 7866
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9894770537376685e-05,
+      "loss": 1.0615,
+      "step": 7867
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9875729223830197e-05,
+      "loss": 1.1028,
+      "step": 7868
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9856696021343192e-05,
+      "loss": 1.0249,
+      "step": 7869
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9837670931842344e-05,
+      "loss": 1.0574,
+      "step": 7870
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9818653957253718e-05,
+      "loss": 1.0769,
+      "step": 7871
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.979964509950234e-05,
+      "loss": 1.0104,
+      "step": 7872
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.978064436051249e-05,
+      "loss": 1.0283,
+      "step": 7873
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9761651742207677e-05,
+      "loss": 1.095,
+      "step": 7874
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9742667246510484e-05,
+      "loss": 0.98,
+      "step": 7875
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9723690875342792e-05,
+      "loss": 1.077,
+      "step": 7876
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.970472263062557e-05,
+      "loss": 1.0616,
+      "step": 7877
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.968576251427896e-05,
+      "loss": 1.093,
+      "step": 7878
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9666810528222367e-05,
+      "loss": 1.0849,
+      "step": 7879
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9647866674374282e-05,
+      "loss": 1.0797,
+      "step": 7880
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9628930954652437e-05,
+      "loss": 1.1473,
+      "step": 7881
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9610003370973707e-05,
+      "loss": 1.0715,
+      "step": 7882
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9591083925254115e-05,
+      "loss": 1.0594,
+      "step": 7883
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9572172619408935e-05,
+      "loss": 1.0515,
+      "step": 7884
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9553269455352573e-05,
+      "loss": 1.0181,
+      "step": 7885
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9534374434998604e-05,
+      "loss": 1.0951,
+      "step": 7886
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9515487560259803e-05,
+      "loss": 1.0555,
+      "step": 7887
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9496608833048112e-05,
+      "loss": 1.081,
+      "step": 7888
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.947773825527459e-05,
+      "loss": 1.0464,
+      "step": 7889
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.94588758288496e-05,
+      "loss": 1.0408,
+      "step": 7890
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9440021555682542e-05,
+      "loss": 1.095,
+      "step": 7891
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9421175437682105e-05,
+      "loss": 1.0637,
+      "step": 7892
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9402337476756072e-05,
+      "loss": 1.0252,
+      "step": 7893
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9383507674811397e-05,
+      "loss": 1.0957,
+      "step": 7894
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.936468603375431e-05,
+      "loss": 1.0776,
+      "step": 7895
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9345872555490076e-05,
+      "loss": 1.0566,
+      "step": 7896
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9327067241923235e-05,
+      "loss": 1.0855,
+      "step": 7897
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9308270094957493e-05,
+      "loss": 1.0717,
+      "step": 7898
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9289481116495657e-05,
+      "loss": 1.09,
+      "step": 7899
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9270700308439792e-05,
+      "loss": 1.0745,
+      "step": 7900
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9251927672691085e-05,
+      "loss": 1.1136,
+      "step": 7901
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.9233163211149874e-05,
+      "loss": 1.0645,
+      "step": 7902
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9214406925715755e-05,
+      "loss": 1.0651,
+      "step": 7903
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.91956588182874e-05,
+      "loss": 1.1004,
+      "step": 7904
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9176918890762753e-05,
+      "loss": 1.0443,
+      "step": 7905
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.915818714503884e-05,
+      "loss": 1.0822,
+      "step": 7906
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9139463583011884e-05,
+      "loss": 1.1036,
+      "step": 7907
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9120748206577322e-05,
+      "loss": 1.098,
+      "step": 7908
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.910204101762969e-05,
+      "loss": 1.0666,
+      "step": 7909
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9083342018062777e-05,
+      "loss": 1.023,
+      "step": 7910
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9064651209769468e-05,
+      "loss": 1.0437,
+      "step": 7911
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9045968594641862e-05,
+      "loss": 1.0318,
+      "step": 7912
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.902729417457124e-05,
+      "loss": 1.0524,
+      "step": 7913
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.9008627951448034e-05,
+      "loss": 1.1314,
+      "step": 7914
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8989969927161788e-05,
+      "loss": 1.0826,
+      "step": 7915
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8971320103601332e-05,
+      "loss": 1.0748,
+      "step": 7916
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.895267848265456e-05,
+      "loss": 1.0651,
+      "step": 7917
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.893404506620864e-05,
+      "loss": 1.0791,
+      "step": 7918
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8915419856149807e-05,
+      "loss": 1.0812,
+      "step": 7919
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8896802854363503e-05,
+      "loss": 1.0622,
+      "step": 7920
+    },
+    {
+      "epoch": 3.17,
+      "eval_loss": 0.8191506266593933,
+      "eval_runtime": 501.8288,
+      "eval_samples_per_second": 51.611,
+      "eval_steps_per_second": 25.806,
+      "step": 7920
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.887819406273439e-05,
+      "loss": 1.0843,
+      "step": 7921
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.88595934831462e-05,
+      "loss": 1.0769,
+      "step": 7922
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8841001117481926e-05,
+      "loss": 1.0051,
+      "step": 7923
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.882241696762369e-05,
+      "loss": 1.0688,
+      "step": 7924
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.880384103545273e-05,
+      "loss": 1.0304,
+      "step": 7925
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8785273322849593e-05,
+      "loss": 1.0561,
+      "step": 7926
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8766713831693872e-05,
+      "loss": 1.0685,
+      "step": 7927
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8748162563864324e-05,
+      "loss": 1.0938,
+      "step": 7928
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8729619521238973e-05,
+      "loss": 1.1206,
+      "step": 7929
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8711084705694892e-05,
+      "loss": 1.0773,
+      "step": 7930
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8692558119108428e-05,
+      "loss": 1.0519,
+      "step": 7931
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8674039763355034e-05,
+      "loss": 1.0765,
+      "step": 7932
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.86555296403093e-05,
+      "loss": 1.0608,
+      "step": 7933
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.863702775184508e-05,
+      "loss": 1.0908,
+      "step": 7934
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8618534099835326e-05,
+      "loss": 1.0403,
+      "step": 7935
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8600048686152127e-05,
+      "loss": 1.0532,
+      "step": 7936
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.858157151266683e-05,
+      "loss": 0.9766,
+      "step": 7937
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.856310258124986e-05,
+      "loss": 1.0825,
+      "step": 7938
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8544641893770854e-05,
+      "loss": 1.0808,
+      "step": 7939
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.852618945209864e-05,
+      "loss": 1.0667,
+      "step": 7940
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.850774525810114e-05,
+      "loss": 1.1246,
+      "step": 7941
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.848930931364551e-05,
+      "loss": 1.0911,
+      "step": 7942
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8470881620598012e-05,
+      "loss": 1.088,
+      "step": 7943
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.845246218082409e-05,
+      "loss": 1.0413,
+      "step": 7944
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8434050996188402e-05,
+      "loss": 1.0571,
+      "step": 7945
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.841564806855468e-05,
+      "loss": 1.1037,
+      "step": 7946
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.839725339978592e-05,
+      "loss": 1.0621,
+      "step": 7947
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8378866991744215e-05,
+      "loss": 1.0692,
+      "step": 7948
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8360488846290813e-05,
+      "loss": 1.1054,
+      "step": 7949
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.834211896528619e-05,
+      "loss": 1.0792,
+      "step": 7950
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.832375735058991e-05,
+      "loss": 0.9942,
+      "step": 7951
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8305404004060768e-05,
+      "loss": 1.0359,
+      "step": 7952
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.82870589275567e-05,
+      "loss": 1.0629,
+      "step": 7953
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8268722122934744e-05,
+      "loss": 1.0161,
+      "step": 7954
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8250393592051217e-05,
+      "loss": 1.106,
+      "step": 7955
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8232073336761513e-05,
+      "loss": 1.079,
+      "step": 7956
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8213761358920177e-05,
+      "loss": 1.0988,
+      "step": 7957
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.819545766038099e-05,
+      "loss": 1.0732,
+      "step": 7958
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8177162242996814e-05,
+      "loss": 1.0426,
+      "step": 7959
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8158875108619756e-05,
+      "loss": 1.1231,
+      "step": 7960
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.814059625910103e-05,
+      "loss": 1.1391,
+      "step": 7961
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8122325696290975e-05,
+      "loss": 1.0189,
+      "step": 7962
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8104063422039208e-05,
+      "loss": 1.0841,
+      "step": 7963
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.808580943819438e-05,
+      "loss": 1.056,
+      "step": 7964
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.80675637466044e-05,
+      "loss": 1.0364,
+      "step": 7965
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.804932634911627e-05,
+      "loss": 1.099,
+      "step": 7966
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.803109724757618e-05,
+      "loss": 1.0719,
+      "step": 7967
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.801287644382952e-05,
+      "loss": 1.0765,
+      "step": 7968
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.799466393972077e-05,
+      "loss": 1.0448,
+      "step": 7969
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7976459737093577e-05,
+      "loss": 1.1007,
+      "step": 7970
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7958263837790824e-05,
+      "loss": 1.0392,
+      "step": 7971
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7940076243654437e-05,
+      "loss": 1.0819,
+      "step": 7972
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.792189695652563e-05,
+      "loss": 1.0953,
+      "step": 7973
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7903725978244678e-05,
+      "loss": 1.0843,
+      "step": 7974
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.788556331065102e-05,
+      "loss": 1.0165,
+      "step": 7975
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7867408955583333e-05,
+      "loss": 1.0392,
+      "step": 7976
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.784926291487935e-05,
+      "loss": 1.0672,
+      "step": 7977
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7831125190376064e-05,
+      "loss": 1.1145,
+      "step": 7978
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7812995783909548e-05,
+      "loss": 1.0743,
+      "step": 7979
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7794874697315046e-05,
+      "loss": 1.0845,
+      "step": 7980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7776761932426976e-05,
+      "loss": 1.1074,
+      "step": 7981
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7758657491078954e-05,
+      "loss": 1.111,
+      "step": 7982
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.774056137510365e-05,
+      "loss": 1.1146,
+      "step": 7983
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.772247358633301e-05,
+      "loss": 0.9865,
+      "step": 7984
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7704394126598056e-05,
+      "loss": 1.0686,
+      "step": 7985
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7686322997728956e-05,
+      "loss": 1.0938,
+      "step": 7986
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.766826020155513e-05,
+      "loss": 1.0431,
+      "step": 7987
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7650205739905024e-05,
+      "loss": 1.0629,
+      "step": 7988
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.763215961460637e-05,
+      "loss": 1.1005,
+      "step": 7989
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7614121827485973e-05,
+      "loss": 1.1352,
+      "step": 7990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7596092380369788e-05,
+      "loss": 1.0882,
+      "step": 7991
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7578071275083007e-05,
+      "loss": 1.1139,
+      "step": 7992
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.756005851344987e-05,
+      "loss": 1.0331,
+      "step": 7993
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7542054097293847e-05,
+      "loss": 1.1336,
+      "step": 7994
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7524058028437585e-05,
+      "loss": 1.078,
+      "step": 7995
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.750607030870277e-05,
+      "loss": 1.0537,
+      "step": 7996
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7488090939910397e-05,
+      "loss": 1.0214,
+      "step": 7997
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7470119923880478e-05,
+      "loss": 1.0828,
+      "step": 7998
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7452157262432235e-05,
+      "loss": 1.1298,
+      "step": 7999
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7434202957384083e-05,
+      "loss": 1.0022,
+      "step": 8000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7416257010553506e-05,
+      "loss": 1.09,
+      "step": 8001
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.739831942375725e-05,
+      "loss": 1.0732,
+      "step": 8002
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7380390198811114e-05,
+      "loss": 1.0907,
+      "step": 8003
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7362469337530084e-05,
+      "loss": 1.0387,
+      "step": 8004
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7344556841728342e-05,
+      "loss": 1.0967,
+      "step": 8005
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.732665271321914e-05,
+      "loss": 1.0642,
+      "step": 8006
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7308756953814964e-05,
+      "loss": 1.0559,
+      "step": 8007
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.729086956532744e-05,
+      "loss": 1.0918,
+      "step": 8008
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.727299054956728e-05,
+      "loss": 1.1022,
+      "step": 8009
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.725511990834443e-05,
+      "loss": 1.0486,
+      "step": 8010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7237257643467942e-05,
+      "loss": 1.1067,
+      "step": 8011
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7219403756746e-05,
+      "loss": 1.0435,
+      "step": 8012
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7201558249986026e-05,
+      "loss": 1.0873,
+      "step": 8013
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.718372112499449e-05,
+      "loss": 1.0633,
+      "step": 8014
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7165892383577097e-05,
+      "loss": 1.0213,
+      "step": 8015
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7148072027538653e-05,
+      "loss": 1.0796,
+      "step": 8016
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.713026005868311e-05,
+      "loss": 1.0716,
+      "step": 8017
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7112456478813633e-05,
+      "loss": 1.0929,
+      "step": 8018
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.709466128973245e-05,
+      "loss": 1.074,
+      "step": 8019
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7076874493241036e-05,
+      "loss": 1.1552,
+      "step": 8020
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7059096091139915e-05,
+      "loss": 1.017,
+      "step": 8021
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.704132608522886e-05,
+      "loss": 1.0694,
+      "step": 8022
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7023564477306697e-05,
+      "loss": 1.1238,
+      "step": 8023
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.70058112691715e-05,
+      "loss": 1.018,
+      "step": 8024
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.698806646262041e-05,
+      "loss": 1.1037,
+      "step": 8025
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6970330059449778e-05,
+      "loss": 1.088,
+      "step": 8026
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.695260206145507e-05,
+      "loss": 1.108,
+      "step": 8027
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6934882470430867e-05,
+      "loss": 1.0981,
+      "step": 8028
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.691717128817101e-05,
+      "loss": 1.0732,
+      "step": 8029
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6899468516468354e-05,
+      "loss": 1.0422,
+      "step": 8030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6881774157115027e-05,
+      "loss": 1.0045,
+      "step": 8031
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.686408821190222e-05,
+      "loss": 1.0972,
+      "step": 8032
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6846410682620274e-05,
+      "loss": 1.0378,
+      "step": 8033
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.682874157105875e-05,
+      "loss": 1.0703,
+      "step": 8034
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6811080879006257e-05,
+      "loss": 1.1042,
+      "step": 8035
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.679342860825064e-05,
+      "loss": 1.0586,
+      "step": 8036
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6775784760578872e-05,
+      "loss": 1.1146,
+      "step": 8037
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6758149337777006e-05,
+      "loss": 1.0423,
+      "step": 8038
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.674052234163035e-05,
+      "loss": 1.0479,
+      "step": 8039
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.672290377392327e-05,
+      "loss": 1.1024,
+      "step": 8040
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.670529363643929e-05,
+      "loss": 1.1184,
+      "step": 8041
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6687691930961158e-05,
+      "loss": 1.0642,
+      "step": 8042
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6670098659270662e-05,
+      "loss": 1.096,
+      "step": 8043
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.665251382314882e-05,
+      "loss": 1.06,
+      "step": 8044
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.663493742437575e-05,
+      "loss": 1.0623,
+      "step": 8045
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6617369464730705e-05,
+      "loss": 1.1225,
+      "step": 8046
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.659980994599215e-05,
+      "loss": 1.1489,
+      "step": 8047
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6582258869937606e-05,
+      "loss": 1.1002,
+      "step": 8048
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.6564716238343814e-05,
+      "loss": 1.1061,
+      "step": 8049
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.654718205298664e-05,
+      "loss": 1.0232,
+      "step": 8050
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6529656315641073e-05,
+      "loss": 1.0394,
+      "step": 8051
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6512139028081274e-05,
+      "loss": 1.0437,
+      "step": 8052
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6494630192080528e-05,
+      "loss": 1.1459,
+      "step": 8053
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.647712980941124e-05,
+      "loss": 1.097,
+      "step": 8054
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6459637881845048e-05,
+      "loss": 1.0989,
+      "step": 8055
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6442154411152633e-05,
+      "loss": 1.0101,
+      "step": 8056
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.642467939910389e-05,
+      "loss": 1.1026,
+      "step": 8057
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6407212847467835e-05,
+      "loss": 1.0553,
+      "step": 8058
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6389754758012577e-05,
+      "loss": 1.0962,
+      "step": 8059
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6372305132505473e-05,
+      "loss": 1.0528,
+      "step": 8060
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.635486397271293e-05,
+      "loss": 1.0855,
+      "step": 8061
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.633743128040054e-05,
+      "loss": 1.0504,
+      "step": 8062
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6320007057333064e-05,
+      "loss": 1.069,
+      "step": 8063
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6302591305274317e-05,
+      "loss": 1.0478,
+      "step": 8064
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6285184025987378e-05,
+      "loss": 1.0175,
+      "step": 8065
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6267785221234366e-05,
+      "loss": 1.0767,
+      "step": 8066
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6250394892776555e-05,
+      "loss": 1.029,
+      "step": 8067
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6233013042374447e-05,
+      "loss": 1.1014,
+      "step": 8068
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6215639671787574e-05,
+      "loss": 1.1128,
+      "step": 8069
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.619827478277466e-05,
+      "loss": 1.021,
+      "step": 8070
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6180918377093612e-05,
+      "loss": 1.143,
+      "step": 8071
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6163570456501377e-05,
+      "loss": 1.1216,
+      "step": 8072
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6146231022754167e-05,
+      "loss": 1.0513,
+      "step": 8073
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6128900077607233e-05,
+      "loss": 1.0238,
+      "step": 8074
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.6111577622814978e-05,
+      "loss": 1.0722,
+      "step": 8075
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.6094263660131016e-05,
+      "loss": 1.0596,
+      "step": 8076
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.6076958191308057e-05,
+      "loss": 1.1191,
+      "step": 8077
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.605966121809791e-05,
+      "loss": 1.0363,
+      "step": 8078
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.604237274225162e-05,
+      "loss": 1.0684,
+      "step": 8079
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.6025092765519255e-05,
+      "loss": 1.0751,
+      "step": 8080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.600782128965015e-05,
+      "loss": 1.093,
+      "step": 8081
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.599055831639268e-05,
+      "loss": 1.0327,
+      "step": 8082
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.597330384749436e-05,
+      "loss": 1.0094,
+      "step": 8083
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5956057884701946e-05,
+      "loss": 1.0702,
+      "step": 8084
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5938820429761192e-05,
+      "loss": 1.0517,
+      "step": 8085
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5921591484417132e-05,
+      "loss": 1.0671,
+      "step": 8086
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5904371050413836e-05,
+      "loss": 1.0426,
+      "step": 8087
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5887159129494523e-05,
+      "loss": 1.0877,
+      "step": 8088
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.586995572340161e-05,
+      "loss": 1.1149,
+      "step": 8089
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.585276083387658e-05,
+      "loss": 1.0904,
+      "step": 8090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.583557446266011e-05,
+      "loss": 1.1024,
+      "step": 8091
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5818396611492014e-05,
+      "loss": 1.1423,
+      "step": 8092
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5801227282111185e-05,
+      "loss": 1.0288,
+      "step": 8093
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5784066476255734e-05,
+      "loss": 1.089,
+      "step": 8094
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.576691419566283e-05,
+      "loss": 1.0281,
+      "step": 8095
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.57497704420688e-05,
+      "loss": 1.087,
+      "step": 8096
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.573263521720918e-05,
+      "loss": 1.0849,
+      "step": 8097
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5715508522818534e-05,
+      "loss": 1.0627,
+      "step": 8098
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5698390360630655e-05,
+      "loss": 1.1189,
+      "step": 8099
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5681280732378402e-05,
+      "loss": 1.1086,
+      "step": 8100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5664179639793798e-05,
+      "loss": 1.0258,
+      "step": 8101
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5647087084608026e-05,
+      "loss": 1.0933,
+      "step": 8102
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5630003068551358e-05,
+      "loss": 1.007,
+      "step": 8103
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5612927593353232e-05,
+      "loss": 1.0144,
+      "step": 8104
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5595860660742255e-05,
+      "loss": 1.0915,
+      "step": 8105
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.557880227244607e-05,
+      "loss": 1.0948,
+      "step": 8106
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5561752430191557e-05,
+      "loss": 1.0796,
+      "step": 8107
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.554471113570468e-05,
+      "loss": 1.0485,
+      "step": 8108
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.552767839071051e-05,
+      "loss": 1.0234,
+      "step": 8109
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.551065419693334e-05,
+      "loss": 1.056,
+      "step": 8110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.54936385560965e-05,
+      "loss": 1.1014,
+      "step": 8111
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5476631469922554e-05,
+      "loss": 1.0991,
+      "step": 8112
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5459632940133118e-05,
+      "loss": 1.0988,
+      "step": 8113
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.544264296844894e-05,
+      "loss": 1.0297,
+      "step": 8114
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.542566155658999e-05,
+      "loss": 1.0026,
+      "step": 8115
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5408688706275266e-05,
+      "loss": 1.084,
+      "step": 8116
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.539172441922293e-05,
+      "loss": 1.0781,
+      "step": 8117
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5374768697150365e-05,
+      "loss": 1.0487,
+      "step": 8118
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5357821541773987e-05,
+      "loss": 1.1311,
+      "step": 8119
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5340882954809333e-05,
+      "loss": 1.0711,
+      "step": 8120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5323952937971177e-05,
+      "loss": 1.112,
+      "step": 8121
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.53070314929733e-05,
+      "loss": 1.1083,
+      "step": 8122
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5290118621528725e-05,
+      "loss": 1.1401,
+      "step": 8123
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.527321432534954e-05,
+      "loss": 1.0385,
+      "step": 8124
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5256318606146958e-05,
+      "loss": 1.0259,
+      "step": 8125
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5239431465631404e-05,
+      "loss": 1.0774,
+      "step": 8126
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5222552905512321e-05,
+      "loss": 1.0726,
+      "step": 8127
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5205682927498399e-05,
+      "loss": 1.0392,
+      "step": 8128
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.518882153329737e-05,
+      "loss": 1.04,
+      "step": 8129
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5171968724616114e-05,
+      "loss": 1.0817,
+      "step": 8130
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5155124503160678e-05,
+      "loss": 1.0833,
+      "step": 8131
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5138288870636241e-05,
+      "loss": 1.0917,
+      "step": 8132
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5121461828747042e-05,
+      "loss": 1.0477,
+      "step": 8133
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.510464337919656e-05,
+      "loss": 1.0212,
+      "step": 8134
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5087833523687278e-05,
+      "loss": 1.1289,
+      "step": 8135
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5071032263920937e-05,
+      "loss": 1.1201,
+      "step": 8136
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5054239601598307e-05,
+      "loss": 1.0949,
+      "step": 8137
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.503745553841931e-05,
+      "loss": 1.1197,
+      "step": 8138
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5020680076083071e-05,
+      "loss": 1.0848,
+      "step": 8139
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5003913216287713e-05,
+      "loss": 1.0344,
+      "step": 8140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4987154960730632e-05,
+      "loss": 1.0705,
+      "step": 8141
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.497040531110826e-05,
+      "loss": 1.1267,
+      "step": 8142
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.495366426911614e-05,
+      "loss": 1.0567,
+      "step": 8143
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4936931836449043e-05,
+      "loss": 1.0788,
+      "step": 8144
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4920208014800752e-05,
+      "loss": 1.0714,
+      "step": 8145
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4903492805864283e-05,
+      "loss": 1.0599,
+      "step": 8146
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4886786211331726e-05,
+      "loss": 1.0008,
+      "step": 8147
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4870088232894286e-05,
+      "loss": 1.0869,
+      "step": 8148
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.485339887224234e-05,
+      "loss": 1.0637,
+      "step": 8149
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4836718131065364e-05,
+      "loss": 1.0348,
+      "step": 8150
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4820046011051936e-05,
+      "loss": 1.0739,
+      "step": 8151
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4803382513889841e-05,
+      "loss": 1.1031,
+      "step": 8152
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4786727641265885e-05,
+      "loss": 1.074,
+      "step": 8153
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4770081394866108e-05,
+      "loss": 1.1358,
+      "step": 8154
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4753443776375608e-05,
+      "loss": 1.0543,
+      "step": 8155
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4736814787478604e-05,
+      "loss": 1.0658,
+      "step": 8156
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4720194429858503e-05,
+      "loss": 1.0984,
+      "step": 8157
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4703582705197782e-05,
+      "loss": 1.0539,
+      "step": 8158
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.468697961517802e-05,
+      "loss": 1.0915,
+      "step": 8159
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.467038516148006e-05,
+      "loss": 1.1072,
+      "step": 8160
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4653799345783714e-05,
+      "loss": 1.0124,
+      "step": 8161
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4637222169767962e-05,
+      "loss": 1.0779,
+      "step": 8162
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4620653635110981e-05,
+      "loss": 1.1139,
+      "step": 8163
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4604093743489977e-05,
+      "loss": 1.0571,
+      "step": 8164
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4587542496581363e-05,
+      "loss": 1.0646,
+      "step": 8165
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.457099989606061e-05,
+      "loss": 1.0328,
+      "step": 8166
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4554465943602325e-05,
+      "loss": 1.0927,
+      "step": 8167
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.453794064088031e-05,
+      "loss": 1.037,
+      "step": 8168
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4521423989567383e-05,
+      "loss": 1.0595,
+      "step": 8169
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4504915991335588e-05,
+      "loss": 1.0487,
+      "step": 8170
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4488416647856018e-05,
+      "loss": 1.0249,
+      "step": 8171
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4471925960798882e-05,
+      "loss": 1.1268,
+      "step": 8172
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4455443931833634e-05,
+      "loss": 1.0028,
+      "step": 8173
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4438970562628728e-05,
+      "loss": 1.0813,
+      "step": 8174
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4422505854851742e-05,
+      "loss": 0.9765,
+      "step": 8175
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4406049810169475e-05,
+      "loss": 1.0606,
+      "step": 8176
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.438960243024774e-05,
+      "loss": 1.0751,
+      "step": 8177
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4373163716751559e-05,
+      "loss": 1.1101,
+      "step": 8178
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4356733671345024e-05,
+      "loss": 1.0833,
+      "step": 8179
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4340312295691338e-05,
+      "loss": 1.0955,
+      "step": 8180
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4323899591452904e-05,
+      "loss": 1.0411,
+      "step": 8181
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.430749556029115e-05,
+      "loss": 1.0601,
+      "step": 8182
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4291100203866715e-05,
+      "loss": 1.0498,
+      "step": 8183
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.427471352383929e-05,
+      "loss": 1.0886,
+      "step": 8184
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4258335521867695e-05,
+      "loss": 1.0598,
+      "step": 8185
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4241966199609925e-05,
+      "loss": 1.1562,
+      "step": 8186
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4225605558723076e-05,
+      "loss": 1.0568,
+      "step": 8187
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4209253600863303e-05,
+      "loss": 1.0935,
+      "step": 8188
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4192910327685981e-05,
+      "loss": 1.064,
+      "step": 8189
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.417657574084551e-05,
+      "loss": 1.1073,
+      "step": 8190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.41602498419955e-05,
+      "loss": 1.0015,
+      "step": 8191
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4143932632788626e-05,
+      "loss": 1.0654,
+      "step": 8192
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4127624114876659e-05,
+      "loss": 1.0597,
+      "step": 8193
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4111324289910578e-05,
+      "loss": 1.0323,
+      "step": 8194
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4095033159540372e-05,
+      "loss": 1.0373,
+      "step": 8195
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4078750725415279e-05,
+      "loss": 1.0488,
+      "step": 8196
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4062476989183537e-05,
+      "loss": 1.066,
+      "step": 8197
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.404621195249255e-05,
+      "loss": 1.0358,
+      "step": 8198
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.4029955616988888e-05,
+      "loss": 1.0806,
+      "step": 8199
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.4013707984318137e-05,
+      "loss": 1.0414,
+      "step": 8200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.399746905612509e-05,
+      "loss": 1.0271,
+      "step": 8201
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3981238834053656e-05,
+      "loss": 1.0989,
+      "step": 8202
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3965017319746809e-05,
+      "loss": 1.0604,
+      "step": 8203
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3948804514846648e-05,
+      "loss": 1.0285,
+      "step": 8204
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3932600420994458e-05,
+      "loss": 1.0653,
+      "step": 8205
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3916405039830548e-05,
+      "loss": 1.1208,
+      "step": 8206
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3900218372994445e-05,
+      "loss": 1.0959,
+      "step": 8207
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.38840404221247e-05,
+      "loss": 1.0781,
+      "step": 8208
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3867871188859027e-05,
+      "loss": 1.0356,
+      "step": 8209
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3851710674834284e-05,
+      "loss": 1.133,
+      "step": 8210
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3835558881686361e-05,
+      "loss": 1.0485,
+      "step": 8211
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3819415811050384e-05,
+      "loss": 1.1124,
+      "step": 8212
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3803281464560502e-05,
+      "loss": 1.0699,
+      "step": 8213
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3787155843849964e-05,
+      "loss": 1.0874,
+      "step": 8214
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3771038950551274e-05,
+      "loss": 1.124,
+      "step": 8215
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3754930786295906e-05,
+      "loss": 1.0324,
+      "step": 8216
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.373883135271451e-05,
+      "loss": 1.1059,
+      "step": 8217
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.372274065143686e-05,
+      "loss": 1.1136,
+      "step": 8218
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3706658684091812e-05,
+      "loss": 1.1088,
+      "step": 8219
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3690585452307403e-05,
+      "loss": 1.1079,
+      "step": 8220
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3674520957710701e-05,
+      "loss": 1.0535,
+      "step": 8221
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3658465201927927e-05,
+      "loss": 1.1002,
+      "step": 8222
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3642418186584461e-05,
+      "loss": 1.1213,
+      "step": 8223
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.362637991330471e-05,
+      "loss": 1.0293,
+      "step": 8224
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3610350383712289e-05,
+      "loss": 1.0358,
+      "step": 8225
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3594329599429867e-05,
+      "loss": 1.109,
+      "step": 8226
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3578317562079202e-05,
+      "loss": 1.0753,
+      "step": 8227
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3562314273281285e-05,
+      "loss": 1.0321,
+      "step": 8228
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3546319734656098e-05,
+      "loss": 1.0827,
+      "step": 8229
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3530333947822782e-05,
+      "loss": 1.0923,
+      "step": 8230
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3514356914399617e-05,
+      "loss": 1.0244,
+      "step": 8231
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3498388636003945e-05,
+      "loss": 1.0735,
+      "step": 8232
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3482429114252292e-05,
+      "loss": 1.1341,
+      "step": 8233
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3466478350760225e-05,
+      "loss": 1.0695,
+      "step": 8234
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.345053634714244e-05,
+      "loss": 1.0429,
+      "step": 8235
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.343460310501281e-05,
+      "loss": 1.1089,
+      "step": 8236
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3418678625984227e-05,
+      "loss": 1.0864,
+      "step": 8237
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3402762911668775e-05,
+      "loss": 1.1151,
+      "step": 8238
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3386855963677603e-05,
+      "loss": 1.016,
+      "step": 8239
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3370957783620974e-05,
+      "loss": 1.0119,
+      "step": 8240
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3355068373108292e-05,
+      "loss": 1.0408,
+      "step": 8241
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3339187733748071e-05,
+      "loss": 0.9945,
+      "step": 8242
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.332331586714789e-05,
+      "loss": 1.0801,
+      "step": 8243
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3307452774914519e-05,
+      "loss": 1.0703,
+      "step": 8244
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3291598458653743e-05,
+      "loss": 1.0975,
+      "step": 8245
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3275752919970552e-05,
+      "loss": 1.1353,
+      "step": 8246
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3259916160469e-05,
+      "loss": 1.0369,
+      "step": 8247
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3244088181752223e-05,
+      "loss": 1.0072,
+      "step": 8248
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3228268985422553e-05,
+      "loss": 1.0506,
+      "step": 8249
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3212458573081343e-05,
+      "loss": 1.1328,
+      "step": 8250
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.31966569463291e-05,
+      "loss": 1.1089,
+      "step": 8251
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3180864106765478e-05,
+      "loss": 1.0418,
+      "step": 8252
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3165080055989143e-05,
+      "loss": 1.0525,
+      "step": 8253
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3149304795597983e-05,
+      "loss": 1.0942,
+      "step": 8254
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.31335383271889e-05,
+      "loss": 1.0636,
+      "step": 8255
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.311778065235797e-05,
+      "loss": 1.0951,
+      "step": 8256
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3102031772700374e-05,
+      "loss": 1.0949,
+      "step": 8257
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.308629168981037e-05,
+      "loss": 1.0586,
+      "step": 8258
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3070560405281329e-05,
+      "loss": 1.0552,
+      "step": 8259
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3054837920705786e-05,
+      "loss": 1.0275,
+      "step": 8260
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3039124237675283e-05,
+      "loss": 1.0858,
+      "step": 8261
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3023419357780586e-05,
+      "loss": 1.0486,
+      "step": 8262
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3007723282611506e-05,
+      "loss": 1.0836,
+      "step": 8263
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2992036013756925e-05,
+      "loss": 1.0756,
+      "step": 8264
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2976357552804952e-05,
+      "loss": 1.0551,
+      "step": 8265
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2960687901342673e-05,
+      "loss": 0.9989,
+      "step": 8266
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2945027060956383e-05,
+      "loss": 1.1551,
+      "step": 8267
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2929375033231429e-05,
+      "loss": 1.0808,
+      "step": 8268
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.291373181975224e-05,
+      "loss": 1.0562,
+      "step": 8269
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2898097422102474e-05,
+      "loss": 1.0717,
+      "step": 8270
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2882471841864786e-05,
+      "loss": 1.0387,
+      "step": 8271
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2866855080620932e-05,
+      "loss": 1.0765,
+      "step": 8272
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.2851247139951873e-05,
+      "loss": 1.0689,
+      "step": 8273
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2835648021437553e-05,
+      "loss": 1.0546,
+      "step": 8274
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2820057726657131e-05,
+      "loss": 1.0682,
+      "step": 8275
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2804476257188824e-05,
+      "loss": 1.0879,
+      "step": 8276
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2788903614609938e-05,
+      "loss": 1.0418,
+      "step": 8277
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.277333980049693e-05,
+      "loss": 1.0988,
+      "step": 8278
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.27577848164253e-05,
+      "loss": 1.0453,
+      "step": 8279
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2742238663969763e-05,
+      "loss": 1.0438,
+      "step": 8280
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2726701344704018e-05,
+      "loss": 1.0725,
+      "step": 8281
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2711172860200904e-05,
+      "loss": 1.018,
+      "step": 8282
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2695653212032454e-05,
+      "loss": 1.0543,
+      "step": 8283
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2680142401769713e-05,
+      "loss": 1.0626,
+      "step": 8284
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.266464043098281e-05,
+      "loss": 1.0966,
+      "step": 8285
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2649147301241082e-05,
+      "loss": 1.0846,
+      "step": 8286
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2633663014112885e-05,
+      "loss": 1.0331,
+      "step": 8287
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2618187571165717e-05,
+      "loss": 1.0432,
+      "step": 8288
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2602720973966186e-05,
+      "loss": 1.0933,
+      "step": 8289
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2587263224079948e-05,
+      "loss": 1.0924,
+      "step": 8290
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2571814323071851e-05,
+      "loss": 1.0851,
+      "step": 8291
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.255637427250579e-05,
+      "loss": 1.0152,
+      "step": 8292
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2540943073944755e-05,
+      "loss": 1.0069,
+      "step": 8293
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2525520728950902e-05,
+      "loss": 1.0414,
+      "step": 8294
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.25101072390854e-05,
+      "loss": 1.1582,
+      "step": 8295
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2494702605908593e-05,
+      "loss": 1.0431,
+      "step": 8296
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2479306830979942e-05,
+      "loss": 1.103,
+      "step": 8297
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2463919915857925e-05,
+      "loss": 1.1398,
+      "step": 8298
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2448541862100217e-05,
+      "loss": 1.0009,
+      "step": 8299
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2433172671263538e-05,
+      "loss": 1.1312,
+      "step": 8300
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2417812344903701e-05,
+      "loss": 1.0877,
+      "step": 8301
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2402460884575684e-05,
+      "loss": 1.0947,
+      "step": 8302
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2387118291833499e-05,
+      "loss": 1.096,
+      "step": 8303
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2371784568230327e-05,
+      "loss": 1.017,
+      "step": 8304
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2356459715318403e-05,
+      "loss": 1.0066,
+      "step": 8305
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2341143734649052e-05,
+      "loss": 1.0883,
+      "step": 8306
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2325836627772758e-05,
+      "loss": 1.1237,
+      "step": 8307
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2310538396239047e-05,
+      "loss": 1.1903,
+      "step": 8308
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2295249041596602e-05,
+      "loss": 1.0796,
+      "step": 8309
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2279968565393152e-05,
+      "loss": 1.1255,
+      "step": 8310
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2264696969175571e-05,
+      "loss": 1.0405,
+      "step": 8311
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2249434254489834e-05,
+      "loss": 1.1001,
+      "step": 8312
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2234180422880981e-05,
+      "loss": 1.1219,
+      "step": 8313
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2218935475893156e-05,
+      "loss": 1.0582,
+      "step": 8314
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2203699415069658e-05,
+      "loss": 1.141,
+      "step": 8315
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2188472241952809e-05,
+      "loss": 1.0361,
+      "step": 8316
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2173253958084107e-05,
+      "loss": 1.0969,
+      "step": 8317
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2158044565004101e-05,
+      "loss": 1.0717,
+      "step": 8318
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2142844064252435e-05,
+      "loss": 1.1562,
+      "step": 8319
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.21276524573679e-05,
+      "loss": 1.0485,
+      "step": 8320
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2112469745888333e-05,
+      "loss": 1.0108,
+      "step": 8321
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2097295931350716e-05,
+      "loss": 1.0914,
+      "step": 8322
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2082131015291098e-05,
+      "loss": 1.1232,
+      "step": 8323
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2066974999244607e-05,
+      "loss": 1.0466,
+      "step": 8324
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2051827884745559e-05,
+      "loss": 1.0608,
+      "step": 8325
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2036689673327296e-05,
+      "loss": 1.0759,
+      "step": 8326
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2021560366522245e-05,
+      "loss": 1.0497,
+      "step": 8327
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2006439965862004e-05,
+      "loss": 1.0368,
+      "step": 8328
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1991328472877173e-05,
+      "loss": 1.0509,
+      "step": 8329
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1976225889097548e-05,
+      "loss": 1.0777,
+      "step": 8330
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1961132216051973e-05,
+      "loss": 1.0365,
+      "step": 8331
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1946047455268362e-05,
+      "loss": 1.0978,
+      "step": 8332
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1930971608273811e-05,
+      "loss": 1.0682,
+      "step": 8333
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1915904676594402e-05,
+      "loss": 1.0439,
+      "step": 8334
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1900846661755438e-05,
+      "loss": 1.0615,
+      "step": 8335
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1885797565281231e-05,
+      "loss": 1.039,
+      "step": 8336
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1870757388695187e-05,
+      "loss": 1.0792,
+      "step": 8337
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1855726133519862e-05,
+      "loss": 1.0516,
+      "step": 8338
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.184070380127692e-05,
+      "loss": 1.0855,
+      "step": 8339
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1825690393487021e-05,
+      "loss": 1.0483,
+      "step": 8340
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1810685911670039e-05,
+      "loss": 1.0932,
+      "step": 8341
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.179569035734488e-05,
+      "loss": 1.0443,
+      "step": 8342
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.178070373202953e-05,
+      "loss": 1.0769,
+      "step": 8343
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1765726037241143e-05,
+      "loss": 1.0457,
+      "step": 8344
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1750757274495882e-05,
+      "loss": 1.0153,
+      "step": 8345
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1735797445309094e-05,
+      "loss": 1.0378,
+      "step": 8346
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.1720846551195165e-05,
+      "loss": 1.0502,
+      "step": 8347
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.170590459366755e-05,
+      "loss": 1.1158,
+      "step": 8348
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1690971574238884e-05,
+      "loss": 1.0313,
+      "step": 8349
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1676047494420827e-05,
+      "loss": 1.0817,
+      "step": 8350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1661132355724158e-05,
+      "loss": 1.0384,
+      "step": 8351
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.164622615965878e-05,
+      "loss": 1.0749,
+      "step": 8352
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.163132890773363e-05,
+      "loss": 1.1213,
+      "step": 8353
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1616440601456802e-05,
+      "loss": 1.1218,
+      "step": 8354
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1601561242335424e-05,
+      "loss": 1.094,
+      "step": 8355
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1586690831875757e-05,
+      "loss": 1.079,
+      "step": 8356
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1571829371583165e-05,
+      "loss": 1.1101,
+      "step": 8357
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1556976862962043e-05,
+      "loss": 1.0574,
+      "step": 8358
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.154213330751599e-05,
+      "loss": 1.0369,
+      "step": 8359
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1527298706747591e-05,
+      "loss": 1.1172,
+      "step": 8360
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1512473062158569e-05,
+      "loss": 1.1158,
+      "step": 8361
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1497656375249754e-05,
+      "loss": 1.1465,
+      "step": 8362
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1482848647521027e-05,
+      "loss": 1.1083,
+      "step": 8363
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1468049880471431e-05,
+      "loss": 1.093,
+      "step": 8364
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1453260075599015e-05,
+      "loss": 1.111,
+      "step": 8365
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.143847923440099e-05,
+      "loss": 1.0912,
+      "step": 8366
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.142370735837366e-05,
+      "loss": 1.1062,
+      "step": 8367
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1408944449012359e-05,
+      "loss": 1.0942,
+      "step": 8368
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.1394190507811542e-05,
+      "loss": 1.13,
+      "step": 8369
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.137944553626481e-05,
+      "loss": 1.0968,
+      "step": 8370
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.136470953586476e-05,
+      "loss": 1.0951,
+      "step": 8371
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1349982508103175e-05,
+      "loss": 1.1328,
+      "step": 8372
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1335264454470874e-05,
+      "loss": 1.0807,
+      "step": 8373
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1320555376457754e-05,
+      "loss": 1.0594,
+      "step": 8374
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1305855275552879e-05,
+      "loss": 1.0194,
+      "step": 8375
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1291164153244294e-05,
+      "loss": 1.055,
+      "step": 8376
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1276482011019262e-05,
+      "loss": 1.0858,
+      "step": 8377
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1261808850364031e-05,
+      "loss": 1.055,
+      "step": 8378
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1247144672763943e-05,
+      "loss": 1.0995,
+      "step": 8379
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1232489479703545e-05,
+      "loss": 1.1124,
+      "step": 8380
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1217843272666373e-05,
+      "loss": 1.0767,
+      "step": 8381
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.120320605313504e-05,
+      "loss": 1.0751,
+      "step": 8382
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.118857782259134e-05,
+      "loss": 1.0293,
+      "step": 8383
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1173958582516076e-05,
+      "loss": 1.0638,
+      "step": 8384
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.115934833438914e-05,
+      "loss": 1.0525,
+      "step": 8385
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1144747079689589e-05,
+      "loss": 1.1567,
+      "step": 8386
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1130154819895489e-05,
+      "loss": 1.0639,
+      "step": 8387
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1115571556484061e-05,
+      "loss": 0.9787,
+      "step": 8388
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.110099729093157e-05,
+      "loss": 1.0314,
+      "step": 8389
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1086432024713355e-05,
+      "loss": 1.1074,
+      "step": 8390
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1071875759303918e-05,
+      "loss": 1.1311,
+      "step": 8391
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1057328496176755e-05,
+      "loss": 1.0888,
+      "step": 8392
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1042790236804535e-05,
+      "loss": 1.1296,
+      "step": 8393
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1028260982658988e-05,
+      "loss": 1.0197,
+      "step": 8394
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.1013740735210899e-05,
+      "loss": 1.1207,
+      "step": 8395
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.0999229495930185e-05,
+      "loss": 1.101,
+      "step": 8396
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0984727266285833e-05,
+      "loss": 1.0283,
+      "step": 8397
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0970234047745887e-05,
+      "loss": 1.0839,
+      "step": 8398
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0955749841777563e-05,
+      "loss": 1.1321,
+      "step": 8399
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0941274649847055e-05,
+      "loss": 1.0556,
+      "step": 8400
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0926808473419747e-05,
+      "loss": 1.0737,
+      "step": 8401
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0912351313960057e-05,
+      "loss": 1.1061,
+      "step": 8402
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0897903172931455e-05,
+      "loss": 1.113,
+      "step": 8403
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0883464051796588e-05,
+      "loss": 1.0518,
+      "step": 8404
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0869033952017115e-05,
+      "loss": 1.0651,
+      "step": 8405
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0854612875053826e-05,
+      "loss": 1.1135,
+      "step": 8406
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0840200822366585e-05,
+      "loss": 1.0618,
+      "step": 8407
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0825797795414316e-05,
+      "loss": 1.1025,
+      "step": 8408
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0811403795655084e-05,
+      "loss": 1.0035,
+      "step": 8409
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.079701882454599e-05,
+      "loss": 1.0601,
+      "step": 8410
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0782642883543215e-05,
+      "loss": 1.0147,
+      "step": 8411
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0768275974102094e-05,
+      "loss": 1.0714,
+      "step": 8412
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0753918097676963e-05,
+      "loss": 1.1678,
+      "step": 8413
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0739569255721315e-05,
+      "loss": 1.0807,
+      "step": 8414
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.07252294496877e-05,
+      "loss": 1.1111,
+      "step": 8415
+    },
+    {
+      "epoch": 3.37,
+      "eval_loss": 0.8187162280082703,
+      "eval_runtime": 500.7883,
+      "eval_samples_per_second": 51.718,
+      "eval_steps_per_second": 25.859,
+      "step": 8415
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0710898681027704e-05,
+      "loss": 1.0501,
+      "step": 8416
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0696576951192106e-05,
+      "loss": 1.0435,
+      "step": 8417
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0682264261630658e-05,
+      "loss": 1.0788,
+      "step": 8418
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0667960613792294e-05,
+      "loss": 1.0965,
+      "step": 8419
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.065366600912493e-05,
+      "loss": 1.1241,
+      "step": 8420
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.0639380449075664e-05,
+      "loss": 1.1434,
+      "step": 8421
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0625103935090653e-05,
+      "loss": 1.0515,
+      "step": 8422
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0610836468615093e-05,
+      "loss": 1.1433,
+      "step": 8423
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0596578051093286e-05,
+      "loss": 1.1161,
+      "step": 8424
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0582328683968656e-05,
+      "loss": 1.0549,
+      "step": 8425
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0568088368683659e-05,
+      "loss": 1.0636,
+      "step": 8426
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0553857106679844e-05,
+      "loss": 1.1007,
+      "step": 8427
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.05396348993979e-05,
+      "loss": 1.0621,
+      "step": 8428
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0525421748277498e-05,
+      "loss": 1.0775,
+      "step": 8429
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.05112176547575e-05,
+      "loss": 1.0852,
+      "step": 8430
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0497022620275788e-05,
+      "loss": 1.096,
+      "step": 8431
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.04828366462693e-05,
+      "loss": 1.0442,
+      "step": 8432
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0468659734174159e-05,
+      "loss": 1.1057,
+      "step": 8433
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0454491885425454e-05,
+      "loss": 1.0536,
+      "step": 8434
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0440333101457423e-05,
+      "loss": 1.0055,
+      "step": 8435
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0426183383703413e-05,
+      "loss": 1.0486,
+      "step": 8436
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0412042733595762e-05,
+      "loss": 1.0464,
+      "step": 8437
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0397911152565986e-05,
+      "loss": 1.077,
+      "step": 8438
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0383788642044622e-05,
+      "loss": 1.0476,
+      "step": 8439
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0369675203461281e-05,
+      "loss": 1.126,
+      "step": 8440
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.035557083824471e-05,
+      "loss": 1.1243,
+      "step": 8441
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0341475547822688e-05,
+      "loss": 1.1391,
+      "step": 8442
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0327389333622128e-05,
+      "loss": 1.0376,
+      "step": 8443
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0313312197068959e-05,
+      "loss": 1.0712,
+      "step": 8444
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0299244139588226e-05,
+      "loss": 1.0573,
+      "step": 8445
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.0285185162604071e-05,
+      "loss": 1.0546,
+      "step": 8446
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0271135267539678e-05,
+      "loss": 1.0198,
+      "step": 8447
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0257094455817329e-05,
+      "loss": 1.1419,
+      "step": 8448
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0243062728858432e-05,
+      "loss": 1.0492,
+      "step": 8449
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0229040088083376e-05,
+      "loss": 1.1234,
+      "step": 8450
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.021502653491173e-05,
+      "loss": 1.074,
+      "step": 8451
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0201022070762089e-05,
+      "loss": 1.0505,
+      "step": 8452
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0187026697052115e-05,
+      "loss": 1.1059,
+      "step": 8453
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0173040415198608e-05,
+      "loss": 1.0996,
+      "step": 8454
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0159063226617383e-05,
+      "loss": 1.0889,
+      "step": 8455
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0145095132723392e-05,
+      "loss": 1.1018,
+      "step": 8456
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.013113613493062e-05,
+      "loss": 1.0439,
+      "step": 8457
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0117186234652144e-05,
+      "loss": 1.0665,
+      "step": 8458
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0103245433300156e-05,
+      "loss": 1.0643,
+      "step": 8459
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0089313732285854e-05,
+      "loss": 1.0223,
+      "step": 8460
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0075391133019596e-05,
+      "loss": 1.0467,
+      "step": 8461
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0061477636910777e-05,
+      "loss": 1.0522,
+      "step": 8462
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0047573245367846e-05,
+      "loss": 1.1052,
+      "step": 8463
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0033677959798393e-05,
+      "loss": 1.078,
+      "step": 8464
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0019791781609045e-05,
+      "loss": 1.0008,
+      "step": 8465
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.0005914712205477e-05,
+      "loss": 1.0832,
+      "step": 8466
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.992046752992523e-06,
+      "loss": 1.0465,
+      "step": 8467
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.978187905374025e-06,
+      "loss": 1.0939,
+      "step": 8468
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.964338170752941e-06,
+      "loss": 1.0221,
+      "step": 8469
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.950497550531301e-06,
+      "loss": 1.1426,
+      "step": 8470
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.936666046110166e-06,
+      "loss": 1.0742,
+      "step": 8471
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.922843658889757e-06,
+      "loss": 1.0821,
+      "step": 8472
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.909030390269302e-06,
+      "loss": 1.1094,
+      "step": 8473
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.895226241647137e-06,
+      "loss": 1.0615,
+      "step": 8474
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.881431214420666e-06,
+      "loss": 1.1015,
+      "step": 8475
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.867645309986395e-06,
+      "loss": 1.0756,
+      "step": 8476
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.853868529739851e-06,
+      "loss": 1.0918,
+      "step": 8477
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.840100875075709e-06,
+      "loss": 1.1088,
+      "step": 8478
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.826342347387652e-06,
+      "loss": 1.112,
+      "step": 8479
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.812592948068489e-06,
+      "loss": 0.9937,
+      "step": 8480
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.798852678510073e-06,
+      "loss": 1.0372,
+      "step": 8481
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.785121540103337e-06,
+      "loss": 1.0361,
+      "step": 8482
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.771399534238335e-06,
+      "loss": 1.05,
+      "step": 8483
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.757686662304111e-06,
+      "loss": 1.015,
+      "step": 8484
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.743982925688866e-06,
+      "loss": 1.0946,
+      "step": 8485
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.730288325779835e-06,
+      "loss": 1.0789,
+      "step": 8486
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.71660286396332e-06,
+      "loss": 1.0626,
+      "step": 8487
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.702926541624746e-06,
+      "loss": 1.1233,
+      "step": 8488
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.68925936014854e-06,
+      "loss": 1.0859,
+      "step": 8489
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.67560132091826e-06,
+      "loss": 1.052,
+      "step": 8490
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.661952425316556e-06,
+      "loss": 1.052,
+      "step": 8491
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.648312674725068e-06,
+      "loss": 1.0584,
+      "step": 8492
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.634682070524614e-06,
+      "loss": 1.0531,
+      "step": 8493
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.621060614095001e-06,
+      "loss": 1.057,
+      "step": 8494
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.607448306815125e-06,
+      "loss": 1.0505,
+      "step": 8495
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.59384515006302e-06,
+      "loss": 1.0876,
+      "step": 8496
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.580251145215703e-06,
+      "loss": 1.0218,
+      "step": 8497
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.56666629364934e-06,
+      "loss": 1.0926,
+      "step": 8498
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.553090596739122e-06,
+      "loss": 1.0656,
+      "step": 8499
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.539524055859327e-06,
+      "loss": 1.0914,
+      "step": 8500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.525966672383336e-06,
+      "loss": 1.0759,
+      "step": 8501
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.512418447683535e-06,
+      "loss": 1.063,
+      "step": 8502
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.498879383131455e-06,
+      "loss": 1.0781,
+      "step": 8503
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.485349480097671e-06,
+      "loss": 1.1017,
+      "step": 8504
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.471828739951804e-06,
+      "loss": 1.0487,
+      "step": 8505
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.458317164062613e-06,
+      "loss": 1.0944,
+      "step": 8506
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.444814753797849e-06,
+      "loss": 1.0942,
+      "step": 8507
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.431321510524372e-06,
+      "loss": 1.0157,
+      "step": 8508
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.417837435608156e-06,
+      "loss": 1.1031,
+      "step": 8509
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.404362530414156e-06,
+      "loss": 1.0876,
+      "step": 8510
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.390896796306503e-06,
+      "loss": 1.0625,
+      "step": 8511
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.377440234648317e-06,
+      "loss": 1.0686,
+      "step": 8512
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.363992846801806e-06,
+      "loss": 1.1874,
+      "step": 8513
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.350554634128306e-06,
+      "loss": 1.0988,
+      "step": 8514
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.337125597988116e-06,
+      "loss": 1.1015,
+      "step": 8515
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.32370573974073e-06,
+      "loss": 1.0966,
+      "step": 8516
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.310295060744633e-06,
+      "loss": 1.0548,
+      "step": 8517
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.296893562357412e-06,
+      "loss": 1.0746,
+      "step": 8518
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.283501245935677e-06,
+      "loss": 1.1175,
+      "step": 8519
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.27011811283519e-06,
+      "loss": 1.0712,
+      "step": 8520
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.256744164410702e-06,
+      "loss": 1.0524,
+      "step": 8521
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.243379402016105e-06,
+      "loss": 1.096,
+      "step": 8522
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.230023827004319e-06,
+      "loss": 1.0061,
+      "step": 8523
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.216677440727305e-06,
+      "loss": 1.0533,
+      "step": 8524
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.203340244536186e-06,
+      "loss": 1.0596,
+      "step": 8525
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.190012239781054e-06,
+      "loss": 1.0844,
+      "step": 8526
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.176693427811157e-06,
+      "loss": 1.025,
+      "step": 8527
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.163383809974745e-06,
+      "loss": 1.0274,
+      "step": 8528
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.150083387619157e-06,
+      "loss": 1.0926,
+      "step": 8529
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.136792162090812e-06,
+      "loss": 1.0629,
+      "step": 8530
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.123510134735224e-06,
+      "loss": 1.0608,
+      "step": 8531
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.110237306896907e-06,
+      "loss": 1.0632,
+      "step": 8532
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.096973679919519e-06,
+      "loss": 1.0185,
+      "step": 8533
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.083719255145707e-06,
+      "loss": 1.0464,
+      "step": 8534
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.070474033917276e-06,
+      "loss": 0.9957,
+      "step": 8535
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.057238017575032e-06,
+      "loss": 1.0431,
+      "step": 8536
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.044011207458847e-06,
+      "loss": 1.0983,
+      "step": 8537
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.030793604907716e-06,
+      "loss": 1.124,
+      "step": 8538
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.01758521125965e-06,
+      "loss": 1.0552,
+      "step": 8539
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.004386027851775e-06,
+      "loss": 1.1706,
+      "step": 8540
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.991196056020235e-06,
+      "loss": 1.0861,
+      "step": 8541
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.978015297100263e-06,
+      "loss": 1.0846,
+      "step": 8542
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.964843752426166e-06,
+      "loss": 1.04,
+      "step": 8543
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.951681423331304e-06,
+      "loss": 1.0579,
+      "step": 8544
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.93852831114812e-06,
+      "loss": 1.0614,
+      "step": 8545
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.925384417208139e-06,
+      "loss": 1.0508,
+      "step": 8546
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.912249742841882e-06,
+      "loss": 1.0564,
+      "step": 8547
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.899124289379035e-06,
+      "loss": 1.1226,
+      "step": 8548
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.886008058148276e-06,
+      "loss": 1.087,
+      "step": 8549
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.872901050477367e-06,
+      "loss": 1.0688,
+      "step": 8550
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.859803267693156e-06,
+      "loss": 1.0545,
+      "step": 8551
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.84671471112154e-06,
+      "loss": 1.0712,
+      "step": 8552
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.833635382087491e-06,
+      "loss": 1.078,
+      "step": 8553
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.820565281915039e-06,
+      "loss": 1.0242,
+      "step": 8554
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.80750441192727e-06,
+      "loss": 1.0824,
+      "step": 8555
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.794452773446383e-06,
+      "loss": 1.0536,
+      "step": 8556
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.781410367793563e-06,
+      "loss": 1.0272,
+      "step": 8557
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.768377196289134e-06,
+      "loss": 1.0419,
+      "step": 8558
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.755353260252475e-06,
+      "loss": 1.1332,
+      "step": 8559
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.742338561001984e-06,
+      "loss": 1.0943,
+      "step": 8560
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.729333099855142e-06,
+      "loss": 1.1331,
+      "step": 8561
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.716336878128528e-06,
+      "loss": 1.1385,
+      "step": 8562
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.703349897137747e-06,
+      "loss": 1.1089,
+      "step": 8563
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.690372158197514e-06,
+      "loss": 1.0519,
+      "step": 8564
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.677403662621553e-06,
+      "loss": 1.1153,
+      "step": 8565
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.664444411722661e-06,
+      "loss": 1.037,
+      "step": 8566
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.651494406812744e-06,
+      "loss": 1.1144,
+      "step": 8567
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.63855364920273e-06,
+      "loss": 1.0271,
+      "step": 8568
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.62562214020265e-06,
+      "loss": 1.0496,
+      "step": 8569
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.612699881121545e-06,
+      "loss": 1.1106,
+      "step": 8570
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.599786873267523e-06,
+      "loss": 1.0734,
+      "step": 8571
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.586883117947853e-06,
+      "loss": 1.1021,
+      "step": 8572
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.573988616468765e-06,
+      "loss": 1.0471,
+      "step": 8573
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.561103370135548e-06,
+      "loss": 1.0486,
+      "step": 8574
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.548227380252627e-06,
+      "loss": 1.0332,
+      "step": 8575
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.535360648123425e-06,
+      "loss": 1.0909,
+      "step": 8576
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.522503175050489e-06,
+      "loss": 1.1211,
+      "step": 8577
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.509654962335367e-06,
+      "loss": 1.0701,
+      "step": 8578
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.496816011278686e-06,
+      "loss": 1.062,
+      "step": 8579
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.483986323180172e-06,
+      "loss": 1.0965,
+      "step": 8580
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.471165899338562e-06,
+      "loss": 1.0846,
+      "step": 8581
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.458354741051722e-06,
+      "loss": 1.0358,
+      "step": 8582
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.4455528496165e-06,
+      "loss": 1.0607,
+      "step": 8583
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.432760226328839e-06,
+      "loss": 1.0755,
+      "step": 8584
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.41997687248377e-06,
+      "loss": 1.0543,
+      "step": 8585
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.407202789375367e-06,
+      "loss": 1.0705,
+      "step": 8586
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.394437978296754e-06,
+      "loss": 1.1079,
+      "step": 8587
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.381682440540129e-06,
+      "loss": 0.9867,
+      "step": 8588
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.368936177396735e-06,
+      "loss": 1.0987,
+      "step": 8589
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.356199190156922e-06,
+      "loss": 1.1468,
+      "step": 8590
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.343471480110043e-06,
+      "loss": 1.0851,
+      "step": 8591
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.330753048544515e-06,
+      "loss": 1.0701,
+      "step": 8592
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.318043896747884e-06,
+      "loss": 1.0887,
+      "step": 8593
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.305344026006668e-06,
+      "loss": 1.1041,
+      "step": 8594
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.292653437606524e-06,
+      "loss": 1.0849,
+      "step": 8595
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.279972132832126e-06,
+      "loss": 1.0257,
+      "step": 8596
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.26730011296718e-06,
+      "loss": 1.1126,
+      "step": 8597
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.254637379294527e-06,
+      "loss": 1.0458,
+      "step": 8598
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.241983933095998e-06,
+      "loss": 1.07,
+      "step": 8599
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.229339775652533e-06,
+      "loss": 1.0249,
+      "step": 8600
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.216704908244134e-06,
+      "loss": 1.0679,
+      "step": 8601
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.204079332149783e-06,
+      "loss": 1.116,
+      "step": 8602
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.191463048647641e-06,
+      "loss": 1.0524,
+      "step": 8603
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.178856059014827e-06,
+      "loss": 1.0681,
+      "step": 8604
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.166258364527568e-06,
+      "loss": 1.0981,
+      "step": 8605
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.153669966461152e-06,
+      "loss": 1.0538,
+      "step": 8606
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.141090866089906e-06,
+      "loss": 1.1042,
+      "step": 8607
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.128521064687211e-06,
+      "loss": 1.0258,
+      "step": 8608
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.11596056352556e-06,
+      "loss": 1.1391,
+      "step": 8609
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.103409363876414e-06,
+      "loss": 1.0706,
+      "step": 8610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.090867467010388e-06,
+      "loss": 1.0506,
+      "step": 8611
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.07833487419708e-06,
+      "loss": 1.069,
+      "step": 8612
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.065811586705186e-06,
+      "loss": 1.0655,
+      "step": 8613
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.053297605802468e-06,
+      "loss": 1.127,
+      "step": 8614
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.040792932755715e-06,
+      "loss": 1.1187,
+      "step": 8615
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.028297568830779e-06,
+      "loss": 1.0573,
+      "step": 8616
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.015811515292592e-06,
+      "loss": 1.086,
+      "step": 8617
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.003334773405113e-06,
+      "loss": 1.087,
+      "step": 8618
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.990867344431408e-06,
+      "loss": 1.0654,
+      "step": 8619
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.978409229633532e-06,
+      "loss": 1.0118,
+      "step": 8620
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.965960430272634e-06,
+      "loss": 1.0842,
+      "step": 8621
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.95352094760895e-06,
+      "loss": 1.0847,
+      "step": 8622
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.941090782901706e-06,
+      "loss": 1.0848,
+      "step": 8623
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.92866993740924e-06,
+      "loss": 1.1142,
+      "step": 8624
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.916258412388933e-06,
+      "loss": 1.0701,
+      "step": 8625
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.90385620909716e-06,
+      "loss": 1.051,
+      "step": 8626
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.891463328789495e-06,
+      "loss": 1.0587,
+      "step": 8627
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.879079772720444e-06,
+      "loss": 1.0669,
+      "step": 8628
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.86670554214357e-06,
+      "loss": 1.059,
+      "step": 8629
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.854340638311586e-06,
+      "loss": 1.0916,
+      "step": 8630
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.841985062476165e-06,
+      "loss": 1.0782,
+      "step": 8631
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.829638815888096e-06,
+      "loss": 1.0308,
+      "step": 8632
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.817301899797203e-06,
+      "loss": 1.0703,
+      "step": 8633
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.80497431545233e-06,
+      "loss": 1.1228,
+      "step": 8634
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.792656064101445e-06,
+      "loss": 1.0839,
+      "step": 8635
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.780347146991519e-06,
+      "loss": 1.0991,
+      "step": 8636
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.76804756536862e-06,
+      "loss": 1.052,
+      "step": 8637
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.75575732047783e-06,
+      "loss": 1.0529,
+      "step": 8638
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.743476413563288e-06,
+      "loss": 1.1092,
+      "step": 8639
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.731204845868212e-06,
+      "loss": 1.0755,
+      "step": 8640
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.718942618634895e-06,
+      "loss": 1.0913,
+      "step": 8641
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.70668973310461e-06,
+      "loss": 1.0306,
+      "step": 8642
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.694446190517757e-06,
+      "loss": 1.1135,
+      "step": 8643
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.682211992113752e-06,
+      "loss": 1.0624,
+      "step": 8644
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.669987139131085e-06,
+      "loss": 1.1029,
+      "step": 8645
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.657771632807275e-06,
+      "loss": 1.0449,
+      "step": 8646
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.645565474378913e-06,
+      "loss": 1.0648,
+      "step": 8647
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.633368665081652e-06,
+      "loss": 1.1706,
+      "step": 8648
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.6211812061501745e-06,
+      "loss": 1.0601,
+      "step": 8649
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.609003098818235e-06,
+      "loss": 1.0684,
+      "step": 8650
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.596834344318649e-06,
+      "loss": 1.0459,
+      "step": 8651
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.58467494388323e-06,
+      "loss": 1.0539,
+      "step": 8652
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.572524898742928e-06,
+      "loss": 1.1372,
+      "step": 8653
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.56038421012768e-06,
+      "loss": 1.0363,
+      "step": 8654
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.548252879266504e-06,
+      "loss": 1.0241,
+      "step": 8655
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.536130907387484e-06,
+      "loss": 1.0316,
+      "step": 8656
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.524018295717739e-06,
+      "loss": 1.0739,
+      "step": 8657
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.511915045483409e-06,
+      "loss": 1.0331,
+      "step": 8658
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.4998211579097474e-06,
+      "loss": 1.0986,
+      "step": 8659
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.4877366342210095e-06,
+      "loss": 1.0538,
+      "step": 8660
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.4756614756405496e-06,
+      "loss": 1.0699,
+      "step": 8661
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.463595683390734e-06,
+      "loss": 1.1234,
+      "step": 8662
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.451539258692974e-06,
+      "loss": 1.0775,
+      "step": 8663
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.439492202767795e-06,
+      "loss": 1.0925,
+      "step": 8664
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.427454516834698e-06,
+      "loss": 1.061,
+      "step": 8665
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.415426202112296e-06,
+      "loss": 1.0836,
+      "step": 8666
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.403407259818196e-06,
+      "loss": 1.0406,
+      "step": 8667
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.391397691169099e-06,
+      "loss": 1.0952,
+      "step": 8668
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.37939749738078e-06,
+      "loss": 1.06,
+      "step": 8669
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.367406679667998e-06,
+      "loss": 1.0315,
+      "step": 8670
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.355425239244585e-06,
+      "loss": 1.1139,
+      "step": 8671
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.3434531773234674e-06,
+      "loss": 1.057,
+      "step": 8672
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.331490495116544e-06,
+      "loss": 1.1069,
+      "step": 8673
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.319537193834858e-06,
+      "loss": 1.0734,
+      "step": 8674
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.3075932746884405e-06,
+      "loss": 1.0123,
+      "step": 8675
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.295658738886347e-06,
+      "loss": 1.0693,
+      "step": 8676
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.283733587636776e-06,
+      "loss": 1.0909,
+      "step": 8677
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2718178221468866e-06,
+      "loss": 1.0619,
+      "step": 8678
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.259911443622947e-06,
+      "loss": 1.1032,
+      "step": 8679
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.248014453270246e-06,
+      "loss": 1.0286,
+      "step": 8680
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.236126852293079e-06,
+      "loss": 1.0566,
+      "step": 8681
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.224248641894926e-06,
+      "loss": 1.1376,
+      "step": 8682
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.212379823278193e-06,
+      "loss": 1.0729,
+      "step": 8683
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.200520397644339e-06,
+      "loss": 1.0738,
+      "step": 8684
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.18867036619395e-06,
+      "loss": 1.1015,
+      "step": 8685
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.176829730126589e-06,
+      "loss": 1.0959,
+      "step": 8686
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.164998490640917e-06,
+      "loss": 1.0878,
+      "step": 8687
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.153176648934612e-06,
+      "loss": 1.0962,
+      "step": 8688
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.141364206204393e-06,
+      "loss": 1.0945,
+      "step": 8689
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.129561163646081e-06,
+      "loss": 1.0346,
+      "step": 8690
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.117767522454477e-06,
+      "loss": 1.0242,
+      "step": 8691
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.105983283823492e-06,
+      "loss": 1.1346,
+      "step": 8692
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.094208448946038e-06,
+      "loss": 1.0746,
+      "step": 8693
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.082443019014073e-06,
+      "loss": 1.0456,
+      "step": 8694
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.070686995218656e-06,
+      "loss": 1.0338,
+      "step": 8695
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.058940378749867e-06,
+      "loss": 1.0535,
+      "step": 8696
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.04720317079679e-06,
+      "loss": 1.0876,
+      "step": 8697
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.03547537254764e-06,
+      "loss": 1.0571,
+      "step": 8698
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.023756985189611e-06,
+      "loss": 1.1264,
+      "step": 8699
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.012048009908945e-06,
+      "loss": 1.0692,
+      "step": 8700
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.000348447890992e-06,
+      "loss": 1.0661,
+      "step": 8701
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.988658300320072e-06,
+      "loss": 1.0813,
+      "step": 8702
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.976977568379628e-06,
+      "loss": 1.0343,
+      "step": 8703
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.965306253252091e-06,
+      "loss": 1.0229,
+      "step": 8704
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.95364435611896e-06,
+      "loss": 1.0878,
+      "step": 8705
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.941991878160792e-06,
+      "loss": 1.021,
+      "step": 8706
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.930348820557153e-06,
+      "loss": 1.038,
+      "step": 8707
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.918715184486712e-06,
+      "loss": 1.0325,
+      "step": 8708
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.9070909711271285e-06,
+      "loss": 1.0591,
+      "step": 8709
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.895476181655147e-06,
+      "loss": 1.1043,
+      "step": 8710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.883870817246563e-06,
+      "loss": 1.0475,
+      "step": 8711
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.872274879076168e-06,
+      "loss": 1.0809,
+      "step": 8712
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.860688368317825e-06,
+      "loss": 1.1584,
+      "step": 8713
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.8491112861444825e-06,
+      "loss": 1.063,
+      "step": 8714
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.837543633728072e-06,
+      "loss": 1.0584,
+      "step": 8715
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.825985412239622e-06,
+      "loss": 1.0564,
+      "step": 8716
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.814436622849163e-06,
+      "loss": 1.0982,
+      "step": 8717
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.8028972667257825e-06,
+      "loss": 1.1582,
+      "step": 8718
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.791367345037647e-06,
+      "loss": 1.13,
+      "step": 8719
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.779846858951921e-06,
+      "loss": 1.0714,
+      "step": 8720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.768335809634852e-06,
+      "loss": 1.0944,
+      "step": 8721
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.756834198251693e-06,
+      "loss": 1.0277,
+      "step": 8722
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.745342025966772e-06,
+      "loss": 1.1059,
+      "step": 8723
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.733859293943478e-06,
+      "loss": 1.0513,
+      "step": 8724
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.722386003344206e-06,
+      "loss": 1.0582,
+      "step": 8725
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.710922155330379e-06,
+      "loss": 1.0816,
+      "step": 8726
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.699467751062527e-06,
+      "loss": 1.0649,
+      "step": 8727
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.688022791700166e-06,
+      "loss": 1.0863,
+      "step": 8728
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.676587278401913e-06,
+      "loss": 1.0392,
+      "step": 8729
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.665161212325377e-06,
+      "loss": 1.0507,
+      "step": 8730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.65374459462722e-06,
+      "loss": 1.0165,
+      "step": 8731
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.6423374264631745e-06,
+      "loss": 1.045,
+      "step": 8732
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.630939708987982e-06,
+      "loss": 1.1194,
+      "step": 8733
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.619551443355476e-06,
+      "loss": 1.0994,
+      "step": 8734
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.6081726307184785e-06,
+      "loss": 1.1126,
+      "step": 8735
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.596803272228846e-06,
+      "loss": 0.9877,
+      "step": 8736
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.585443369037592e-06,
+      "loss": 1.0702,
+      "step": 8737
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.5740929222946275e-06,
+      "loss": 1.133,
+      "step": 8738
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.5627519331489805e-06,
+      "loss": 1.0963,
+      "step": 8739
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.551420402748731e-06,
+      "loss": 1.105,
+      "step": 8740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.540098332240963e-06,
+      "loss": 1.1119,
+      "step": 8741
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.528785722771802e-06,
+      "loss": 1.0903,
+      "step": 8742
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.517482575486489e-06,
+      "loss": 1.0927,
+      "step": 8743
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.506188891529197e-06,
+      "loss": 1.0969,
+      "step": 8744
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.494904672043234e-06,
+      "loss": 1.0568,
+      "step": 8745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.483629918170897e-06,
+      "loss": 1.0633,
+      "step": 8746
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.472364631053529e-06,
+      "loss": 1.0725,
+      "step": 8747
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.4611088118315625e-06,
+      "loss": 1.0044,
+      "step": 8748
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.4498624616443955e-06,
+      "loss": 1.0696,
+      "step": 8749
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.4386255816305195e-06,
+      "loss": 1.1076,
+      "step": 8750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.427398172927479e-06,
+      "loss": 1.0793,
+      "step": 8751
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.416180236671809e-06,
+      "loss": 1.0675,
+      "step": 8752
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.404971773999124e-06,
+      "loss": 1.0522,
+      "step": 8753
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.393772786044072e-06,
+      "loss": 1.0356,
+      "step": 8754
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.382583273940313e-06,
+      "loss": 1.046,
+      "step": 8755
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.371403238820594e-06,
+      "loss": 1.1066,
+      "step": 8756
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.360232681816669e-06,
+      "loss": 1.0229,
+      "step": 8757
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.349071604059376e-06,
+      "loss": 1.0193,
+      "step": 8758
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.3379200066785215e-06,
+      "loss": 1.0896,
+      "step": 8759
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.326777890803004e-06,
+      "loss": 1.0626,
+      "step": 8760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.315645257560765e-06,
+      "loss": 1.0378,
+      "step": 8761
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.3045221080787475e-06,
+      "loss": 1.1081,
+      "step": 8762
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.293408443482995e-06,
+      "loss": 1.0423,
+      "step": 8763
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.28230426489852e-06,
+      "loss": 1.0214,
+      "step": 8764
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.271209573449421e-06,
+      "loss": 1.0632,
+      "step": 8765
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.260124370258846e-06,
+      "loss": 1.0624,
+      "step": 8766
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.249048656448941e-06,
+      "loss": 1.0527,
+      "step": 8767
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.237982433140887e-06,
+      "loss": 1.074,
+      "step": 8768
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.226925701454988e-06,
+      "loss": 1.0465,
+      "step": 8769
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.215878462510471e-06,
+      "loss": 1.0998,
+      "step": 8770
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.204840717425697e-06,
+      "loss": 1.0352,
+      "step": 8771
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.193812467318005e-06,
+      "loss": 1.0422,
+      "step": 8772
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.182793713303803e-06,
+      "loss": 1.0489,
+      "step": 8773
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.171784456498542e-06,
+      "loss": 1.0411,
+      "step": 8774
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.160784698016675e-06,
+      "loss": 0.9875,
+      "step": 8775
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.149794438971745e-06,
+      "loss": 1.0783,
+      "step": 8776
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.138813680476285e-06,
+      "loss": 1.0538,
+      "step": 8777
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.127842423641894e-06,
+      "loss": 1.0357,
+      "step": 8778
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.116880669579228e-06,
+      "loss": 1.0673,
+      "step": 8779
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.105928419397933e-06,
+      "loss": 1.069,
+      "step": 8780
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.094985674206699e-06,
+      "loss": 1.0635,
+      "step": 8781
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.084052435113308e-06,
+      "loss": 1.0413,
+      "step": 8782
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.073128703224518e-06,
+      "loss": 1.0911,
+      "step": 8783
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.0622144796461666e-06,
+      "loss": 1.0781,
+      "step": 8784
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.051309765483093e-06,
+      "loss": 1.0453,
+      "step": 8785
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.0404145618391915e-06,
+      "loss": 1.0334,
+      "step": 8786
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.029528869817425e-06,
+      "loss": 1.0272,
+      "step": 8787
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.018652690519732e-06,
+      "loss": 1.026,
+      "step": 8788
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.007786025047124e-06,
+      "loss": 1.0956,
+      "step": 8789
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9969288744996525e-06,
+      "loss": 1.0571,
+      "step": 8790
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.986081239976382e-06,
+      "loss": 1.1065,
+      "step": 8791
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9752431225754466e-06,
+      "loss": 1.0354,
+      "step": 8792
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.964414523394002e-06,
+      "loss": 1.0177,
+      "step": 8793
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.9535954435282146e-06,
+      "loss": 1.1448,
+      "step": 8794
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.942785884073354e-06,
+      "loss": 1.0922,
+      "step": 8795
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.931985846123645e-06,
+      "loss": 1.031,
+      "step": 8796
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.921195330772378e-06,
+      "loss": 1.1344,
+      "step": 8797
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.910414339111925e-06,
+      "loss": 1.0979,
+      "step": 8798
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.899642872233624e-06,
+      "loss": 1.0221,
+      "step": 8799
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.888880931227903e-06,
+      "loss": 1.0376,
+      "step": 8800
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8781285171842005e-06,
+      "loss": 1.0536,
+      "step": 8801
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8673856311909805e-06,
+      "loss": 1.0676,
+      "step": 8802
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.856652274335772e-06,
+      "loss": 1.0971,
+      "step": 8803
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.845928447705095e-06,
+      "loss": 1.0039,
+      "step": 8804
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.835214152384561e-06,
+      "loss": 1.0977,
+      "step": 8805
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.824509389458799e-06,
+      "loss": 1.1063,
+      "step": 8806
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.813814160011432e-06,
+      "loss": 1.1029,
+      "step": 8807
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8031284651251735e-06,
+      "loss": 1.0252,
+      "step": 8808
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.792452305881735e-06,
+      "loss": 1.0776,
+      "step": 8809
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.781785683361851e-06,
+      "loss": 1.1325,
+      "step": 8810
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.77112859864537e-06,
+      "loss": 1.1061,
+      "step": 8811
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.760481052811062e-06,
+      "loss": 1.0238,
+      "step": 8812
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.749843046936831e-06,
+      "loss": 1.1163,
+      "step": 8813
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.739214582099562e-06,
+      "loss": 1.0689,
+      "step": 8814
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.728595659375158e-06,
+      "loss": 1.1079,
+      "step": 8815
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.717986279838628e-06,
+      "loss": 1.0501,
+      "step": 8816
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.7073864445639225e-06,
+      "loss": 1.0288,
+      "step": 8817
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.696796154624118e-06,
+      "loss": 1.0257,
+      "step": 8818
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.686215411091245e-06,
+      "loss": 1.069,
+      "step": 8819
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.6756442150364134e-06,
+      "loss": 1.1405,
+      "step": 8820
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.665082567529778e-06,
+      "loss": 1.0691,
+      "step": 8821
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.654530469640484e-06,
+      "loss": 1.062,
+      "step": 8822
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.643987922436722e-06,
+      "loss": 1.048,
+      "step": 8823
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.633454926985749e-06,
+      "loss": 1.109,
+      "step": 8824
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.622931484353799e-06,
+      "loss": 1.0335,
+      "step": 8825
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.6124175956062116e-06,
+      "loss": 1.0828,
+      "step": 8826
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.60191326180729e-06,
+      "loss": 1.115,
+      "step": 8827
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.591418484020394e-06,
+      "loss": 1.1136,
+      "step": 8828
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.580933263307942e-06,
+      "loss": 1.0765,
+      "step": 8829
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.570457600731327e-06,
+      "loss": 1.0106,
+      "step": 8830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.559991497351058e-06,
+      "loss": 1.1161,
+      "step": 8831
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.549534954226599e-06,
+      "loss": 1.1245,
+      "step": 8832
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.53908797241649e-06,
+      "loss": 1.0459,
+      "step": 8833
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.528650552978265e-06,
+      "loss": 1.0898,
+      "step": 8834
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.518222696968556e-06,
+      "loss": 1.0605,
+      "step": 8835
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.5078044054429404e-06,
+      "loss": 0.992,
+      "step": 8836
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.497395679456108e-06,
+      "loss": 1.1027,
+      "step": 8837
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.486996520061738e-06,
+      "loss": 1.0961,
+      "step": 8838
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.476606928312511e-06,
+      "loss": 1.0999,
+      "step": 8839
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.46622690526023e-06,
+      "loss": 1.0148,
+      "step": 8840
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.455856451955632e-06,
+      "loss": 1.0962,
+      "step": 8841
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.445495569448556e-06,
+      "loss": 1.0515,
+      "step": 8842
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.435144258787839e-06,
+      "loss": 1.057,
+      "step": 8843
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.424802521021333e-06,
+      "loss": 1.0342,
+      "step": 8844
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.4144703571959665e-06,
+      "loss": 1.0684,
+      "step": 8845
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.404147768357659e-06,
+      "loss": 1.0676,
+      "step": 8846
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.393834755551385e-06,
+      "loss": 1.03,
+      "step": 8847
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.383531319821155e-06,
+      "loss": 1.0811,
+      "step": 8848
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.373237462209968e-06,
+      "loss": 1.1056,
+      "step": 8849
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.3629531837599e-06,
+      "loss": 1.0634,
+      "step": 8850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.352678485512042e-06,
+      "loss": 1.1452,
+      "step": 8851
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.342413368506483e-06,
+      "loss": 1.0851,
+      "step": 8852
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.332157833782414e-06,
+      "loss": 1.0831,
+      "step": 8853
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.321911882377961e-06,
+      "loss": 1.0335,
+      "step": 8854
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.311675515330394e-06,
+      "loss": 1.0558,
+      "step": 8855
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.301448733675906e-06,
+      "loss": 0.9987,
+      "step": 8856
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.291231538449759e-06,
+      "loss": 1.098,
+      "step": 8857
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.281023930686291e-06,
+      "loss": 1.07,
+      "step": 8858
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.270825911418787e-06,
+      "loss": 1.0643,
+      "step": 8859
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.26063748167962e-06,
+      "loss": 1.0611,
+      "step": 8860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.250458642500178e-06,
+      "loss": 1.0479,
+      "step": 8861
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.240289394910869e-06,
+      "loss": 1.0285,
+      "step": 8862
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.230129739941147e-06,
+      "loss": 1.069,
+      "step": 8863
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.21997967861948e-06,
+      "loss": 1.0781,
+      "step": 8864
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.209839211973355e-06,
+      "loss": 1.065,
+      "step": 8865
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.199708341029319e-06,
+      "loss": 1.0867,
+      "step": 8866
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.1895870668129065e-06,
+      "loss": 1.0805,
+      "step": 8867
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.179475390348731e-06,
+      "loss": 1.0822,
+      "step": 8868
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.1693733126603975e-06,
+      "loss": 1.0881,
+      "step": 8869
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.15928083477053e-06,
+      "loss": 1.0834,
+      "step": 8870
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.149197957700847e-06,
+      "loss": 1.0678,
+      "step": 8871
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.1391246824719964e-06,
+      "loss": 1.084,
+      "step": 8872
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.129061010103742e-06,
+      "loss": 1.093,
+      "step": 8873
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.119006941614812e-06,
+      "loss": 1.0792,
+      "step": 8874
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.108962478023016e-06,
+      "loss": 1.0973,
+      "step": 8875
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.09892762034514e-06,
+      "loss": 1.0915,
+      "step": 8876
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.088902369597048e-06,
+      "loss": 1.061,
+      "step": 8877
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.078886726793586e-06,
+      "loss": 1.1075,
+      "step": 8878
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.068880692948663e-06,
+      "loss": 1.072,
+      "step": 8879
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.058884269075192e-06,
+      "loss": 1.1006,
+      "step": 8880
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.048897456185109e-06,
+      "loss": 1.0843,
+      "step": 8881
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.038920255289404e-06,
+      "loss": 1.0482,
+      "step": 8882
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.028952667398068e-06,
+      "loss": 1.1418,
+      "step": 8883
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.018994693520163e-06,
+      "loss": 1.0784,
+      "step": 8884
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.009046334663703e-06,
+      "loss": 1.1302,
+      "step": 8885
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.999107591835783e-06,
+      "loss": 1.0288,
+      "step": 8886
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.98917846604251e-06,
+      "loss": 1.0368,
+      "step": 8887
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.979258958289035e-06,
+      "loss": 1.0477,
+      "step": 8888
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.9693490695794985e-06,
+      "loss": 1.0929,
+      "step": 8889
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.95944880091711e-06,
+      "loss": 1.0998,
+      "step": 8890
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.949558153304068e-06,
+      "loss": 1.04,
+      "step": 8891
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.9396771277416155e-06,
+      "loss": 1.0827,
+      "step": 8892
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.92980572523003e-06,
+      "loss": 1.1172,
+      "step": 8893
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.919943946768579e-06,
+      "loss": 1.0864,
+      "step": 8894
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.910091793355609e-06,
+      "loss": 1.0932,
+      "step": 8895
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.900249265988422e-06,
+      "loss": 1.1558,
+      "step": 8896
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.890416365663442e-06,
+      "loss": 1.0946,
+      "step": 8897
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.880593093376029e-06,
+      "loss": 1.076,
+      "step": 8898
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.8707794501205995e-06,
+      "loss": 1.0687,
+      "step": 8899
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.860975436890614e-06,
+      "loss": 1.0768,
+      "step": 8900
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.851181054678522e-06,
+      "loss": 1.0834,
+      "step": 8901
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.841396304475832e-06,
+      "loss": 1.0554,
+      "step": 8902
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.831621187273083e-06,
+      "loss": 1.0839,
+      "step": 8903
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.821855704059786e-06,
+      "loss": 1.1444,
+      "step": 8904
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.812099855824537e-06,
+      "loss": 1.1911,
+      "step": 8905
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.802353643554913e-06,
+      "loss": 1.0865,
+      "step": 8906
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.792617068237537e-06,
+      "loss": 1.0137,
+      "step": 8907
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.782890130858053e-06,
+      "loss": 1.1096,
+      "step": 8908
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.773172832401129e-06,
+      "loss": 1.0516,
+      "step": 8909
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.763465173850468e-06,
+      "loss": 1.1022,
+      "step": 8910
+    },
+    {
+      "epoch": 3.57,
+      "eval_loss": 0.8183498978614807,
+      "eval_runtime": 501.46,
+      "eval_samples_per_second": 51.649,
+      "eval_steps_per_second": 25.825,
+      "step": 8910
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.753767156188771e-06,
+      "loss": 1.0728,
+      "step": 8911
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.744078780397765e-06,
+      "loss": 1.0962,
+      "step": 8912
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.734400047458243e-06,
+      "loss": 1.0067,
+      "step": 8913
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.724730958349977e-06,
+      "loss": 1.0771,
+      "step": 8914
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.715071514051772e-06,
+      "loss": 0.9993,
+      "step": 8915
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.705421715541491e-06,
+      "loss": 1.1344,
+      "step": 8916
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.695781563795965e-06,
+      "loss": 1.005,
+      "step": 8917
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.6861510597911e-06,
+      "loss": 1.072,
+      "step": 8918
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.676530204501783e-06,
+      "loss": 1.0726,
+      "step": 8919
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.666918998901937e-06,
+      "loss": 1.0772,
+      "step": 8920
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.657317443964526e-06,
+      "loss": 1.0533,
+      "step": 8921
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.647725540661541e-06,
+      "loss": 1.1144,
+      "step": 8922
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.6381432899639365e-06,
+      "loss": 1.0862,
+      "step": 8923
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.628570692841771e-06,
+      "loss": 1.0784,
+      "step": 8924
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.619007750264071e-06,
+      "loss": 1.0604,
+      "step": 8925
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.609454463198915e-06,
+      "loss": 1.0658,
+      "step": 8926
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.599910832613375e-06,
+      "loss": 1.0873,
+      "step": 8927
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.590376859473566e-06,
+      "loss": 1.0469,
+      "step": 8928
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.5808525447446295e-06,
+      "loss": 1.0153,
+      "step": 8929
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.571337889390726e-06,
+      "loss": 1.0196,
+      "step": 8930
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.561832894375018e-06,
+      "loss": 1.1342,
+      "step": 8931
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.5523375606597255e-06,
+      "loss": 1.1095,
+      "step": 8932
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.5428518892060454e-06,
+      "loss": 1.084,
+      "step": 8933
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.533375880974256e-06,
+      "loss": 1.0476,
+      "step": 8934
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.523909536923588e-06,
+      "loss": 1.0362,
+      "step": 8935
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.514452858012342e-06,
+      "loss": 1.1048,
+      "step": 8936
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.50500584519783e-06,
+      "loss": 1.0749,
+      "step": 8937
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.495568499436375e-06,
+      "loss": 1.042,
+      "step": 8938
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.486140821683338e-06,
+      "loss": 1.0635,
+      "step": 8939
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 4.476722812893097e-06,
+      "loss": 1.0881,
+      "step": 8940
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.4673144740190245e-06,
+      "loss": 1.0937,
+      "step": 8941
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.457915806013546e-06,
+      "loss": 1.1316,
+      "step": 8942
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.448526809828113e-06,
+      "loss": 1.0386,
+      "step": 8943
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.439147486413153e-06,
+      "loss": 1.0732,
+      "step": 8944
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.429777836718185e-06,
+      "loss": 1.0513,
+      "step": 8945
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.420417861691672e-06,
+      "loss": 1.0927,
+      "step": 8946
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.411067562281157e-06,
+      "loss": 1.0837,
+      "step": 8947
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.40172693943317e-06,
+      "loss": 1.1096,
+      "step": 8948
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.392395994093257e-06,
+      "loss": 1.0906,
+      "step": 8949
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.383074727206027e-06,
+      "loss": 0.9909,
+      "step": 8950
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.373763139715048e-06,
+      "loss": 1.1416,
+      "step": 8951
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.364461232562988e-06,
+      "loss": 1.0544,
+      "step": 8952
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.355169006691451e-06,
+      "loss": 1.0279,
+      "step": 8953
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.345886463041094e-06,
+      "loss": 1.0927,
+      "step": 8954
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.336613602551631e-06,
+      "loss": 0.975,
+      "step": 8955
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.327350426161725e-06,
+      "loss": 1.0232,
+      "step": 8956
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.3180969348091235e-06,
+      "loss": 1.1061,
+      "step": 8957
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.308853129430568e-06,
+      "loss": 1.0432,
+      "step": 8958
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.299619010961797e-06,
+      "loss": 0.9996,
+      "step": 8959
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.2903945803376086e-06,
+      "loss": 1.077,
+      "step": 8960
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.2811798384918e-06,
+      "loss": 1.0339,
+      "step": 8961
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.271974786357169e-06,
+      "loss": 1.0946,
+      "step": 8962
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.262779424865582e-06,
+      "loss": 1.0687,
+      "step": 8963
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.25359375494786e-06,
+      "loss": 1.0276,
+      "step": 8964
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.244417777533904e-06,
+      "loss": 1.107,
+      "step": 8965
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.235251493552605e-06,
+      "loss": 1.0453,
+      "step": 8966
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.226094903931854e-06,
+      "loss": 1.1223,
+      "step": 8967
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.21694800959862e-06,
+      "loss": 1.0882,
+      "step": 8968
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.2078108114788185e-06,
+      "loss": 1.1157,
+      "step": 8969
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.198683310497398e-06,
+      "loss": 1.0626,
+      "step": 8970
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.189565507578408e-06,
+      "loss": 1.0488,
+      "step": 8971
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.18045740364481e-06,
+      "loss": 1.1,
+      "step": 8972
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.171358999618635e-06,
+      "loss": 1.0882,
+      "step": 8973
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.162270296420923e-06,
+      "loss": 1.0245,
+      "step": 8974
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.153191294971737e-06,
+      "loss": 1.0626,
+      "step": 8975
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.144121996190142e-06,
+      "loss": 1.0706,
+      "step": 8976
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.135062400994249e-06,
+      "loss": 1.0616,
+      "step": 8977
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.126012510301158e-06,
+      "loss": 1.1073,
+      "step": 8978
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.116972325027002e-06,
+      "loss": 1.0258,
+      "step": 8979
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.107941846086915e-06,
+      "loss": 1.0363,
+      "step": 8980
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.09892107439509e-06,
+      "loss": 1.0085,
+      "step": 8981
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.0899100108646835e-06,
+      "loss": 1.0061,
+      "step": 8982
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.0809086564079e-06,
+      "loss": 1.0979,
+      "step": 8983
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.071917011935944e-06,
+      "loss": 1.1116,
+      "step": 8984
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.062935078359087e-06,
+      "loss": 1.066,
+      "step": 8985
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.053962856586546e-06,
+      "loss": 1.059,
+      "step": 8986
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.045000347526595e-06,
+      "loss": 1.0533,
+      "step": 8987
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.036047552086519e-06,
+      "loss": 1.1438,
+      "step": 8988
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 4.027104471172627e-06,
+      "loss": 1.0854,
+      "step": 8989
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.018171105690227e-06,
+      "loss": 1.043,
+      "step": 8990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.009247456543641e-06,
+      "loss": 1.1093,
+      "step": 8991
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.000333524636246e-06,
+      "loss": 1.1235,
+      "step": 8992
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.991429310870376e-06,
+      "loss": 1.0467,
+      "step": 8993
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.982534816147454e-06,
+      "loss": 1.1004,
+      "step": 8994
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.973650041367838e-06,
+      "loss": 1.0779,
+      "step": 8995
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.964774987430963e-06,
+      "loss": 1.0763,
+      "step": 8996
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.955909655235246e-06,
+      "loss": 1.064,
+      "step": 8997
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.947054045678167e-06,
+      "loss": 1.0565,
+      "step": 8998
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.9382081596561425e-06,
+      "loss": 1.0266,
+      "step": 8999
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.929371998064691e-06,
+      "loss": 1.1122,
+      "step": 9000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.920545561798272e-06,
+      "loss": 1.0804,
+      "step": 9001
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.911728851750418e-06,
+      "loss": 1.1168,
+      "step": 9002
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.902921868813647e-06,
+      "loss": 1.0969,
+      "step": 9003
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.89412461387949e-06,
+      "loss": 1.0173,
+      "step": 9004
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.885337087838526e-06,
+      "loss": 1.1243,
+      "step": 9005
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.876559291580284e-06,
+      "loss": 1.0936,
+      "step": 9006
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.867791225993389e-06,
+      "loss": 1.0872,
+      "step": 9007
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.859032891965419e-06,
+      "loss": 1.0898,
+      "step": 9008
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.850284290382988e-06,
+      "loss": 1.0964,
+      "step": 9009
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.841545422131754e-06,
+      "loss": 1.0192,
+      "step": 9010
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.832816288096308e-06,
+      "loss": 1.0461,
+      "step": 9011
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.824096889160356e-06,
+      "loss": 1.0694,
+      "step": 9012
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.815387226206557e-06,
+      "loss": 1.0677,
+      "step": 9013
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8066873001166072e-06,
+      "loss": 1.0338,
+      "step": 9014
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7979971117711787e-06,
+      "loss": 1.1138,
+      "step": 9015
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.789316662050024e-06,
+      "loss": 1.0464,
+      "step": 9016
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7806459518318514e-06,
+      "loss": 1.0951,
+      "step": 9017
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7719849819944142e-06,
+      "loss": 1.1173,
+      "step": 9018
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7633337534144773e-06,
+      "loss": 1.0659,
+      "step": 9019
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7546922669677965e-06,
+      "loss": 1.0372,
+      "step": 9020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.746060523529171e-06,
+      "loss": 1.0457,
+      "step": 9021
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7374385239723808e-06,
+      "loss": 1.0816,
+      "step": 9022
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7288262691702715e-06,
+      "loss": 1.023,
+      "step": 9023
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7202237599946452e-06,
+      "loss": 1.1333,
+      "step": 9024
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.71163099731634e-06,
+      "loss": 1.078,
+      "step": 9025
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7030479820052365e-06,
+      "loss": 1.1155,
+      "step": 9026
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6944747149301852e-06,
+      "loss": 1.0921,
+      "step": 9027
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.685911196959058e-06,
+      "loss": 1.0754,
+      "step": 9028
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.677357428958772e-06,
+      "loss": 1.0241,
+      "step": 9029
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.668813411795202e-06,
+      "loss": 1.1044,
+      "step": 9030
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.660279146333301e-06,
+      "loss": 1.1031,
+      "step": 9031
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6517546334369877e-06,
+      "loss": 1.0524,
+      "step": 9032
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6432398739691842e-06,
+      "loss": 1.0667,
+      "step": 9033
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.634734868791878e-06,
+      "loss": 1.0968,
+      "step": 9034
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6262396187660253e-06,
+      "loss": 1.1105,
+      "step": 9035
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6177541247516155e-06,
+      "loss": 1.0406,
+      "step": 9036
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.609278387607651e-06,
+      "loss": 1.027,
+      "step": 9037
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6008124081921113e-06,
+      "loss": 1.0705,
+      "step": 9038
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.592356187362034e-06,
+      "loss": 1.046,
+      "step": 9039
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5839097259734557e-06,
+      "loss": 1.1383,
+      "step": 9040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5754730248814037e-06,
+      "loss": 1.0856,
+      "step": 9041
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5670460849399624e-06,
+      "loss": 1.0268,
+      "step": 9042
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5586289070021706e-06,
+      "loss": 1.1061,
+      "step": 9043
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5502214919201247e-06,
+      "loss": 1.0862,
+      "step": 9044
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5418238405449113e-06,
+      "loss": 1.0369,
+      "step": 9045
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.533435953726616e-06,
+      "loss": 1.0473,
+      "step": 9046
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.525057832314371e-06,
+      "loss": 1.0691,
+      "step": 9047
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5166894771562987e-06,
+      "loss": 1.0811,
+      "step": 9048
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.5083308890995425e-06,
+      "loss": 0.9885,
+      "step": 9049
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4999820689902486e-06,
+      "loss": 1.0281,
+      "step": 9050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4916430176735625e-06,
+      "loss": 1.034,
+      "step": 9051
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.483313735993654e-06,
+      "loss": 1.0344,
+      "step": 9052
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.474994224793737e-06,
+      "loss": 1.0375,
+      "step": 9053
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.466684484915961e-06,
+      "loss": 1.0624,
+      "step": 9054
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4583845172015628e-06,
+      "loss": 1.1043,
+      "step": 9055
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.450094322490738e-06,
+      "loss": 1.1311,
+      "step": 9056
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4418139016227146e-06,
+      "loss": 1.0428,
+      "step": 9057
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4335432554357338e-06,
+      "loss": 1.0792,
+      "step": 9058
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4252823847670257e-06,
+      "loss": 1.0985,
+      "step": 9059
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4170312904528766e-06,
+      "loss": 1.0949,
+      "step": 9060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.40878997332853e-06,
+      "loss": 1.0557,
+      "step": 9061
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.4005584342282515e-06,
+      "loss": 1.0875,
+      "step": 9062
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.3923366739853522e-06,
+      "loss": 1.0917,
+      "step": 9063
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.3841246934321113e-06,
+      "loss": 1.0801,
+      "step": 9064
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3759224933998634e-06,
+      "loss": 1.1334,
+      "step": 9065
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3677300747188887e-06,
+      "loss": 1.0666,
+      "step": 9066
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3595474382185355e-06,
+      "loss": 1.0658,
+      "step": 9067
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3513745847271404e-06,
+      "loss": 1.049,
+      "step": 9068
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3432115150720533e-06,
+      "loss": 1.091,
+      "step": 9069
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.335058230079602e-06,
+      "loss": 1.1127,
+      "step": 9070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3269147305751813e-06,
+      "loss": 1.0099,
+      "step": 9071
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3187810173831545e-06,
+      "loss": 1.0578,
+      "step": 9072
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.3106570913269185e-06,
+      "loss": 1.0883,
+      "step": 9073
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.302542953228849e-06,
+      "loss": 1.1266,
+      "step": 9074
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2944386039103437e-06,
+      "loss": 1.0525,
+      "step": 9075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.286344044191836e-06,
+      "loss": 1.1173,
+      "step": 9076
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2782592748927256e-06,
+      "loss": 1.0835,
+      "step": 9077
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.270184296831458e-06,
+      "loss": 1.0849,
+      "step": 9078
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2621191108254676e-06,
+      "loss": 1.0048,
+      "step": 9079
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2540637176911695e-06,
+      "loss": 1.0514,
+      "step": 9080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.246018118244076e-06,
+      "loss": 1.0852,
+      "step": 9081
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2379823132986154e-06,
+      "loss": 1.1141,
+      "step": 9082
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2299563036682577e-06,
+      "loss": 1.0629,
+      "step": 9083
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.2219400901655094e-06,
+      "loss": 1.058,
+      "step": 9084
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.213933673601821e-06,
+      "loss": 1.1057,
+      "step": 9085
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.205937054787733e-06,
+      "loss": 1.1067,
+      "step": 9086
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.197950234532732e-06,
+      "loss": 1.0958,
+      "step": 9087
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1899732136453143e-06,
+      "loss": 1.1063,
+      "step": 9088
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1820059929330347e-06,
+      "loss": 1.0672,
+      "step": 9089
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.174048573202393e-06,
+      "loss": 0.9948,
+      "step": 9090
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1661009552589437e-06,
+      "loss": 1.0735,
+      "step": 9091
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.158163139907233e-06,
+      "loss": 1.0421,
+      "step": 9092
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1502351279507957e-06,
+      "loss": 1.0515,
+      "step": 9093
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.142316920192212e-06,
+      "loss": 1.0377,
+      "step": 9094
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.134408517433052e-06,
+      "loss": 1.0706,
+      "step": 9095
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1265099204738633e-06,
+      "loss": 1.0446,
+      "step": 9096
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1186211301142633e-06,
+      "loss": 1.1614,
+      "step": 9097
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.110742147152823e-06,
+      "loss": 1.0465,
+      "step": 9098
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1028729723871387e-06,
+      "loss": 1.0427,
+      "step": 9099
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.095013606613828e-06,
+      "loss": 1.0708,
+      "step": 9100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0871640506284884e-06,
+      "loss": 1.0511,
+      "step": 9101
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0793243052257502e-06,
+      "loss": 1.0692,
+      "step": 9102
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.071494371199224e-06,
+      "loss": 1.0611,
+      "step": 9103
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0636742493415526e-06,
+      "loss": 1.0218,
+      "step": 9104
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.05586394044437e-06,
+      "loss": 1.0452,
+      "step": 9105
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0480634452983216e-06,
+      "loss": 1.01,
+      "step": 9106
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.040272764693064e-06,
+      "loss": 1.0865,
+      "step": 9107
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0324918994172447e-06,
+      "loss": 1.0507,
+      "step": 9108
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0247208502585446e-06,
+      "loss": 1.0998,
+      "step": 9109
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.016959618003623e-06,
+      "loss": 1.1123,
+      "step": 9110
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.009208203438174e-06,
+      "loss": 1.1146,
+      "step": 9111
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.0014666073468478e-06,
+      "loss": 1.0147,
+      "step": 9112
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.993734830513362e-06,
+      "loss": 1.1332,
+      "step": 9113
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9860128737204005e-06,
+      "loss": 1.1492,
+      "step": 9114
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.978300737749673e-06,
+      "loss": 1.0216,
+      "step": 9115
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9705984233818872e-06,
+      "loss": 1.0826,
+      "step": 9116
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9629059313967422e-06,
+      "loss": 1.0445,
+      "step": 9117
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9552232625729593e-06,
+      "loss": 1.054,
+      "step": 9118
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.947550417688272e-06,
+      "loss": 1.0149,
+      "step": 9119
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.939887397519414e-06,
+      "loss": 1.0434,
+      "step": 9120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9322342028420992e-06,
+      "loss": 1.0997,
+      "step": 9121
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9245908344310846e-06,
+      "loss": 1.1089,
+      "step": 9122
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9169572930601188e-06,
+      "loss": 1.0922,
+      "step": 9123
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9093335795019494e-06,
+      "loss": 1.0731,
+      "step": 9124
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9017196945283154e-06,
+      "loss": 1.0516,
+      "step": 9125
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.894115638910011e-06,
+      "loss": 1.1246,
+      "step": 9126
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.886521413416765e-06,
+      "loss": 1.0725,
+      "step": 9127
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8789370188173846e-06,
+      "loss": 1.0923,
+      "step": 9128
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8713624558796226e-06,
+      "loss": 1.026,
+      "step": 9129
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8637977253702543e-06,
+      "loss": 1.0181,
+      "step": 9130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.856242828055089e-06,
+      "loss": 1.0737,
+      "step": 9131
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.848697764698893e-06,
+      "loss": 1.1062,
+      "step": 9132
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8411625360654782e-06,
+      "loss": 1.0888,
+      "step": 9133
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8336371429176333e-06,
+      "loss": 1.082,
+      "step": 9134
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8261215860171495e-06,
+      "loss": 1.0736,
+      "step": 9135
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.818615866124863e-06,
+      "loss": 1.0383,
+      "step": 9136
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.8111199840005765e-06,
+      "loss": 1.0555,
+      "step": 9137
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.803633940403083e-06,
+      "loss": 1.0467,
+      "step": 9138
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.7961577360902323e-06,
+      "loss": 1.0317,
+      "step": 9139
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.78869137181883e-06,
+      "loss": 1.1092,
+      "step": 9140
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.7812348483447047e-06,
+      "loss": 1.0649,
+      "step": 9141
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.7737881664227085e-06,
+      "loss": 1.0068,
+      "step": 9142
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.7663513268066377e-06,
+      "loss": 1.0845,
+      "step": 9143
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.758924330249357e-06,
+      "loss": 1.133,
+      "step": 9144
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.75150717750271e-06,
+      "loss": 1.0606,
+      "step": 9145
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.744099869317529e-06,
+      "loss": 1.0934,
+      "step": 9146
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.736702406443681e-06,
+      "loss": 1.0598,
+      "step": 9147
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.729314789629989e-06,
+      "loss": 1.0942,
+      "step": 9148
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.721937019624332e-06,
+      "loss": 1.0302,
+      "step": 9149
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.7145690971735806e-06,
+      "loss": 1.0559,
+      "step": 9150
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.7072110230235593e-06,
+      "loss": 1.103,
+      "step": 9151
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6998627979191724e-06,
+      "loss": 1.0534,
+      "step": 9152
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6925244226042703e-06,
+      "loss": 1.0555,
+      "step": 9153
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.685195897821713e-06,
+      "loss": 1.0403,
+      "step": 9154
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6778772243133965e-06,
+      "loss": 1.0936,
+      "step": 9155
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.670568402820173e-06,
+      "loss": 0.9882,
+      "step": 9156
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.66326943408195e-06,
+      "loss": 1.0625,
+      "step": 9157
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.655980318837592e-06,
+      "loss": 1.0406,
+      "step": 9158
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.648701057824976e-06,
+      "loss": 1.0803,
+      "step": 9159
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.641431651781012e-06,
+      "loss": 1.0426,
+      "step": 9160
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6341721014415676e-06,
+      "loss": 1.0238,
+      "step": 9161
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.626922407541543e-06,
+      "loss": 1.0773,
+      "step": 9162
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6196825708148298e-06,
+      "loss": 1.0209,
+      "step": 9163
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.612452591994319e-06,
+      "loss": 1.0698,
+      "step": 9164
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.605232471811936e-06,
+      "loss": 1.048,
+      "step": 9165
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5980222109985517e-06,
+      "loss": 1.0638,
+      "step": 9166
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.590821810284061e-06,
+      "loss": 1.0839,
+      "step": 9167
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.583631270397402e-06,
+      "loss": 0.9874,
+      "step": 9168
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.576450592066448e-06,
+      "loss": 1.0643,
+      "step": 9169
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.56927977601813e-06,
+      "loss": 1.0217,
+      "step": 9170
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5621188229783433e-06,
+      "loss": 1.0588,
+      "step": 9171
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5549677336719867e-06,
+      "loss": 1.1196,
+      "step": 9172
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5478265088230035e-06,
+      "loss": 1.0879,
+      "step": 9173
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5406951491542817e-06,
+      "loss": 1.0381,
+      "step": 9174
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.533573655387744e-06,
+      "loss": 1.0502,
+      "step": 9175
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.526462028244303e-06,
+      "loss": 1.0427,
+      "step": 9176
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5193602684438824e-06,
+      "loss": 1.0364,
+      "step": 9177
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.5122683767054066e-06,
+      "loss": 1.0418,
+      "step": 9178
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.50518635374678e-06,
+      "loss": 1.0617,
+      "step": 9179
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.498114200284918e-06,
+      "loss": 1.0661,
+      "step": 9180
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.4910519170357695e-06,
+      "loss": 1.0605,
+      "step": 9181
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.4839995047142183e-06,
+      "loss": 0.9806,
+      "step": 9182
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.4769569640342272e-06,
+      "loss": 1.0777,
+      "step": 9183
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.469924295708692e-06,
+      "loss": 1.0902,
+      "step": 9184
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.4629015004495438e-06,
+      "loss": 1.1491,
+      "step": 9185
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.455888578967702e-06,
+      "loss": 1.0785,
+      "step": 9186
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.4488855319730995e-06,
+      "loss": 1.0756,
+      "step": 9187
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.441892360174669e-06,
+      "loss": 1.0676,
+      "step": 9188
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.4349090642803328e-06,
+      "loss": 1.0763,
+      "step": 9189
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.4279356449969927e-06,
+      "loss": 1.0676,
+      "step": 9190
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.4209721030305943e-06,
+      "loss": 1.0719,
+      "step": 9191
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.4140184390860744e-06,
+      "loss": 1.0546,
+      "step": 9192
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.407074653867336e-06,
+      "loss": 1.0859,
+      "step": 9193
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.4001407480773287e-06,
+      "loss": 1.0514,
+      "step": 9194
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.393216722417968e-06,
+      "loss": 1.0461,
+      "step": 9195
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3863025775901603e-06,
+      "loss": 1.0964,
+      "step": 9196
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.379398314293868e-06,
+      "loss": 1.0508,
+      "step": 9197
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3725039332279873e-06,
+      "loss": 1.0572,
+      "step": 9198
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3656194350904603e-06,
+      "loss": 1.1093,
+      "step": 9199
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.358744820578218e-06,
+      "loss": 1.0571,
+      "step": 9200
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3518800903871486e-06,
+      "loss": 1.0727,
+      "step": 9201
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3450252452122177e-06,
+      "loss": 1.0735,
+      "step": 9202
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3381802857473154e-06,
+      "loss": 1.0275,
+      "step": 9203
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3313452126853764e-06,
+      "loss": 1.0467,
+      "step": 9204
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3245200267183356e-06,
+      "loss": 1.101,
+      "step": 9205
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3177047285370956e-06,
+      "loss": 1.0733,
+      "step": 9206
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.310899318831583e-06,
+      "loss": 1.0763,
+      "step": 9207
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.3041037982907243e-06,
+      "loss": 0.9972,
+      "step": 9208
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.2973181676024136e-06,
+      "loss": 1.0104,
+      "step": 9209
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.290542427453579e-06,
+      "loss": 1.0627,
+      "step": 9210
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.283776578530139e-06,
+      "loss": 1.042,
+      "step": 9211
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.277020621517012e-06,
+      "loss": 1.0432,
+      "step": 9212
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.2702745570981063e-06,
+      "loss": 1.0488,
+      "step": 9213
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.2635383859563208e-06,
+      "loss": 1.0981,
+      "step": 9214
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.256812108773576e-06,
+      "loss": 1.0795,
+      "step": 9215
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.250095726230772e-06,
+      "loss": 1.0798,
+      "step": 9216
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.2433892390078317e-06,
+      "loss": 1.0903,
+      "step": 9217
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.2366926477836336e-06,
+      "loss": 1.0887,
+      "step": 9218
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.230005953236092e-06,
+      "loss": 1.0974,
+      "step": 9219
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.223329156042131e-06,
+      "loss": 1.1361,
+      "step": 9220
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.216662256877622e-06,
+      "loss": 1.1501,
+      "step": 9221
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.2100052564174577e-06,
+      "loss": 1.1062,
+      "step": 9222
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.2033581553355554e-06,
+      "loss": 1.0545,
+      "step": 9223
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1967209543047875e-06,
+      "loss": 1.0041,
+      "step": 9224
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.190093653997061e-06,
+      "loss": 1.0691,
+      "step": 9225
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.18347625508325e-06,
+      "loss": 1.1112,
+      "step": 9226
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1768687582332414e-06,
+      "loss": 1.0629,
+      "step": 9227
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1702711641159336e-06,
+      "loss": 1.0868,
+      "step": 9228
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1636834733991916e-06,
+      "loss": 1.0852,
+      "step": 9229
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.157105686749905e-06,
+      "loss": 1.1471,
+      "step": 9230
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1505378048339407e-06,
+      "loss": 1.1023,
+      "step": 9231
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.143979828316178e-06,
+      "loss": 1.0599,
+      "step": 9232
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1374317578604753e-06,
+      "loss": 1.03,
+      "step": 9233
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1308935941297236e-06,
+      "loss": 1.0943,
+      "step": 9234
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.124365337785772e-06,
+      "loss": 1.0648,
+      "step": 9235
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.1178469894894804e-06,
+      "loss": 1.0592,
+      "step": 9236
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.111338549900721e-06,
+      "loss": 1.0183,
+      "step": 9237
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.1048400196783224e-06,
+      "loss": 1.0573,
+      "step": 9238
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0983513994801695e-06,
+      "loss": 1.1331,
+      "step": 9239
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0918726899630923e-06,
+      "loss": 1.1404,
+      "step": 9240
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0854038917829555e-06,
+      "loss": 1.0921,
+      "step": 9241
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0789450055945792e-06,
+      "loss": 1.0729,
+      "step": 9242
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.072496032051807e-06,
+      "loss": 1.0651,
+      "step": 9243
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.066056971807484e-06,
+      "loss": 1.1165,
+      "step": 9244
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0596278255134326e-06,
+      "loss": 1.1013,
+      "step": 9245
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.053208593820488e-06,
+      "loss": 1.1413,
+      "step": 9246
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.046799277378486e-06,
+      "loss": 1.1215,
+      "step": 9247
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.04039987683623e-06,
+      "loss": 1.074,
+      "step": 9248
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0340103928415564e-06,
+      "loss": 1.0777,
+      "step": 9249
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0276308260412603e-06,
+      "loss": 1.0798,
+      "step": 9250
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0212611770811573e-06,
+      "loss": 1.1102,
+      "step": 9251
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0149014466060655e-06,
+      "loss": 1.0503,
+      "step": 9252
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.008551635259759e-06,
+      "loss": 1.0671,
+      "step": 9253
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.0022117436850675e-06,
+      "loss": 1.1121,
+      "step": 9254
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.995881772523767e-06,
+      "loss": 1.0985,
+      "step": 9255
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.989561722416633e-06,
+      "loss": 1.1232,
+      "step": 9256
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9832515940034878e-06,
+      "loss": 1.0389,
+      "step": 9257
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.976951387923076e-06,
+      "loss": 0.9728,
+      "step": 9258
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9706611048131986e-06,
+      "loss": 1.0266,
+      "step": 9259
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.964380745310612e-06,
+      "loss": 1.0267,
+      "step": 9260
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9581103100510865e-06,
+      "loss": 1.0624,
+      "step": 9261
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.951849799669403e-06,
+      "loss": 1.0639,
+      "step": 9262
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9455992147992984e-06,
+      "loss": 1.0697,
+      "step": 9263
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.939358556073512e-06,
+      "loss": 1.1467,
+      "step": 9264
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.933127824123826e-06,
+      "loss": 1.0734,
+      "step": 9265
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.926907019580959e-06,
+      "loss": 1.0569,
+      "step": 9266
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9206961430746738e-06,
+      "loss": 1.1208,
+      "step": 9267
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9144951952336787e-06,
+      "loss": 1.1073,
+      "step": 9268
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9083041766857044e-06,
+      "loss": 1.1108,
+      "step": 9269
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9021230880574835e-06,
+      "loss": 1.0915,
+      "step": 9270
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8959519299747263e-06,
+      "loss": 1.0244,
+      "step": 9271
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8897907030621554e-06,
+      "loss": 1.0862,
+      "step": 9272
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8836394079434716e-06,
+      "loss": 1.0846,
+      "step": 9273
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8774980452413659e-06,
+      "loss": 1.0928,
+      "step": 9274
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8713666155775633e-06,
+      "loss": 1.1038,
+      "step": 9275
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8652451195727339e-06,
+      "loss": 1.0341,
+      "step": 9276
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8591335578465484e-06,
+      "loss": 1.118,
+      "step": 9277
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8530319310177235e-06,
+      "loss": 1.0517,
+      "step": 9278
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8469402397039094e-06,
+      "loss": 1.0604,
+      "step": 9279
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.840858484521768e-06,
+      "loss": 1.0635,
+      "step": 9280
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.834786666086985e-06,
+      "loss": 1.1086,
+      "step": 9281
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8287247850141797e-06,
+      "loss": 1.0463,
+      "step": 9282
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8226728419170502e-06,
+      "loss": 0.9999,
+      "step": 9283
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8166308374082064e-06,
+      "loss": 1.1123,
+      "step": 9284
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8105987720992922e-06,
+      "loss": 1.0841,
+      "step": 9285
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8045766466009418e-06,
+      "loss": 1.0666,
+      "step": 9286
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7985644615227893e-06,
+      "loss": 1.0769,
+      "step": 9287
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7925622174734368e-06,
+      "loss": 1.0684,
+      "step": 9288
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.786569915060521e-06,
+      "loss": 1.0183,
+      "step": 9289
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7805875548906226e-06,
+      "loss": 1.0462,
+      "step": 9290
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7746151375693686e-06,
+      "loss": 1.0587,
+      "step": 9291
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7686526637013312e-06,
+      "loss": 1.1011,
+      "step": 9292
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.762700133890094e-06,
+      "loss": 1.1839,
+      "step": 9293
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7567575487382638e-06,
+      "loss": 1.0507,
+      "step": 9294
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7508249088473927e-06,
+      "loss": 1.043,
+      "step": 9295
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7449022148180672e-06,
+      "loss": 1.0858,
+      "step": 9296
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7389894672498296e-06,
+      "loss": 1.0042,
+      "step": 9297
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7330866667412238e-06,
+      "loss": 1.0333,
+      "step": 9298
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.727193813889827e-06,
+      "loss": 1.0464,
+      "step": 9299
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7213109092921508e-06,
+      "loss": 1.0872,
+      "step": 9300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7154379535437304e-06,
+      "loss": 1.0901,
+      "step": 9301
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.709574947239123e-06,
+      "loss": 1.0482,
+      "step": 9302
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7037218909717989e-06,
+      "loss": 1.0987,
+      "step": 9303
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.697878785334317e-06,
+      "loss": 1.1037,
+      "step": 9304
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6920456309181488e-06,
+      "loss": 1.0524,
+      "step": 9305
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6862224283137883e-06,
+      "loss": 1.0857,
+      "step": 9306
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6804091781107422e-06,
+      "loss": 1.1546,
+      "step": 9307
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6746058808974729e-06,
+      "loss": 1.075,
+      "step": 9308
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.668812537261477e-06,
+      "loss": 1.1035,
+      "step": 9309
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6630291477892079e-06,
+      "loss": 1.1125,
+      "step": 9310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6572557130661081e-06,
+      "loss": 1.1041,
+      "step": 9311
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6514922336766547e-06,
+      "loss": 1.0762,
+      "step": 9312
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6457387102042808e-06,
+      "loss": 1.0705,
+      "step": 9313
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6399951432314099e-06,
+      "loss": 1.0395,
+      "step": 9314
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6342615333394873e-06,
+      "loss": 1.0268,
+      "step": 9315
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6285378811089158e-06,
+      "loss": 1.0589,
+      "step": 9316
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6228241871191207e-06,
+      "loss": 1.0365,
+      "step": 9317
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6171204519485061e-06,
+      "loss": 1.0221,
+      "step": 9318
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6114266761744434e-06,
+      "loss": 1.0839,
+      "step": 9319
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6057428603733492e-06,
+      "loss": 1.0239,
+      "step": 9320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6000690051205748e-06,
+      "loss": 1.0265,
+      "step": 9321
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5944051109905155e-06,
+      "loss": 1.0968,
+      "step": 9322
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5887511785565134e-06,
+      "loss": 1.0488,
+      "step": 9323
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5831072083909326e-06,
+      "loss": 0.9976,
+      "step": 9324
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.577473201065105e-06,
+      "loss": 1.0747,
+      "step": 9325
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5718491571493853e-06,
+      "loss": 1.1373,
+      "step": 9326
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5662350772130851e-06,
+      "loss": 1.0746,
+      "step": 9327
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5606309618245274e-06,
+      "loss": 1.1571,
+      "step": 9328
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5550368115510362e-06,
+      "loss": 1.1086,
+      "step": 9329
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5494526269588805e-06,
+      "loss": 1.0777,
+      "step": 9330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5438784086133973e-06,
+      "loss": 1.1245,
+      "step": 9331
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.538314157078824e-06,
+      "loss": 1.0947,
+      "step": 9332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5327598729184767e-06,
+      "loss": 1.096,
+      "step": 9333
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5272155566945944e-06,
+      "loss": 1.1281,
+      "step": 9334
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5216812089684397e-06,
+      "loss": 1.0781,
+      "step": 9335
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5161568303002637e-06,
+      "loss": 1.046,
+      "step": 9336
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5106424212492975e-06,
+      "loss": 1.0804,
+      "step": 9337
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5051379823737833e-06,
+      "loss": 1.0562,
+      "step": 9338
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4996435142309417e-06,
+      "loss": 1.1084,
+      "step": 9339
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4941590173769615e-06,
+      "loss": 1.1083,
+      "step": 9340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.488684492367054e-06,
+      "loss": 1.0272,
+      "step": 9341
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4832199397554203e-06,
+      "loss": 1.0828,
+      "step": 9342
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.477765360095229e-06,
+      "loss": 1.0833,
+      "step": 9343
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4723207539386718e-06,
+      "loss": 1.0553,
+      "step": 9344
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4668861218368857e-06,
+      "loss": 1.0392,
+      "step": 9345
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4614614643400527e-06,
+      "loss": 1.0536,
+      "step": 9346
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4560467819972889e-06,
+      "loss": 1.0463,
+      "step": 9347
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.450642075356745e-06,
+      "loss": 1.0332,
+      "step": 9348
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.445247344965539e-06,
+      "loss": 1.067,
+      "step": 9349
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4398625913697895e-06,
+      "loss": 1.1125,
+      "step": 9350
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4344878151145935e-06,
+      "loss": 1.0554,
+      "step": 9351
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4291230167440495e-06,
+      "loss": 1.0606,
+      "step": 9352
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.423768196801234e-06,
+      "loss": 1.0849,
+      "step": 9353
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4184233558282356e-06,
+      "loss": 1.0867,
+      "step": 9354
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4130884943660993e-06,
+      "loss": 1.0841,
+      "step": 9355
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4077636129548933e-06,
+      "loss": 1.1337,
+      "step": 9356
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4024487121336637e-06,
+      "loss": 1.1062,
+      "step": 9357
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3971437924404251e-06,
+      "loss": 1.1367,
+      "step": 9358
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.3918488544122144e-06,
+      "loss": 1.0362,
+      "step": 9359
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.386563898585036e-06,
+      "loss": 1.0994,
+      "step": 9360
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3812889254938954e-06,
+      "loss": 1.0434,
+      "step": 9361
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3760239356727766e-06,
+      "loss": 1.0486,
+      "step": 9362
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3707689296546645e-06,
+      "loss": 1.0357,
+      "step": 9363
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3655239079715444e-06,
+      "loss": 1.0769,
+      "step": 9364
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.360288871154347e-06,
+      "loss": 1.0289,
+      "step": 9365
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3550638197330379e-06,
+      "loss": 1.0817,
+      "step": 9366
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.349848754236549e-06,
+      "loss": 1.0395,
+      "step": 9367
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3446436751928026e-06,
+      "loss": 1.0346,
+      "step": 9368
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3394485831287219e-06,
+      "loss": 1.0225,
+      "step": 9369
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.334263478570208e-06,
+      "loss": 1.1113,
+      "step": 9370
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3290883620421524e-06,
+      "loss": 1.064,
+      "step": 9371
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.323923234068436e-06,
+      "loss": 1.0625,
+      "step": 9372
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3187680951719405e-06,
+      "loss": 1.0547,
+      "step": 9373
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3136229458745043e-06,
+      "loss": 1.0318,
+      "step": 9374
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.3084877866970103e-06,
+      "loss": 1.0938,
+      "step": 9375
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.303362618159265e-06,
+      "loss": 1.0563,
+      "step": 9376
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2982474407800981e-06,
+      "loss": 1.0772,
+      "step": 9377
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2931422550773399e-06,
+      "loss": 1.1165,
+      "step": 9378
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2880470615677653e-06,
+      "loss": 1.074,
+      "step": 9379
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2829618607672066e-06,
+      "loss": 0.9858,
+      "step": 9380
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2778866531904188e-06,
+      "loss": 1.094,
+      "step": 9381
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2728214393511573e-06,
+      "loss": 1.0717,
+      "step": 9382
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2677662197622008e-06,
+      "loss": 1.0767,
+      "step": 9383
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2627209949352958e-06,
+      "loss": 1.1217,
+      "step": 9384
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2576857653811557e-06,
+      "loss": 1.0087,
+      "step": 9385
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2526605316095175e-06,
+      "loss": 1.1012,
+      "step": 9386
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.247645294129085e-06,
+      "loss": 1.029,
+      "step": 9387
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2426400534475634e-06,
+      "loss": 1.0348,
+      "step": 9388
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2376448100716365e-06,
+      "loss": 1.0686,
+      "step": 9389
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2326595645069549e-06,
+      "loss": 1.0671,
+      "step": 9390
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2276843172582153e-06,
+      "loss": 1.0634,
+      "step": 9391
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2227190688290368e-06,
+      "loss": 1.0553,
+      "step": 9392
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.217763819722084e-06,
+      "loss": 1.0479,
+      "step": 9393
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.212818570438956e-06,
+      "loss": 1.074,
+      "step": 9394
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2078833214802853e-06,
+      "loss": 1.048,
+      "step": 9395
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2029580733456502e-06,
+      "loss": 1.159,
+      "step": 9396
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1980428265336629e-06,
+      "loss": 1.0687,
+      "step": 9397
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.193137581541881e-06,
+      "loss": 1.0781,
+      "step": 9398
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.188242338866874e-06,
+      "loss": 1.0376,
+      "step": 9399
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1833570990041897e-06,
+      "loss": 1.1043,
+      "step": 9400
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1784818624483774e-06,
+      "loss": 1.0959,
+      "step": 9401
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.173616629692942e-06,
+      "loss": 1.0476,
+      "step": 9402
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1687614012304116e-06,
+      "loss": 1.0583,
+      "step": 9403
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1639161775522821e-06,
+      "loss": 1.1153,
+      "step": 9404
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1590809591490392e-06,
+      "loss": 1.0261,
+      "step": 9405
+    },
+    {
+      "epoch": 3.77,
+      "eval_loss": 0.818169116973877,
+      "eval_runtime": 503.6475,
+      "eval_samples_per_second": 51.425,
+      "eval_steps_per_second": 25.712,
+      "step": 9405
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1542557465101467e-06,
+      "loss": 1.1165,
+      "step": 9406
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1494405401240914e-06,
+      "loss": 1.0492,
+      "step": 9407
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.144635340478295e-06,
+      "loss": 1.0992,
+      "step": 9408
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1398401480592013e-06,
+      "loss": 1.0859,
+      "step": 9409
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.1350549633522333e-06,
+      "loss": 1.0763,
+      "step": 9410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1302797868418035e-06,
+      "loss": 1.1119,
+      "step": 9411
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.125514619011314e-06,
+      "loss": 1.0372,
+      "step": 9412
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1207594603431348e-06,
+      "loss": 1.1019,
+      "step": 9413
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.116014311318625e-06,
+      "loss": 1.1321,
+      "step": 9414
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1112791724181782e-06,
+      "loss": 1.0893,
+      "step": 9415
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1065540441210887e-06,
+      "loss": 1.0578,
+      "step": 9416
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.1018389269057295e-06,
+      "loss": 1.0009,
+      "step": 9417
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0971338212493964e-06,
+      "loss": 1.0409,
+      "step": 9418
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0924387276283865e-06,
+      "loss": 1.067,
+      "step": 9419
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0877536465179973e-06,
+      "loss": 0.9793,
+      "step": 9420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0830785783925045e-06,
+      "loss": 1.1028,
+      "step": 9421
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0784135237251747e-06,
+      "loss": 1.0553,
+      "step": 9422
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0737584829882407e-06,
+      "loss": 1.012,
+      "step": 9423
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0691134566529482e-06,
+      "loss": 1.0508,
+      "step": 9424
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0644784451895096e-06,
+      "loss": 1.0742,
+      "step": 9425
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0598534490671496e-06,
+      "loss": 1.0212,
+      "step": 9426
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0552384687540383e-06,
+      "loss": 1.0985,
+      "step": 9427
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0506335047173799e-06,
+      "loss": 1.0676,
+      "step": 9428
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0460385574233121e-06,
+      "loss": 1.0076,
+      "step": 9429
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0414536273370079e-06,
+      "loss": 1.0861,
+      "step": 9430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0368787149225955e-06,
+      "loss": 1.1296,
+      "step": 9431
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0323138206431826e-06,
+      "loss": 1.1423,
+      "step": 9432
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0277589449609105e-06,
+      "loss": 1.0257,
+      "step": 9433
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.0232140883368435e-06,
+      "loss": 0.9441,
+      "step": 9434
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0186792512310806e-06,
+      "loss": 1.0339,
+      "step": 9435
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0141544341026876e-06,
+      "loss": 1.0527,
+      "step": 9436
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0096396374096983e-06,
+      "loss": 1.0883,
+      "step": 9437
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0051348616091694e-06,
+      "loss": 1.0781,
+      "step": 9438
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.0006401071571247e-06,
+      "loss": 1.0416,
+      "step": 9439
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.96155374508545e-07,
+      "loss": 1.0746,
+      "step": 9440
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.916806641174669e-07,
+      "loss": 1.1749,
+      "step": 9441
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.872159764368282e-07,
+      "loss": 1.0917,
+      "step": 9442
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.82761311918623e-07,
+      "loss": 1.1414,
+      "step": 9443
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.783166710137904e-07,
+      "loss": 1.0156,
+      "step": 9444
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.738820541722704e-07,
+      "loss": 1.0403,
+      "step": 9445
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.694574618429707e-07,
+      "loss": 1.0811,
+      "step": 9446
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.65042894473811e-07,
+      "loss": 1.1097,
+      "step": 9447
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.60638352511678e-07,
+      "loss": 1.0588,
+      "step": 9448
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.562438364024485e-07,
+      "loss": 1.0857,
+      "step": 9449
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.518593465909664e-07,
+      "loss": 1.0498,
+      "step": 9450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.474848835211103e-07,
+      "loss": 1.1125,
+      "step": 9451
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.431204476356925e-07,
+      "loss": 1.0907,
+      "step": 9452
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.387660393765263e-07,
+      "loss": 1.0539,
+      "step": 9453
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.344216591844257e-07,
+      "loss": 1.085,
+      "step": 9454
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.300873074991612e-07,
+      "loss": 1.0771,
+      "step": 9455
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.257629847595262e-07,
+      "loss": 1.0574,
+      "step": 9456
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.214486914032484e-07,
+      "loss": 1.0373,
+      "step": 9457
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.171444278670782e-07,
+      "loss": 1.087,
+      "step": 9458
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.128501945867563e-07,
+      "loss": 1.0673,
+      "step": 9459
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.085659919969791e-07,
+      "loss": 1.0468,
+      "step": 9460
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.042918205314333e-07,
+      "loss": 1.0867,
+      "step": 9461
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.000276806228059e-07,
+      "loss": 1.1001,
+      "step": 9462
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.957735727027516e-07,
+      "loss": 1.113,
+      "step": 9463
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.915294972019372e-07,
+      "loss": 1.0349,
+      "step": 9464
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.872954545499746e-07,
+      "loss": 0.9884,
+      "step": 9465
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.830714451754985e-07,
+      "loss": 1.0585,
+      "step": 9466
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.788574695060891e-07,
+      "loss": 1.0436,
+      "step": 9467
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.746535279683499e-07,
+      "loss": 1.0017,
+      "step": 9468
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.704596209878401e-07,
+      "loss": 1.1222,
+      "step": 9469
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.662757489891205e-07,
+      "loss": 1.0778,
+      "step": 9470
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.621019123957186e-07,
+      "loss": 1.1419,
+      "step": 9471
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.579381116301521e-07,
+      "loss": 1.0934,
+      "step": 9472
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.537843471139507e-07,
+      "loss": 1.0921,
+      "step": 9473
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.496406192675777e-07,
+      "loss": 1.0985,
+      "step": 9474
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.455069285105199e-07,
+      "loss": 1.139,
+      "step": 9475
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.413832752612316e-07,
+      "loss": 1.1259,
+      "step": 9476
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.372696599371566e-07,
+      "loss": 1.0665,
+      "step": 9477
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.331660829547283e-07,
+      "loss": 1.09,
+      "step": 9478
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.290725447293257e-07,
+      "loss": 1.0586,
+      "step": 9479
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.24989045675384e-07,
+      "loss": 1.1033,
+      "step": 9480
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.209155862062501e-07,
+      "loss": 1.0942,
+      "step": 9481
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.168521667342832e-07,
+      "loss": 1.063,
+      "step": 9482
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.127987876708542e-07,
+      "loss": 1.0991,
+      "step": 9483
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.087554494262572e-07,
+      "loss": 1.1271,
+      "step": 9484
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.047221524098314e-07,
+      "loss": 1.0695,
+      "step": 9485
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.006988970298612e-07,
+      "loss": 1.0553,
+      "step": 9486
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.966856836936098e-07,
+      "loss": 1.0682,
+      "step": 9487
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.926825128073634e-07,
+      "loss": 1.0823,
+      "step": 9488
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.886893847763532e-07,
+      "loss": 1.05,
+      "step": 9489
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.847063000048005e-07,
+      "loss": 1.109,
+      "step": 9490
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.807332588959381e-07,
+      "loss": 1.0883,
+      "step": 9491
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.767702618519446e-07,
+      "loss": 1.0573,
+      "step": 9492
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.728173092739988e-07,
+      "loss": 1.0697,
+      "step": 9493
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.688744015622806e-07,
+      "loss": 1.0218,
+      "step": 9494
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.649415391159043e-07,
+      "loss": 1.0935,
+      "step": 9495
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.610187223330178e-07,
+      "loss": 1.0626,
+      "step": 9496
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.571059516107259e-07,
+      "loss": 1.0765,
+      "step": 9497
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.532032273451228e-07,
+      "loss": 1.003,
+      "step": 9498
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.493105499312925e-07,
+      "loss": 1.1076,
+      "step": 9499
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.454279197632863e-07,
+      "loss": 1.0494,
+      "step": 9500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.415553372341455e-07,
+      "loss": 1.0804,
+      "step": 9501
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.376928027358898e-07,
+      "loss": 1.0928,
+      "step": 9502
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.338403166595509e-07,
+      "loss": 1.0752,
+      "step": 9503
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.299978793950945e-07,
+      "loss": 1.1235,
+      "step": 9504
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.261654913315096e-07,
+      "loss": 1.1278,
+      "step": 9505
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.223431528567415e-07,
+      "loss": 1.117,
+      "step": 9506
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.18530864357736e-07,
+      "loss": 1.0772,
+      "step": 9507
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.147286262204178e-07,
+      "loss": 1.0355,
+      "step": 9508
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.109364388296901e-07,
+      "loss": 1.113,
+      "step": 9509
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.07154302569435e-07,
+      "loss": 1.1014,
+      "step": 9510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.033822178225125e-07,
+      "loss": 1.0613,
+      "step": 9511
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.996201849708062e-07,
+      "loss": 1.095,
+      "step": 9512
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.958682043951115e-07,
+      "loss": 1.0693,
+      "step": 9513
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.921262764752801e-07,
+      "loss": 1.0894,
+      "step": 9514
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.883944015900867e-07,
+      "loss": 1.1113,
+      "step": 9515
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.846725801173181e-07,
+      "loss": 1.0566,
+      "step": 9516
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.809608124337508e-07,
+      "loss": 1.0473,
+      "step": 9517
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.772590989151062e-07,
+      "loss": 1.0388,
+      "step": 9518
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.735674399361402e-07,
+      "loss": 1.0625,
+      "step": 9519
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.698858358705429e-07,
+      "loss": 1.0461,
+      "step": 9520
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.662142870910159e-07,
+      "loss": 1.1013,
+      "step": 9521
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.625527939692289e-07,
+      "loss": 1.1113,
+      "step": 9522
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.589013568758518e-07,
+      "loss": 1.0418,
+      "step": 9523
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.552599761805e-07,
+      "loss": 1.107,
+      "step": 9524
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.516286522518122e-07,
+      "loss": 1.061,
+      "step": 9525
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.480073854573832e-07,
+      "loss": 1.0745,
+      "step": 9526
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.443961761638085e-07,
+      "loss": 1.106,
+      "step": 9527
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.407950247366401e-07,
+      "loss": 1.1053,
+      "step": 9528
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.372039315404199e-07,
+      "loss": 1.0771,
+      "step": 9529
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.336228969387015e-07,
+      "loss": 1.0438,
+      "step": 9530
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.300519212939837e-07,
+      "loss": 1.0473,
+      "step": 9531
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.264910049677664e-07,
+      "loss": 1.0532,
+      "step": 9532
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.229401483205055e-07,
+      "loss": 1.0578,
+      "step": 9533
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.193993517116692e-07,
+      "loss": 1.0564,
+      "step": 9534
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.158686154997151e-07,
+      "loss": 1.0906,
+      "step": 9535
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.123479400420351e-07,
+      "loss": 1.1008,
+      "step": 9536
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.088373256950442e-07,
+      "loss": 0.9991,
+      "step": 9537
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.053367728141246e-07,
+      "loss": 1.0588,
+      "step": 9538
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.018462817536375e-07,
+      "loss": 1.0627,
+      "step": 9539
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.983658528669223e-07,
+      "loss": 1.0981,
+      "step": 9540
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.948954865063305e-07,
+      "loss": 1.0227,
+      "step": 9541
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.914351830231368e-07,
+      "loss": 1.0777,
+      "step": 9542
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.879849427676609e-07,
+      "loss": 1.1034,
+      "step": 9543
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.845447660891567e-07,
+      "loss": 1.0641,
+      "step": 9544
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.811146533358902e-07,
+      "loss": 1.0838,
+      "step": 9545
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.776946048550725e-07,
+      "loss": 1.0641,
+      "step": 9546
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.742846209929376e-07,
+      "loss": 1.1301,
+      "step": 9547
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.708847020946762e-07,
+      "loss": 1.0237,
+      "step": 9548
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.674948485044795e-07,
+      "loss": 1.053,
+      "step": 9549
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.641150605654733e-07,
+      "loss": 1.0325,
+      "step": 9550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.60745338619828e-07,
+      "loss": 1.1064,
+      "step": 9551
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.573856830086488e-07,
+      "loss": 1.0991,
+      "step": 9552
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.540360940720413e-07,
+      "loss": 1.127,
+      "step": 9553
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.506965721490787e-07,
+      "loss": 1.0437,
+      "step": 9554
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.47367117577835e-07,
+      "loss": 1.0409,
+      "step": 9555
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.440477306953628e-07,
+      "loss": 1.0929,
+      "step": 9556
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.407384118376602e-07,
+      "loss": 1.1247,
+      "step": 9557
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.374391613397478e-07,
+      "loss": 1.0419,
+      "step": 9558
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.341499795356253e-07,
+      "loss": 1.0161,
+      "step": 9559
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.308708667582374e-07,
+      "loss": 1.0493,
+      "step": 9560
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.276018233395408e-07,
+      "loss": 1.036,
+      "step": 9561
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.243428496104596e-07,
+      "loss": 1.009,
+      "step": 9562
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.210939459009079e-07,
+      "loss": 1.0655,
+      "step": 9563
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.178551125397891e-07,
+      "loss": 1.0814,
+      "step": 9564
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.14626349854952e-07,
+      "loss": 1.0191,
+      "step": 9565
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.114076581732463e-07,
+      "loss": 1.0729,
+      "step": 9566
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.081990378205226e-07,
+      "loss": 1.0998,
+      "step": 9567
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.050004891215764e-07,
+      "loss": 1.066,
+      "step": 9568
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.018120124002157e-07,
+      "loss": 1.0425,
+      "step": 9569
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.986336079791931e-07,
+      "loss": 1.0797,
+      "step": 9570
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.954652761802736e-07,
+      "loss": 1.0737,
+      "step": 9571
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.923070173241894e-07,
+      "loss": 1.0318,
+      "step": 9572
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.891588317306516e-07,
+      "loss": 1.0826,
+      "step": 9573
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.860207197183608e-07,
+      "loss": 1.1194,
+      "step": 9574
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.82892681604985e-07,
+      "loss": 1.0307,
+      "step": 9575
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.79774717707171e-07,
+      "loss": 1.0832,
+      "step": 9576
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.766668283405662e-07,
+      "loss": 1.0531,
+      "step": 9577
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.735690138197968e-07,
+      "loss": 1.0593,
+      "step": 9578
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.7048127445842305e-07,
+      "loss": 1.0488,
+      "step": 9579
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.674036105690505e-07,
+      "loss": 1.0513,
+      "step": 9580
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.643360224632187e-07,
+      "loss": 1.0726,
+      "step": 9581
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.612785104514683e-07,
+      "loss": 1.0198,
+      "step": 9582
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.582310748433072e-07,
+      "loss": 1.04,
+      "step": 9583
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.5519371594724416e-07,
+      "loss": 1.0575,
+      "step": 9584
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.521664340707332e-07,
+      "loss": 1.006,
+      "step": 9585
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.4914922952025153e-07,
+      "loss": 1.0342,
+      "step": 9586
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.4614210260122134e-07,
+      "loss": 1.0416,
+      "step": 9587
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.431450536180659e-07,
+      "loss": 1.0608,
+      "step": 9588
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.401580828741647e-07,
+      "loss": 1.1138,
+      "step": 9589
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.371811906718981e-07,
+      "loss": 1.0909,
+      "step": 9590
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.342143773126361e-07,
+      "loss": 1.1001,
+      "step": 9591
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.31257643096683e-07,
+      "loss": 1.0824,
+      "step": 9592
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.283109883233771e-07,
+      "loss": 1.0443,
+      "step": 9593
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.2537441329099094e-07,
+      "loss": 1.0781,
+      "step": 9594
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.2244791829680886e-07,
+      "loss": 1.0844,
+      "step": 9595
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.195315036370939e-07,
+      "loss": 1.0326,
+      "step": 9596
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.166251696070544e-07,
+      "loss": 1.153,
+      "step": 9597
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.137289165009106e-07,
+      "loss": 1.1062,
+      "step": 9598
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.1084274461186123e-07,
+      "loss": 1.0937,
+      "step": 9599
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.0796665423206146e-07,
+      "loss": 1.0777,
+      "step": 9600
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.0510064565266737e-07,
+      "loss": 1.1164,
+      "step": 9601
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.0224471916381347e-07,
+      "loss": 1.1488,
+      "step": 9602
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.9939887505460185e-07,
+      "loss": 1.0728,
+      "step": 9603
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.965631136131243e-07,
+      "loss": 1.0728,
+      "step": 9604
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.9373743512644004e-07,
+      "loss": 1.0484,
+      "step": 9605
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.9092183988060913e-07,
+      "loss": 1.0299,
+      "step": 9606
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8811632816064814e-07,
+      "loss": 1.0271,
+      "step": 9607
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.85320900250552e-07,
+      "loss": 1.039,
+      "step": 9608
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.825355564333277e-07,
+      "loss": 1.0393,
+      "step": 9609
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7976029699091644e-07,
+      "loss": 1.0594,
+      "step": 9610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7699512220427114e-07,
+      "loss": 0.9802,
+      "step": 9611
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.742400323533235e-07,
+      "loss": 1.0743,
+      "step": 9612
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.714950277169504e-07,
+      "loss": 1.1046,
+      "step": 9613
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.6876010857304076e-07,
+      "loss": 1.0677,
+      "step": 9614
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.660352751984619e-07,
+      "loss": 1.1046,
+      "step": 9615
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.633205278690377e-07,
+      "loss": 1.0767,
+      "step": 9616
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.606158668596038e-07,
+      "loss": 1.0485,
+      "step": 9617
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.579212924439301e-07,
+      "loss": 1.0495,
+      "step": 9618
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.552368048948207e-07,
+      "loss": 1.09,
+      "step": 9619
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.525624044840026e-07,
+      "loss": 1.1134,
+      "step": 9620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.4989809148221473e-07,
+      "loss": 1.0713,
+      "step": 9621
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.472438661591859e-07,
+      "loss": 1.0674,
+      "step": 9622
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.445997287835789e-07,
+      "loss": 1.0522,
+      "step": 9623
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.419656796230908e-07,
+      "loss": 1.1148,
+      "step": 9624
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.3934171894434153e-07,
+      "loss": 1.0572,
+      "step": 9625
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.367278470129742e-07,
+      "loss": 1.0885,
+      "step": 9626
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.341240640935883e-07,
+      "loss": 1.1293,
+      "step": 9627
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.3153037044976187e-07,
+      "loss": 1.055,
+      "step": 9628
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.2894676634407374e-07,
+      "loss": 1.0369,
+      "step": 9629
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.2637325203804805e-07,
+      "loss": 1.0343,
+      "step": 9630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.238098277922097e-07,
+      "loss": 1.0018,
+      "step": 9631
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.2125649386605116e-07,
+      "loss": 0.9894,
+      "step": 9632
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.187132505180657e-07,
+      "loss": 1.0507,
+      "step": 9633
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.161800980056917e-07,
+      "loss": 1.0597,
+      "step": 9634
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.136570365853797e-07,
+      "loss": 1.0606,
+      "step": 9635
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.111440665125254e-07,
+      "loss": 1.058,
+      "step": 9636
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.086411880415141e-07,
+      "loss": 1.0986,
+      "step": 9637
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.061484014257432e-07,
+      "loss": 0.9722,
+      "step": 9638
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0366570691753304e-07,
+      "loss": 1.0791,
+      "step": 9639
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.0119310476821595e-07,
+      "loss": 1.0808,
+      "step": 9640
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.9873059522810276e-07,
+      "loss": 1.0303,
+      "step": 9641
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.962781785464608e-07,
+      "loss": 1.0838,
+      "step": 9642
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.9383585497156917e-07,
+      "loss": 1.149,
+      "step": 9643
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.914036247506635e-07,
+      "loss": 1.0295,
+      "step": 9644
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8898148812994685e-07,
+      "loss": 1.0723,
+      "step": 9645
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8656944535462305e-07,
+      "loss": 1.0729,
+      "step": 9646
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8416749666887455e-07,
+      "loss": 1.0405,
+      "step": 9647
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.817756423158402e-07,
+      "loss": 1.0746,
+      "step": 9648
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.7939388253765963e-07,
+      "loss": 1.1034,
+      "step": 9649
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.770222175754289e-07,
+      "loss": 1.11,
+      "step": 9650
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.746606476692448e-07,
+      "loss": 1.029,
+      "step": 9651
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.723091730581606e-07,
+      "loss": 1.1233,
+      "step": 9652
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.699677939802303e-07,
+      "loss": 1.0572,
+      "step": 9653
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.676365106724643e-07,
+      "loss": 1.0589,
+      "step": 9654
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.6531532337086276e-07,
+      "loss": 1.095,
+      "step": 9655
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.6300423231041535e-07,
+      "loss": 1.1019,
+      "step": 9656
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.6070323772505733e-07,
+      "loss": 1.0621,
+      "step": 9657
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.584123398477245e-07,
+      "loss": 1.0569,
+      "step": 9658
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.561315389103314e-07,
+      "loss": 0.9983,
+      "step": 9659
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.53860835143771e-07,
+      "loss": 1.145,
+      "step": 9660
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.516002287778929e-07,
+      "loss": 1.0465,
+      "step": 9661
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4934972004155844e-07,
+      "loss": 1.0871,
+      "step": 9662
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4710930916258536e-07,
+      "loss": 1.1034,
+      "step": 9663
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.448789963677589e-07,
+      "loss": 1.042,
+      "step": 9664
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.426587818828652e-07,
+      "loss": 1.0605,
+      "step": 9665
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.404486659326577e-07,
+      "loss": 1.1196,
+      "step": 9666
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.382486487408797e-07,
+      "loss": 1.1031,
+      "step": 9667
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.3605873053021976e-07,
+      "loss": 1.0407,
+      "step": 9668
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.338789115223783e-07,
+      "loss": 1.0598,
+      "step": 9669
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.3170919193802322e-07,
+      "loss": 1.0762,
+      "step": 9670
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2954957199678995e-07,
+      "loss": 0.9956,
+      "step": 9671
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2740005191730363e-07,
+      "loss": 1.0269,
+      "step": 9672
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2526063191715685e-07,
+      "loss": 1.054,
+      "step": 9673
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2313131221293194e-07,
+      "loss": 1.0634,
+      "step": 9674
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.2101209302017867e-07,
+      "loss": 1.0851,
+      "step": 9675
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.1890297455342546e-07,
+      "loss": 1.0277,
+      "step": 9676
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.1680395702619038e-07,
+      "loss": 0.9978,
+      "step": 9677
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.1471504065094793e-07,
+      "loss": 1.0334,
+      "step": 9678
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.126362256391623e-07,
+      "loss": 1.0696,
+      "step": 9679
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.1056751220128733e-07,
+      "loss": 1.0181,
+      "step": 9680
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.0850890054671113e-07,
+      "loss": 1.1306,
+      "step": 9681
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.0646039088386694e-07,
+      "loss": 1.0591,
+      "step": 9682
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.0442198342009999e-07,
+      "loss": 1.0985,
+      "step": 9683
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.0239367836176748e-07,
+      "loss": 1.0705,
+      "step": 9684
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.0037547591420514e-07,
+      "loss": 1.0453,
+      "step": 9685
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.9836737628170509e-07,
+      "loss": 1.069,
+      "step": 9686
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.963693796675603e-07,
+      "loss": 1.1646,
+      "step": 9687
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.943814862740201e-07,
+      "loss": 1.0198,
+      "step": 9688
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.9240369630232347e-07,
+      "loss": 1.1032,
+      "step": 9689
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.9043600995268806e-07,
+      "loss": 1.0868,
+      "step": 9690
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.884784274243101e-07,
+      "loss": 1.0905,
+      "step": 9691
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8653094891534217e-07,
+      "loss": 1.1295,
+      "step": 9692
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8459357462293768e-07,
+      "loss": 1.1405,
+      "step": 9693
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8266630474321757e-07,
+      "loss": 1.0436,
+      "step": 9694
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.8074913947128124e-07,
+      "loss": 1.0843,
+      "step": 9695
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7884207900121796e-07,
+      "loss": 1.0537,
+      "step": 9696
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.769451235260622e-07,
+      "loss": 1.1182,
+      "step": 9697
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7505827323784917e-07,
+      "loss": 1.1244,
+      "step": 9698
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7318152832759283e-07,
+      "loss": 1.0832,
+      "step": 9699
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7131488898527448e-07,
+      "loss": 1.0882,
+      "step": 9700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.694583553998541e-07,
+      "loss": 1.096,
+      "step": 9701
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.6761192775927025e-07,
+      "loss": 1.14,
+      "step": 9702
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.657756062504512e-07,
+      "loss": 1.0733,
+      "step": 9703
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.6394939105927044e-07,
+      "loss": 1.0435,
+      "step": 9704
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.6213328237061342e-07,
+      "loss": 1.0932,
+      "step": 9705
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.603272803683109e-07,
+      "loss": 1.0775,
+      "step": 9706
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.585313852351944e-07,
+      "loss": 1.1286,
+      "step": 9707
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.56745597153074e-07,
+      "loss": 1.0946,
+      "step": 9708
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.549699163027163e-07,
+      "loss": 1.1118,
+      "step": 9709
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.5320434286386632e-07,
+      "loss": 1.0292,
+      "step": 9710
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.514488770152811e-07,
+      "loss": 1.0329,
+      "step": 9711
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4970351893464074e-07,
+      "loss": 1.0215,
+      "step": 9712
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4796826879863723e-07,
+      "loss": 1.1031,
+      "step": 9713
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4624312678294117e-07,
+      "loss": 1.1067,
+      "step": 9714
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4452809306219062e-07,
+      "loss": 1.0446,
+      "step": 9715
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4282316780999117e-07,
+      "loss": 1.0208,
+      "step": 9716
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4112835119892698e-07,
+      "loss": 1.071,
+      "step": 9717
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3944364340058302e-07,
+      "loss": 1.0664,
+      "step": 9718
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3776904458550067e-07,
+      "loss": 1.0973,
+      "step": 9719
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3610455492318874e-07,
+      "loss": 1.0488,
+      "step": 9720
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3445017458216803e-07,
+      "loss": 1.0854,
+      "step": 9721
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3280590372988233e-07,
+      "loss": 1.0857,
+      "step": 9722
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.311717425328096e-07,
+      "loss": 1.1043,
+      "step": 9723
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2954769115637312e-07,
+      "loss": 1.1051,
+      "step": 9724
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2793374976495244e-07,
+      "loss": 1.1036,
+      "step": 9725
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.263299185219613e-07,
+      "loss": 1.0205,
+      "step": 9726
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2473619758974763e-07,
+      "loss": 1.0664,
+      "step": 9727
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2315258712962685e-07,
+      "loss": 1.0677,
+      "step": 9728
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2157908730193734e-07,
+      "loss": 1.0233,
+      "step": 9729
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.200156982659406e-07,
+      "loss": 1.1108,
+      "step": 9730
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1846242017992115e-07,
+      "loss": 1.0997,
+      "step": 9731
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1691925320110875e-07,
+      "loss": 1.0611,
+      "step": 9732
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1538619748572288e-07,
+      "loss": 1.0451,
+      "step": 9733
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1386325318893942e-07,
+      "loss": 1.0797,
+      "step": 9734
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1235042046495725e-07,
+      "loss": 1.0896,
+      "step": 9735
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1084769946689832e-07,
+      "loss": 1.0569,
+      "step": 9736
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.093550903468965e-07,
+      "loss": 1.0745,
+      "step": 9737
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0787259325605315e-07,
+      "loss": 1.0575,
+      "step": 9738
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0640020834442599e-07,
+      "loss": 1.0567,
+      "step": 9739
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0493793576108469e-07,
+      "loss": 1.1029,
+      "step": 9740
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0348577565404416e-07,
+      "loss": 1.0692,
+      "step": 9741
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0204372817030905e-07,
+      "loss": 1.0868,
+      "step": 9742
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0061179345586258e-07,
+      "loss": 1.0722,
+      "step": 9743
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.918997165566657e-08,
+      "loss": 1.0522,
+      "step": 9744
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.777826291365033e-08,
+      "loss": 1.0499,
+      "step": 9745
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.637666737271067e-08,
+      "loss": 1.0819,
+      "step": 9746
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.49851851747452e-08,
+      "loss": 1.0386,
+      "step": 9747
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.360381646060789e-08,
+      "loss": 1.0156,
+      "step": 9748
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.223256137014246e-08,
+      "loss": 1.0873,
+      "step": 9749
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.08714200421601e-08,
+      "loss": 1.1487,
+      "step": 9750
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.952039261445055e-08,
+      "loss": 1.0268,
+      "step": 9751
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.817947922378223e-08,
+      "loss": 1.0441,
+      "step": 9752
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.684868000589097e-08,
+      "loss": 1.037,
+      "step": 9753
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.552799509550236e-08,
+      "loss": 1.0914,
+      "step": 9754
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.421742462629834e-08,
+      "loss": 1.0564,
+      "step": 9755
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.291696873096167e-08,
+      "loss": 1.0987,
+      "step": 9756
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.162662754113149e-08,
+      "loss": 1.069,
+      "step": 9757
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.034640118743663e-08,
+      "loss": 1.0577,
+      "step": 9758
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.907628979946236e-08,
+      "loss": 1.0347,
+      "step": 9759
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.781629350580577e-08,
+      "loss": 1.0728,
+      "step": 9760
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.65664124339982e-08,
+      "loss": 1.0305,
+      "step": 9761
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.532664671057177e-08,
+      "loss": 1.0891,
+      "step": 9762
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.409699646103718e-08,
+      "loss": 1.0641,
+      "step": 9763
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.287746180986154e-08,
+      "loss": 1.0972,
+      "step": 9764
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.166804288051277e-08,
+      "loss": 1.0714,
+      "step": 9765
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.046873979540403e-08,
+      "loss": 1.004,
+      "step": 9766
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.927955267594932e-08,
+      "loss": 1.1313,
+      "step": 9767
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.810048164254124e-08,
+      "loss": 1.0942,
+      "step": 9768
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.693152681452874e-08,
+      "loss": 1.0363,
+      "step": 9769
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.577268831023942e-08,
+      "loss": 1.0078,
+      "step": 9770
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.462396624700162e-08,
+      "loss": 1.0507,
+      "step": 9771
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.348536074108902e-08,
+      "loss": 1.0865,
+      "step": 9772
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.235687190776496e-08,
+      "loss": 1.1115,
+      "step": 9773
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.123849986127139e-08,
+      "loss": 1.1048,
+      "step": 9774
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.013024471481776e-08,
+      "loss": 1.089,
+      "step": 9775
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.9032106580592106e-08,
+      "loss": 1.0778,
+      "step": 9776
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.7944085569772154e-08,
+      "loss": 1.0536,
+      "step": 9777
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.686618179248093e-08,
+      "loss": 1.1451,
+      "step": 9778
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.579839535785336e-08,
+      "loss": 1.0562,
+      "step": 9779
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.4740726373980755e-08,
+      "loss": 1.0679,
+      "step": 9780
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.369317494791082e-08,
+      "loss": 1.1419,
+      "step": 9781
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.2655741185714255e-08,
+      "loss": 1.047,
+      "step": 9782
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.162842519240707e-08,
+      "loss": 1.0866,
+      "step": 9783
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.0611227071972745e-08,
+      "loss": 1.03,
+      "step": 9784
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.9604146927384464e-08,
+      "loss": 1.0782,
+      "step": 9785
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.8607184860605114e-08,
+      "loss": 1.1007,
+      "step": 9786
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.762034097254286e-08,
+      "loss": 1.1136,
+      "step": 9787
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.664361536310668e-08,
+      "loss": 1.0449,
+      "step": 9788
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.5677008131161935e-08,
+      "loss": 1.0653,
+      "step": 9789
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.4720519374574774e-08,
+      "loss": 1.131,
+      "step": 9790
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.377414919014555e-08,
+      "loss": 1.0799,
+      "step": 9791
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.283789767370872e-08,
+      "loss": 1.1056,
+      "step": 9792
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.19117649200107e-08,
+      "loss": 1.0617,
+      "step": 9793
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.099575102283204e-08,
+      "loss": 1.0721,
+      "step": 9794
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.0089856074876367e-08,
+      "loss": 1.0835,
+      "step": 9795
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.9194080167870296e-08,
+      "loss": 1.1007,
+      "step": 9796
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.8308423392485746e-08,
+      "loss": 1.013,
+      "step": 9797
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.7432885838373235e-08,
+      "loss": 1.0407,
+      "step": 9798
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.6567467594161855e-08,
+      "loss": 1.1396,
+      "step": 9799
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.571216874748151e-08,
+      "loss": 0.9787,
+      "step": 9800
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.48669893848963e-08,
+      "loss": 1.0637,
+      "step": 9801
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.403192959196e-08,
+      "loss": 1.0945,
+      "step": 9802
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3206989453216096e-08,
+      "loss": 1.1093,
+      "step": 9803
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.239216905217557e-08,
+      "loss": 1.0542,
+      "step": 9804
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.1587468471316886e-08,
+      "loss": 1.1513,
+      "step": 9805
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.07928877920971e-08,
+      "loss": 1.0558,
+      "step": 9806
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.000842709497409e-08,
+      "loss": 1.0624,
+      "step": 9807
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.923408645932879e-08,
+      "loss": 1.0014,
+      "step": 9808
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.846986596357626e-08,
+      "loss": 1.0549,
+      "step": 9809
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7715765685065732e-08,
+      "loss": 1.0836,
+      "step": 9810
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.6971785700136142e-08,
+      "loss": 1.0525,
+      "step": 9811
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.6237926084093923e-08,
+      "loss": 0.9859,
+      "step": 9812
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.5514186911246296e-08,
+      "loss": 1.0695,
+      "step": 9813
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.480056825484578e-08,
+      "loss": 1.0128,
+      "step": 9814
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.4097070187134585e-08,
+      "loss": 1.0541,
+      "step": 9815
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.3403692779333518e-08,
+      "loss": 1.0355,
+      "step": 9816
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.272043610161978e-08,
+      "loss": 1.0652,
+      "step": 9817
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2047300223182466e-08,
+      "loss": 1.0623,
+      "step": 9818
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.138428521214486e-08,
+      "loss": 1.0444,
+      "step": 9819
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.0731391135631052e-08,
+      "loss": 1.0664,
+      "step": 9820
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.008861805973261e-08,
+      "loss": 1.072,
+      "step": 9821
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.9455966049530815e-08,
+      "loss": 1.0745,
+      "step": 9822
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8833435169041125e-08,
+      "loss": 1.0542,
+      "step": 9823
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.8221025481313105e-08,
+      "loss": 1.0516,
+      "step": 9824
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.76187370483305e-08,
+      "loss": 1.087,
+      "step": 9825
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.702656993105567e-08,
+      "loss": 1.0547,
+      "step": 9826
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.6444524189440646e-08,
+      "loss": 1.0721,
+      "step": 9827
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.587259988240497e-08,
+      "loss": 1.0303,
+      "step": 9828
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.5310797067846772e-08,
+      "loss": 1.0275,
+      "step": 9829
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4759115802642776e-08,
+      "loss": 1.0897,
+      "step": 9830
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.4217556142637201e-08,
+      "loss": 1.0624,
+      "step": 9831
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3686118142641757e-08,
+      "loss": 1.0576,
+      "step": 9832
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3164801856468956e-08,
+      "loss": 1.0178,
+      "step": 9833
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2653607336887697e-08,
+      "loss": 1.0698,
+      "step": 9834
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.2152534635645474e-08,
+      "loss": 1.1317,
+      "step": 9835
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1661583803457276e-08,
+      "loss": 1.0641,
+      "step": 9836
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1180754890038892e-08,
+      "loss": 1.0617,
+      "step": 9837
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0710047944062495e-08,
+      "loss": 1.02,
+      "step": 9838
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0249463013167759e-08,
+      "loss": 1.0413,
+      "step": 9839
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.79900014398405e-09,
+      "loss": 1.0757,
+      "step": 9840
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.358659382119328e-09,
+      "loss": 1.037,
+      "step": 9841
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.928440772137947e-09,
+      "loss": 1.127,
+      "step": 9842
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.508344357605058e-09,
+      "loss": 0.9894,
+      "step": 9843
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.098370181042203e-09,
+      "loss": 1.1288,
+      "step": 9844
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.698518283949518e-09,
+      "loss": 1.0995,
+      "step": 9845
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.3087887067946335e-09,
+      "loss": 1.1314,
+      "step": 9846
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.9291814890459775e-09,
+      "loss": 1.0549,
+      "step": 9847
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.559696669128368e-09,
+      "loss": 1.0959,
+      "step": 9848
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.200334284434117e-09,
+      "loss": 1.0842,
+      "step": 9849
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.851094371356336e-09,
+      "loss": 1.0885,
+      "step": 9850
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.511976965244525e-09,
+      "loss": 1.1322,
+      "step": 9851
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.182982100426781e-09,
+      "loss": 1.0947,
+      "step": 9852
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.864109810198691e-09,
+      "loss": 1.0779,
+      "step": 9853
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.5553601268566444e-09,
+      "loss": 1.1635,
+      "step": 9854
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.256733081642317e-09,
+      "loss": 1.0816,
+      "step": 9855
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.968228704798182e-09,
+      "loss": 1.0733,
+      "step": 9856
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.6898470255120053e-09,
+      "loss": 1.0535,
+      "step": 9857
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.4215880719834503e-09,
+      "loss": 1.1031,
+      "step": 9858
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.16345187135747e-09,
+      "loss": 1.062,
+      "step": 9859
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.915438449768715e-09,
+      "loss": 1.0814,
+      "step": 9860
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.6775478323304294e-09,
+      "loss": 1.0666,
+      "step": 9861
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.449780043112249e-09,
+      "loss": 1.0973,
+      "step": 9862
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.232135105173505e-09,
+      "loss": 1.1171,
+      "step": 9863
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.0246130405521257e-09,
+      "loss": 1.0259,
+      "step": 9864
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8272138702535303e-09,
+      "loss": 1.0993,
+      "step": 9865
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.6399376142617329e-09,
+      "loss": 1.0393,
+      "step": 9866
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.4627842915282409e-09,
+      "loss": 1.0867,
+      "step": 9867
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.295753919994258e-09,
+      "loss": 1.0195,
+      "step": 9868
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.1388465165684814e-09,
+      "loss": 1.0409,
+      "step": 9869
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.920620971382022e-10,
+      "loss": 1.0234,
+      "step": 9870
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.554006765471023e-10,
+      "loss": 1.0757,
+      "step": 9871
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 7.288622686396629e-10,
+      "loss": 1.047,
+      "step": 9872
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.124468862278577e-10,
+      "loss": 1.0389,
+      "step": 9873
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.06154541102255e-10,
+      "loss": 1.0819,
+      "step": 9874
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.0998524399871174e-10,
+      "loss": 1.0974,
+      "step": 9875
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.239390046871904e-10,
+      "loss": 1.1084,
+      "step": 9876
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.4801583183853286e-10,
+      "loss": 1.094,
+      "step": 9877
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.8221573316878903e-10,
+      "loss": 1.0691,
+      "step": 9878
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.2653871533929718e-10,
+      "loss": 1.0491,
+      "step": 9879
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.098478396778575e-11,
+      "loss": 1.1001,
+      "step": 9880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.555394367278254e-11,
+      "loss": 0.9869,
+      "step": 9881
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.0246198040307917e-11,
+      "loss": 1.0272,
+      "step": 9882
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.06154963497707e-12,
+      "loss": 1.0826,
+      "step": 9883
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0,
+      "loss": 1.0481,
+      "step": 9884
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 9884,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "total_flos": 1.2994919207031276e+19,
+  "trial_name": null,
+  "trial_params": null
+}